Authors: Gunzung Kim, Jeongsook Eom, Yongwan Park
Categories: Article, LiDAR, Urban air mobility, Wind hazards, Risley prism, Optical OFDMA, DWDMA, Variable field-of-view, Visual perception, Aerospace engineering, Electrical and electronic engineering, Atmospheric science
Source: Scientific Reports
Authors: Gunzung Kim, Jeongsook Eom, Yongwan Park
Urban Air Mobility (UAM) is particularly vulnerable to wind hazards, and conventional weather monitoring tools often do not offer the detailed, real-time information needed for safe operations. This research, which examines the visual perception of the on-board cycloidal scanning LiDAR system in improving UAM safety, is of significant importance. The cycloidal scanning LiDAR system, designed explicitly for on-board integration, delivers high-resolution visual mapping, real-time data processing, and comprehensive environmental scanning with 360° rotational capabilities. Its lightweight design and low power consumption make it well-suited for UAM applications, providing continual visual updates on wind conditions along the flight path. This study underscores the system’s effectiveness in providing advanced visual perception for UAM operations. By emphasizing the crucial role of visual perception in identifying and responding to wind hazards, the research highlights the significance of this technology in guaranteeing safe and efficient UAM operations. Striking a balance between real-time visual capabilities and practical considerations such as power, size, and weight is vital to optimizing UAM safety and efficiency.
The emergence of urban air mobility (UAM), utilizing advanced aerial vehicles (AAV) like electric vertical takeoff and landing (eVTOL) aircraft, introduces significant safety challenges due to low-level wind hazards in the atmospheric surface layer (ASL)^1,2^, such as clear air turbulence (CAT), gusts, and wind shear. Unlike civil aircraft operating at stable altitudes of around 20,000 meters, where existing meteorological services provide sufficient wind hazard information for pre-defined routes, UAM operates at low altitudes (as low as 300 meters), where urban landscapes and atmospheric instability cause wind conditions to change rapidly^3,4^. This dynamic environment necessitates onboard, real-time wind hazard detection, as conventional systems are insufficient, underscoring the critical need for advanced sensing and monitoring technologies specifically designed for UAM operations.
CAT is hazardous as it occurs without visible signs, making it hard to predict and avoid^5–8^. In urban environments, CAT is intensified by airflow interaction with buildings and other structures, leading to sudden, unpredictable changes in wind speed and direction^9–12^. These sudden changes can cause UAM vehicles to experience abrupt altitude shifts and loss of control, increasing the risk of accidents during critical flight phases such as takeoff and landing.
Gust, characterized by rapid and short-term increases in wind speed, is another significant hazard in urban settings^13–15^. With its tall buildings and narrow streets, the urban landscape can create wind tunnels that accelerate wind speeds and generate powerful gust. These gust imposes dynamic loads on UAM vehicles, challenging their structural integrity and flight stability. The sudden nature of gust means that UAM vehicles can be subjected to unexpected forces, leading to structural damage or loss of control.
Wind shear, involving abrupt changes in wind speed and direction over short distances, is particularly dangerous for UAM operations^16–20^. Both vertical and horizontal wind shear are typical in urban areas due to the varied building heights and the uneven distribution of open spaces^21–24^. Vertical wind shear intensified near tall structures, poses severe risks during takeoff, landing, and low-altitude cruising, as it can lead to sudden changes in lift and aerodynamic forces. Horizontal wind shear, influenced by the spatial arrangement of buildings, can cause significant variations in flight conditions over short distances, making it challenging for UAM vehicles to maintain stable flight paths.
The unpredictability and intensity of wind hazards in urban environments pose significant risks for UAM operations^25–30^. CAT, gust, and wind shear can cause sudden and violent changes in airflow, imposing severe stress on vehicle structures and destabilizing shifts in wind direction and speed. It is imperative to underscore the pivotal role of advanced technological solutions and stringent safety protocols in ensuring the safety of UAM operations. These elements are essential and represent the path forward in guaranteeing the security and reliability of UAM operations. Mitigation strategies, including advanced meteorological monitoring technologies such as Doppler radar and LiDAR (light detection and ranging), are crucial for detecting and predicting these hazards. Real-time data from these tools can inform flight planning and operational decision-making, helping to avoid hazardous conditions. Furthermore, UAM vehicles must be designed with robust control systems and resilient structural components to withstand the dynamic loads imposed by these wind hazards. In summary, addressing these low-level wind hazards is vital for the safe integration of UAM into urban transportation systems, requiring advanced monitoring technologies, resilient vehicle design, and comprehensive safety protocols.
This article introduces an on-board cycloidal scanning LiDAR system that significantly enhances UAM safety by offering advanced visual perception capabilities to detect wind hazards, such as clear air turbulence and gusts, through aerosol detection. Building on our previous works^31,32^, which introduced and refined a LiDAR system employing orthogonal optical frequency division multiple access (OOFDMA) technology and a Risley prism for advanced beam steering, this paper extends the application of this technology to UAM. Specifically, this study proposes and validates the use of the cycloidal scanning LiDAR system to improve UAM safety by providing real-time environmental awareness in urban airspaces. This represents the first application of the OOFDMA-based LiDAR system to a UAM-specific context, addressing the unique challenges of detecting wind hazards in complex urban environments. The key contributions of this study are as The system uses visual perception to spot CAT, gusts, and wind shear up to 2 km ahead. This enables early detection of potential hazards, allowing for timely and proactive safety measures.The system’s 360° rotational capability ensures complete visual scanning of the environment, minimizing blind spots and providing full situational awareness. The variable field-of-view (FoV) mechanism optimizes visual scanning for specific regions of interest, ensuring effective monitoring of critical areas.By employing multiple LiDAR units, the system provides high spatial resolution (0.1°) and a rapid scanning rate (50 Hz). These features ensure detailed visual mapping of wind conditions and obstacles, offering comprehensive situational awareness that instills confidence in the system’s safety features.The system incorporates particle image density (PID) to detect and characterize wind hazards efficiently with minimal computational overhead. This method enables real-time hazard avoidance without complex signal processing. Additional quantitative analysis, including figures and tables, demonstrates the system’s performance in UAM-specific scenarios.Detailed visual perception of atmospheric conditions enables UAM vehicles to take preemptive actions, such as adjusting flight paths or altitudes, to avoid adverse weather conditions. This proactive approach significantly reduces the risk of accidents caused by sudden wind changes, enhancing trust in the system’s safety features.The rest of this paper is organized into four sections. Section “Wind hazard in atmospheric surface layer and remote sensing” describes the adaptation of the Risley prism for LiDAR technology. The technical design and operational principle of cycloidal scanning LiDAR is presented in section “Technical design and operational principle of on-board cycloidal scanning LiDAR system”. It also contains the variable FoV mechanism of the cycloidal scanning LiDAR. Section “Visual perception of wind hazards with LiDAR system” explains the performance evaluations of visual perception to detect wind hazards with four LiDAR systems. Section “Conclusion” concludes this study.
The ASL is the lowest part of the Earth’s atmosphere and interacts directly with the surface^33,34^. It typically makes up the bottom 10% of the planetary boundary layer (PBL) and extends a few tens of meters above the surface. This layer is characterized by intense turbulence due to surface friction, which mixes heat, momentum, and substances such as water vapor and pollutants. Wind speed in the ASL increases logarithmically with height and is influenced by surface roughness from vegetation, buildings, and terrain^34,35^. Understanding the ASL, particularly its role in climate change, is crucial for managing weather, climate, agriculture, air quality, and renewable energy.
Low-level wind hazards in the ASL, such as wind shear, microbursts, and turbulence, are not mere potential risks, they pose significant and immediate dangers, especially to aviation^18,36,37^. Wind shear involves rapid changes in wind speed and direction, which are particularly perilous during takeoff and landing. Microbursts are intense downdrafts that can generate surface winds exceeding 100 knots, causing severe risks to aviation and ground-based structures. Turbulence can cause significant structural stress on aircraft and infrastructure. The importance of advanced meteorological monitoring and forecasting techniques like Doppler radar and LiDAR systems in detecting and predicting these hazardous conditions cannot be overstated.
Wind hazards don’t just affect the atmosphere’s behavior, and they significantly influence the behavior of aerosols as well^38–41^. For instance, strong winds can transport aerosols over long distances, influencing air quality and climate far from their source regions. Convective currents and turbulence elevate and mix aerosols, affecting regional air quality and visibility. Wind hazards also contribute to the resuspension of ground aerosols, increasing atmospheric concentrations. This interplay between wind hazards and aerosols, which is often complex and dynamic, is crucial to our understanding of ASL and its implications for various fields.
Detecting low-level wind hazards through aerosol movement involves advanced observational technologies, remote sensing, and atmospheric modeling. Ground-based stations equipped with anemometers^42–44^, weather radar^45–51^, wind LiDAR^1,52–59^, and aerosol spectrometers^60–62^ are crucial. Anemometers measure wind speed and direction, weather radar and wind LiDAR provide detailed profiles of aerosol distribution and wind velocity, and aerosol spectrometers analyze aerosol size and concentration. Remote sensing technologies like satellites^63,64^ offers comprehensive monitoring of wind-driven aerosol movements, aiding in hazard detection and forecasting.
Using remote sensing technologies in the ASL is indispensable for understanding and mitigating wind hazards. For instance, Doppler weather radar^45–51^ emits microwave signals that bounce off atmospheric particles and measures the frequency shift to determine velocity and direction. It is critical for detecting wind patterns and identifying hazards such as gust fronts and tornadoes. Micropulse LiDAR (MPL)^65–67^, known for its compact design and high pulse repetition frequency, monitors aerosol concentrations and distributions, providing high-resolution vertical profiles essential for air quality studies, climate research, and pollution control^53–55,68,69^. Avionics weather radar^70–72^, specifically designed for aircraft, provides real-time information on precipitation, turbulence, and other atmospheric phenomena, enhancing flight safety. Coherent-detection Doppler wind LiDAR^57,73^, which measures wind speed and direction by capturing the Doppler shift in the backscattered laser signal, provides highly accurate vertical and horizontal wind speed profiles, essential for comprehensive wind pattern analysis. These technologies, when integrated, can significantly improve the detection and mitigation of wind hazards in the ASL, enhancing safety and operational efficiency in sectors such as aviation and environmental management.
Various technologies are employed to detect wind hazards in the ASL^16,74–80^, each offering unique advantages and limitations critical for specific applications. Meteorological observations, utilizing instruments such as anemometers and wind vanes, provide precise and real-time data on wind speed and direction at specific locations. However, their spatial coverage is inherently limited and can be influenced by local obstructions. Weather radar systems, which use radio waves to monitor atmospheric phenomena, including wind patterns and precipitation, offer wide-area coverage and early warning capabilities but require substantial investment and maintenance, and they may need more vertical resolution. LiDAR technology uses laser pulses to create detailed three-dimensional measurements of wind fields and turbulence. It provides high-resolution data for atmospheric research and wind energy assessment applications. however, it has a limited range and can be significantly affected by adverse weather conditions. SODAR (sonic detection and ranging) systems, which use sound waves to measure vertical wind speed and turbulence profiles, offer precise data on the dynamics of the lower atmosphere but are limited in altitude coverage and susceptible to interference from background noise and terrain effects. Satellite imagery, leveraging advanced remote sensing technologies like scatterometers, radiometers, and synthetic aperture radar, provides a comprehensive view of wind patterns and extreme weather events over large areas, crucial for disaster management, though it typically offers lower spatial and temporal resolution compared to ground-based technologies. numerical weather prediction (NWP) models synthesize data from diverse sources to simulate and forecast wind hazards over extensive regions and timeframes, offering valuable guidance for weather forecasting and hazard mitigation. However, they require substantial computational resources and may not fully capture the complexity of localized weather phenomena.
Although a dense ground observation network provides detailed information from the ground^81^, weather information at various altitudes is still severely lacking. In particular, high-resolution vertical information 20–50 m from the ground to the UAM flight path is required around vertiports, in addition to high-resolution horizontal information. Therefore, at vertiports^30^, ground-based observations are necessary to obtain high-resolution weather information with high accuracy according to altitude, and weather observation drones can be utilized as a means to fulfill this requirement. By deploying weather drones at or around vertiports, they can repeatedly move between the ground and the UAM flight path, sampling weather information according to altitude. The advantage of this method is that it obtains weather information with high resolution and accuracy according to altitude. However, in conditions of strong winds or hazardous weather, the weather drones themselves may have difficulty operating, so information acquisition will only be possible in limited weather conditions. The CopterSonde and MeteoDrone^81^, developed in Oklahoma, USA, and Switzerland, have continuously conducted experiments on data accuracy and the operational environment through research and development. They compared the wind observed by CopterSonde, radiosonde, and Doppler lidar. The changes in wind according to altitude were well observed. Weather observation by such weather drones can be effectively used in areas where UAM weather data is lacking.
Mitigating wind hazards in UAM operations involves a multifaceted approach^30,82,83^, integrating advanced weather forecasting, real-time weather monitoring, and sophisticated on-board avionics systems to provide crucial information on wind conditions and potential hazards. Effective mitigation strategies begin with precise weather forecasting, enabling UAM operators to anticipate and plan for hazardous conditions well in advance. Real-time weather monitoring further enhances decision-making by providing up-to-date hazard information, which is essential for dynamic route planning to optimize flight paths. Robust aircraft design is another critical component, ensuring UAM vehicles can withstand turbulent conditions and maintain structural integrity under dynamic loads. Autonomous flight systems enhance operational safety by adapting to sudden wind changes and mitigating pilot workload. Ground-based wind mitigation measures, particularly near landing sites, are vital in minimizing wind-induced risks. But the specialized pilot training and procedural adjustments are truly essential. They equip pilots with the skills to handle wind hazards effectively, ensuring safe operations within the ASL.
Proactive management of wind hazards is crucial for the safety and efficiency of UAM operations^30,84^. UAM vehicles should receive comprehensive wind condition information well before their intended flight path. Ideally, operators should access weather forecasts and real-time updates several hours before scheduled flights to assess potential hazards and adjust plans accordingly. Receiving wind hazard data tens of kilometers ahead of the planned flight path allows sufficient time to reroute flights or implement mitigation strategies, such as changing altitude or adjusting airspeed. This proactive approach ensures effective risk management, maintaining operational safety and efficiency in the presence of wind hazards. However, what truly enhances the detection of specific wind hazards within the ASL is the integration of weather radar and LiDAR technologies. On-board LiDAR or radar sensors^85^, providing real-time situational awareness, significantly improve safety and operational efficiency by enabling prompt responses to dynamic atmospheric conditions. The integrated system of on-board and external sensor technologies, supported by advanced data processing and communication networks, is essential for enhancing the safety and efficiency of UAM operations, particularly in the real-time detection and response to dynamic environmental conditions.
This cycloidal scanning LiDAR system, proposed by the authors^31,32^, significantly enhances the ability to detect and respond to low-level wind hazards and other obstacles through its innovative use of Risley prisms^86–89^, OOFDMA^90–93^, and dense wavelength division multiple access (DWDMA) technologies^94–97^, as shown in Fig. 1. The combination of precise signal processing, comprehensive rotational capabilities, and robust security measures makes this system a reliable and effective solution for improving the safety and efficiency of UAM operations. The detailed and flexible design of the system ensures it can adapt to a wide range of environmental conditions and operational requirements, making it a versatile tool for modern autonomous applications. By utilizing multiple LiDAR units to increase measurement resolution^96–99^, the system can detect even the most miniature objects, such as aerosols, ensuring higher safety and operational reliability in UAM environments.
The hardware of the proposed LiDAR system consists of four independent LiDAR units, each equipped with OOFDMA technology for efficient data handling and a Risley prism for precise beam steering, as previously described in the reference^32^. These LiDAR units operate under the control of the edge computer, which dynamically adjusts their scanning patterns based on operational needs. For general monitoring, the LiDAR units evenly divide their FoV to provide wide-area coverage, while for focused monitoring, they concentrate their scans on specific regions of interest. This approach will be applied in the proposed method for detecting wind hazards in UAM operations, ensuring that both broad environmental awareness and detailed hazard detection are achieved. The edge computer handles real-time data synchronization and fusion, integrating the data streams from all four units into a unified dataset.
The software architecture is designed to support three primary (1) controlling the operational parameters of the four LiDAR units, such as scan rates and field-of-view configurations, (2) fusing the collected data to generate detailed point clouds, and (3) calculating PID and wind-related properties, including wind speed, direction, and turbulence intensity. These outputs are critical for real-time wind hazard detection and avoidance, allowing for safe UAM operations. The software also employs advanced data processing algorithms to ensure high accuracy and low latency in hazard detection.
The LiDAR system, developed to identify low-level wind hazards and enhance the safe operation of UAM, is designed with a sophisticated architecture integrating 128 laser diodes and external optical modulators (EOMs). These diodes operate at different wavelengths, enabling a diverse range of signals. These signals are merged into a single optical stream using an optical coupler and directed to the rotating Risley prism along its central axis through an optical cable. The Risley prism serves as the critical component of this system, refracting each laser signal at unique angles based on their wavelengths. To enable simultaneous transmission and reception of data across multiple locations, the system employs OOFDMA and DWDMA technologies, significantly enhancing the functionality and performance of the LiDAR system.
The laser diodes generate signals at various wavelengths specified by DWDMA technology. DWDMA allows multiple optical signals to be transmitted simultaneously over a single optical fiber by utilizing different wavelengths for each signal. An optical coupler merges these multiple signals into a single coherent stream directed toward the Risley prism. Inside the Risley prism, each laser signal is split into multiple beams based on their specific wavelength and other parameters, such as the angle of incidence and refractive index. These split signals are refracted at various angles, enabling the system to transmit multiple pulse streams with different wavelengths toward various target directions. This setup significantly enhances the system’s capacity for comprehensive and detailed scanning.
DWDMA is a technique that increases the bandwidth and efficiency of optical transmission by multiplexing different wavelengths. In the LiDAR system, DWDMA combines laser signals of different wavelengths into a single optical stream. Each wavelength carries a distinct data stream, enabling simultaneous transmission of multiple signals. This multiplexing allows the LiDAR system to handle large amounts of data, improving its ability to perform detailed and extensive scans. DWDMA enhances the system’s data transmission capacity, providing more comprehensive and detailed scanning capabilities.
OOFDMA technology divides the optical spectrum into multiple orthogonal sub-channels. Each sub-channel is modulated with a unique frequency, enabling multiple data streams to be transmitted and received concurrently without interference. This method significantly improves the efficiency and speed of data transmission by reducing the time required for laser pulse transmission and reception. By leveraging OOFDMA, the LiDAR system can effectively distinguish between various laser signals, making it particularly suitable for simultaneously detecting multiple objects and wind hazards. The integration of DWDMA with OOFDMA optimizes the available optical spectrum, enabling the LiDAR system to handle high data rates and complex scanning tasks efficiently.
In the context of the LiDAR system, OOFDMA enables efficient data encoding by assigning different sub-channels to different data streams. This reduces the transmission time and enhances the system’s ability to handle high data rates. The orthogonality of the sub-channels ensures minimal interference between them, allowing for clear and precise data transmission. Upon reception, the system decodes the transmitted signals by separating the combined sub-channels. This process involves complex signal processing techniques to accurately reconstruct the original data streams. Using OOFDMA in the LiDAR system significantly improves its capacity to handle multiple data streams, enhancing its overall performance and reliability.
Once the laser pulse streams are reflected off objects, they are gathered by a lens and sorted by an indium phosphide (InP) arrayed waveguide grating (AWG). The AWG disperses the light into its component wavelengths, amplified using an erbium-doped fiber amplifier (EDFA). The amplified optical signals are converted into electrical signals by avalanche photodiodes (APDs) and transimpedance amplifiers (TIAs). These electrical signals are processed by an descrete Hartley transform precoded (HTP)-based flip-OFDM decoder^91,92^, which decodes the transmitted data. The system also employs advanced encryption standard (AES) encryption to ensure that only authorized signals are processed, maintaining the integrity and security of the data. The system measures a range of parameters, including distance, speed, time of flight, signal strength, pulse width, and Doppler frequency, providing comprehensive data for accurate obstacle detection.
As depicted in Fig. 2a, the LiDAR system can perform rotational measurements centered on the optical part, including rotations around the z-axis in the spherical coordinate system^89,100–102^. The Risley prism can rotate 360° in all directions, enabling the system to conduct comprehensive environmental scans. The entire LiDAR unit, including the Risley prism, can rotate 360° around the z-axis as presented in Fig. 2b,c. This dual-axis rotation capability allows a full 360° scan of the surrounding environment, ensuring no potential hazards are missed. This comprehensive scanning ability is crucial for applications in dynamic environments where obstacles and hazards may appear from any direction.
To ensure accurate measurements and minimize mutual interference, the system employs encryption identifiers modulated through OOFDMA to specify transmission directions. Optical coding techniques generate distinguishable pulse streams, with OOFDMA producing short pulse streams to overcome the limitations of traditional optical coding methods. The system uses AES encryption standards and randomly generated device and angle identifiers to secure data transmission, periodically renewing session keys to prevent unauthorized access. This robust security framework ensures that the data collected by the LiDAR system remains confidential and tamper-proof.
The adjustable Risley prism consists of three independently rotating wedge prisms of different sizes and slopes. This configuration allows the sensor to achieve a variable FoV. By adjusting these prisms’ angles and rotational velocities, the system can modify the scanning patterns and obstacle detection distances according to the specific region of interest (ROI). This function is crucial for adapting to different object sizes, distances, and distributions, making the system highly versatile and practical in various environmental conditions. The system can employ multiple LiDAR units simultaneously to enhance measurement resolution further, as shown in Fig. 3. By coordinating these units, the system achieves higher-resolution scans, allowing the detection of tiny objects, such as aerosols, that might be dispersed in the air. Each LiDAR unit can cover a segment of the FoV (Fig. 3a), and overlapping scans (Fig. 3b) can provide a composite high-resolution image. This method ensures that even the tiniest obstacles are detected, significantly enhancing the safety and reliability of UAM operations. The simultaneous operation of multiple LiDARs also increases data density, allowing for a more detailed and precise mapping of the environment. This approach is beneficial in detecting and characterizing small, dynamic obstacles that could pose a risk to UAM vehicles.
In real-world applications such as self-driving cars, autonomous robots, and UAM, detecting obstacles in the direction of travel is crucial^103,104^. The adjustable Risley prism allows the system to focus on specific directions precisely. By adjusting the distance between prisms or the angles of the internal prisms, the system can modify the scanning area to detect obstacles more effectively, especially in areas where obstacles are concentrated. Multiple LiDAR units can be synchronized to provide detailed obstacle maps, ensuring comprehensive detection and minimizing blind spots. This capability is particularly advantageous for detecting small airborne obstacles, such as aerosols, which are typically challenging to detect with a single LiDAR unit. The system’s ability to dynamically adapt its scanning patterns and densities ensures that even areas not measurable by one LiDAR can be covered by others, minimizing blind spots and enhancing overall detection accuracy.Fig. 1Operating concept of the proposed cycloid scanning LiDAR system.Fig. 2Single operation of cycloidal scanning LiDAR system (a) side view; (b) topview of 360° cycloidal scanning; and (c) sideview of 360° cycloidal scanning.Fig. 3Cooperative operation with multiple cycloidal scanning LiDARs (a) cooperative operation of cycloidal scanning LiDAR system; and (b) focusing on specific directions with high precision.Fig.4Targeted simulation environment for LiDAR aerosol detection in UAM applications.Fig. 5Scanning patterns of four LiDARs (a) VLS-128; (b) M1-R; (c) single cyclodial scanning LiDAR system; and (d) proposed cycloidal scanning LiDAR system.Fig. 6Detected aerosol distribution in calm atmospheric conditions (a) aerosol distribution; (b) VLS-128; (c) M1-R; (d) single cyclodial scanning LiDAR system; and (e) proposed cycloidal scanning LiDAR system.
To ensure the safe operation of UAM systems, a thorough understanding of atmospheric conditions, especially aerosols and low-level wind hazards^38,57^, is crucial. Our cyclodial scanning LiDAR system utilizes multiple LiDAR units working concurrently to provide detailed measurements, effectively overcoming the limitations of traditional single LiDAR setups. This system improves the detection of small particles and intricate wind patterns, which are essential for ensuring the safety and efficiency of UAM flights.
Under calm weather conditions, wind typically exhibits steady behavior, resulting in a consistent and slow movement of aerosols in a single direction. However, during wind hazards such as CAT, the distribution and movement of aerosols become significantly rapid and erratic. Traditional Doppler LiDAR systems, commonly used for weather observation, measure the movement of tiny particles like aerosols in the atmosphere using the Doppler effect and translate this movement into wind velocity data. For UAM operations, it is crucial to determine the presence of wind along the flight path, assess its strength, and understand its distribution. This information is vital for preemptive measures or rerouting to avoid adverse conditions.
Ground-based MPL systems emit laser beams vertically, which limits their ability to provide a comprehensive view of dust and aerosol distribution in the horizontal plane. This limitation means that even if strong winds are detected vertically, they do not necessarily correlate to the conditions affecting the horizontal flight path of the UAM. In contrast, our cyclodial scanning LiDAR system employs multiple units operating simultaneously to scan the airspace ahead of the UAM. By overlapping measurement areas, the system minimizes gaps, enhances the density and range of measurements, and offers a detailed and accurate representation of the atmospheric conditions in the direction of travel.
The conventional pulsed LiDAR systems employ distinct measurement zones for each LiDAR unit to prevent mutual interference, resulting in a trade-off between measurement range and density. This compromise hinders the detection of fine details in aerosol movement essential for understanding wind patterns. Conversely, the cyclodial scanning LiDAR system utilizes an advanced technique that enables multiple LiDAR units to operate simultaneously without interference, significantly enhancing the range and density of measurements. The heightened measurement density is concentrated in the central path of the UAM’s flight, ensuring a high-resolution scan of the immediate travel area. Simultaneously, the peripheral areas are scanned with an extended range, providing a comprehensive understanding of both the direction of travel and potential alternative paths. This dual focus enhances the UAM’s ability to navigate through or around hazardous conditions.
The operation of the cyclodial scanning LiDAR system involves emitting a laser beam in the direction of the UAM’s travel. As the beam travels, it interacts with aerosol particles in the atmosphere, leading to backscattering and reception by the LiDAR system. The time delay between the emission and reception of the laser beam is utilized to calculate the distance to the aerosol, while the intensity of the received laser beam aids in determining the aerosol’s size. The proposed system adopts the meteorological Doppler LiDAR method to analyze the intensity of the reflected wave and infer the aerosol size. By continuously calculating the distance and size of detected aerosols, the LiDAR system can detect small aerosols and infer wind movement, which is crucial for ensuring safe UAM operations. By employing multiple LiDAR units to scan the airspace ahead, the proposed system ensures high-resolution and comprehensive measurement of atmospheric conditions. This is essential for identifying low-level wind hazards and providing data for proactive UAM flight path adjustments, ensuring safe and efficient operations.
The cyclodial scanning LiDAR system, a true innovation in UAM technology, is designed to optimize the utilization of multiple LiDAR units operating concurrently. Unlike traditional LiDAR systems that allocate separate measurement zones to each unit to avoid interference, the proposed system allows for overlapping scanning areas by multiple LiDAR units. This groundbreaking approach enhances the measurement density and range, which is critical for detecting minute aerosols and a comprehensive understanding of intricate wind patterns. The cyclodial scanning LiDAR system provides a detailed and comprehensive environmental scan by increasing the measurement density and range. It focuses on scanning the center of the UAM’s path with higher density to ensure the precise detection of aerosols and wind patterns while scanning the periphery with an extended range to offer a broader view of the environment. This method enables the UAM to navigate safely by providing real-time data on atmospheric conditions, essential for avoiding hazards and ensuring efficient flight operations.
When the LiDAR system emits a laser beam in the direction of the UAM’s travel, the beam travels until it encounters aerosol particles in the atmosphere. The backscattering of the laser beam by these aerosols and its reception by the LiDAR system is used to calculate the distance to the aerosols. Furthermore, the intensity of the received laser beam aids in determining the size of the aerosols. This process involves several Emission of laser The LiDAR system emits a laser beam in the direction of the UAM’s travel.Interaction with The laser beam travels through the atmosphere and encounters aerosol particles.Backscattering of laser Upon collision with the aerosols, part of the laser beam is backscattered while the remainder continues forward.Reception of backscattered The backscattered laser beam is received by the LiDAR system.Calculation of distance and The time difference between emission and reception is used to calculate the distance to the aerosol, and the intensity of the received beam is analyzed to determine the aerosol’s size.The cyclodial scanning LiDAR system, a safety cornerstone in UAM operations, plays a crucial role in ensuring the safe operation of UAMs by detecting aerosols and inferring wind movement. It provides detailed and accurate measurements of atmospheric conditions, enabling UAMs to adjust flight paths and prepare for potential hazards proactively. This capability is not just beneficial, it’s essential for maintaining safety and efficiency in UAM operations. The comprehensive scanning capability of the cyclodial scanning LiDAR system, achieved by using multiple LiDAR units to scan the airspace ahead, offers several key High measurement Multiple LiDAR units ensure high measurement density, crucial for detecting small aerosols and understanding detailed wind patterns.Extended Using multiple LiDAR units increases the overall measurement range, which is the maximum distance at which the system can detect and measure aerosols. This extended range provides a more comprehensive detection area, allowing the LiDAR system to scan a larger portion of the airspace ahead and provide more advanced warning of potential hazards to the UAM.Enhanced Detailed and accurate detection of wind hazards enables UAMs to take preemptive measures or adjust flight paths to avoid adverse conditions, significantly enhancing operational safety.The cyclodial scanning LiDAR system, a game-changer in UAM navigation, offers real-time data on atmospheric conditions, allowing for more efficient navigation. This reduces the risk of encountering unexpected hazards and paves the way for a more streamlined and efficient UAM operation. The future of UAM systems is brighter with this technology in place. The cyclodial scanning LiDAR system significantly advances the detection of aerosols and low-level wind hazards for UAM operations. By utilizing multiple LiDAR units operating concurrently, the system provides detailed and high-density measurements of atmospheric conditions, ensuring the safety and efficiency of UAM flights. This technology enhances the ability to detect and respond to potential hazards, contributing to the overall reliability and effectiveness of urban air transportation and paving the way for the future of UAM systems. The features and differences between wind lidar and our proposed cycloidal scanning lidar are thoroughly compared and summarized in the Table 1^1,52–59^.Table 1Comparative analysis of on-board cycloidal scanning LiDAR with wind LiDAR.ItemProposed cycloidal scanning LiDARWind LiDARReal-time data processingThe system is a robust safety solution, leveraging real-time data processing to detect and analyze wind hazards promptly. Its high processing speed empowers UAM vehicles to adapt their flight paths, improving safety quickly. Advanced algorithms enable the system to efficiently handle large amounts of data, consistently delivering up-to-date information for informed decision-making.While capable of providing real-time data, Wind LiDAR typically lacks the advanced processing speed and algorithms found in the cycloidal system. This can result in slower detection and response times, compromising safety in fast-changing environments.On-board integrationTailored for seamless integration with UAM navigation and control systems, this LiDAR system effortlessly provides real-time data directly to the vehicle’s on-board systems. This cohesive integration ensures instant hazard detection and navigational adjustments, enabling proactive safety measures. The cycloidal scanning LiDAR can also gather data from any point along the flight path, offering continuous real-time updates on environmental conditions. This constant awareness of its surroundings allows the UAM vehicle to adjust its dynamic flight path in response to changing environmental conditions.Wind LiDAR systems typically function as fixed installations on the ground, limiting data collection to specific locations or designated areas. Although they offer valuable data within their range, they cannot provide continuous real-time updates for a moving UAM vehicle. These systems are highly effective for measuring wind speed, direction, and turbulence from the ground to several kilometers in altitude, making them well-suited for weather forecasting and wind energy assessment applications. They can provide detailed information on wind speed and direction at various altitudes, making them valuable for studying atmospheric dynamics and weather patterns.Comprehensive environmental scanningThe system features 360° rotational capabilities, ensuring complete environmental scanning and minimizing blind spots. This comprehensive real-time scanning capability is crucial for maintaining situational awareness and detecting wind hazards from all directions.Wind LiDAR systems can simultaneously measure wind speeds and directions with high spatial resolution, offering 2D maps of wind speed over large areas. This makes them ideal for studying complex wind patterns and atmospheric dynamics, although their fixed scanning area may only cover part of the environment as thoroughly as the cycloidal system.High-resolution real-time mappingThis system offers high spatial resolution and scans rapidly at 50 Hz, creating detailed real-time maps of specific target areas. This high level of detail is crucial for identifying minor and rapidly changing wind hazards, enabling precise and immediate responses.While this system generally provides lower resolution and slower scanning rates compared to the cycloidal system, its capability to generate high-resolution 2D maps of wind speed and direction over large areas is invaluable for detailed atmospheric analysis, despite being less effective in a fast-moving UAM context.System complexityIntegrating multiple LiDAR units and complex real-time processing algorithms increases the system’s complexity, leading to higher costs, more demanding maintenance, and the need for specialized technical expertise.Wind LiDAR systems have a generally more straightforward design and operation, and they are less expensive and easier to maintain. Their simplicity provides a significant advantage in terms of operational efficiency and cost-effectiveness.Power consumption, size, and weightDesigned to be lightweight and consume less power, it is ideal for on-board integration in UAM vehicles. The compact size and lower power requirements are critical for maintaining the vehicle’s performance and efficiency. However, these design constraints typically limit the system to measuring shorter distances than ground-based systems. This means that while the cycloidal system is highly efficient for short-range scanning, it may not be as effective for long-range environmental scanning.As a ground-based system, Wind LiDAR faces fewer constraints regarding size, weight, and power consumption. This allows it to incorporate more significant and more powerful components that can measure wind conditions over much greater distances and provide detailed data from multiple altitudes. For example, a Wind LiDAR system can measure wind conditions from the ground up to several kilometers in altitude, providing a comprehensive view of the atmospheric conditions. The lack of size and weight restrictions enables Wind LiDAR to use more powerful sensors and processors, enhancing its capability to effectively cover extensive areas and heights.
The on-board cycloidal scanning LiDAR system provides superior real-time data processing, comprehensive environmental scanning, and high-resolution mapping capabilities, all crucial for enhancing UAM safety. Its ability to collect data anywhere along the flight path ensures continuous real-time updates, providing a significant advantage over fixed Wind LiDAR systems. Wind LiDAR systems offer significant advantages in providing detailed wind speed and direction information over large areas with high spatial resolution. They are essential for applications such as weather forecasting, wind energy assessment, and studying atmospheric dynamics. Their fixed installation limits data collection to specific areas, making them a practical choice for these specific applications. When selecting the appropriate system, it’s crucial to balance the need for advanced real-time capabilities with practical considerations of power, complexity, and integration constraints specific to UAM applications. This informed decision-making process ensures that the chosen system meets the unique needs of the UAM industry.
LiDAR systems are widely used in automotive settings, as they enable the generation of detailed three-dimensional maps of the vehicle’s surroundings. The generic LiDAR range equation, which describes the power of the received signal, is fundamental to understanding and optimizing LiDAR performance. The general form of the LiDAR range equation^105^ is expressed 1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} P_r = P_t \cdot \frac{G_t \cdot A_r \cdot \sigma \cdot \eta _t \cdot \eta _r \cdot \eta _a}{(4\pi R^2)^2} \end{aligned}
\usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ P_r $$\end{document} denotes the received power. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ P_t $$\end{document} represents the transmitted power. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ G_t $$\end{document} is the transmitter gain. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ A_r $$\end{document} is the receiver aperture area. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \sigma $$\end{document} signifies the target cross-sectional area (reflectivity of the target). *R* denotes the range of the target. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \eta _t $$\end{document} is the transmitter efficiency. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \eta _r $$\end{document} is the receiver efficiency. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \eta _a $$\end{document} represents the atmospheric transmission efficiency. Transmitted power (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ P_t $$\end{document}) refers to the power output of the LiDAR system’s laser source, typically operating in the near-infrared spectrum for automotive applications, chosen for its eye safety and optimal atmospheric transmission characteristics. Transmitter gain (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ G_t $$\end{document}) reflects the gain of the transmitting antenna or optics, which focuses the emitted laser beam into a narrow, high-intensity pulse, enhancing range and accuracy. The receiver aperture area (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ A_r $$\end{document}) denotes the effective area of the LiDAR receiver that captures the backscattered laser light. A larger aperture increases the amount of light collected, enhancing signal strength and detection sensitivity. The target cross-sectional area (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \sigma $$\end{document}) represents the reflectivity and size of the target object. Different materials and surfaces reflect varying amounts of laser light, affecting the intensity of the received signal. Range to target (*R*) denotes the distance from the LiDAR system to the target. The received signal power decreases with the square of this distance, highlighting the challenge of long-range detection. Efficiencies terms \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ ({{\eta _{t}}, \; {\eta _{r}}, \; {\eta _{a}}}) $$\end{document} account for the losses in the transmission, reception, and atmospheric propagation of the laser signal. Higher efficiencies indicate a more effective system with minimal signal loss. In meteorology, LiDAR systems investigate atmospheric particles (aerosols) by measuring the backscattering and extinction coefficients^106–109^. These coefficients offer crucial insights into the concentration and distribution of aerosols, which are essential for comprehending weather patterns, air quality, and climate change. The backscattering coefficient quantifies the amount of laser light scattered back toward the LiDAR system by aerosols in the atmosphere. It depends on the aerosols’ number density, size distribution, and refractive index.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \beta (R) = N(R) \cdot \sigma _{bs}(R) \end{aligned} $$\end{document}Where: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \beta (R) $$\end{document} represents the backscattering coefficient at range *R*. *N*(*R*) denotes the number density of aerosols at range *R*. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \sigma _{bs}(R) $$\end{document} signifies the backscattering cross-section of the aerosols at range *R*. The extinction coefficient signifies the total loss of laser light due to both scattering and absorption by aerosols as the light travels through the atmosphere.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \alpha (R) = N(R) \cdot \sigma _{ext}(R) \end{aligned} $$\end{document}Where: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \alpha (R) $$\end{document} denotes the extinction coefficient at range *R*. *N*(*R*) denotes the number density of aerosols at range *R*. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \sigma _{ext}(R) $$\end{document} signifies the extinction cross-section of the aerosols at range *R*. To gain a deeper understanding, the LiDAR equation for aerosol detection can be expressed by integrating the backscattering and extinction coefficients. The modified LiDAR equation is given 4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} P_r(r) = P_t \cdot \frac{G_t \cdot A_r \cdot \beta (R) \cdot \eta _t \cdot \eta _r \cdot e^{-2 \int _0^R \alpha (s) ds}}{(4\pi R^2)^2} \end{aligned} $$\end{document}In this \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ P_r(R) $$\end{document} represents the received power at range *R*. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ e^{-2 \int _0^R \alpha (s) ds} $$\end{document} is the atmospheric transmission term, accounting for the two-way attenuation of the laser signal due to extinction. The backscattering coefficient (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \beta (r) $$\end{document}) indicates the fraction of the incident laser power scattered back towards the LiDAR receiver by aerosols, depending on the physical properties of the aerosols. Extinction coefficient (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \alpha (r) $$\end{document}) quantifies the attenuation of the laser beam as it travels through the atmosphere due to scattering and absorption by aerosols. The integral of the extinction coefficient over the path length accounts for the cumulative attenuation of the laser signal. Atmospheric transmission \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \left( e^{-2 \int _0^r \alpha (s) ds}\right) $$\end{document} represents the exponential decay of the laser signal as it propagates to the target and back due to the combined effects of scattering and absorption. This term is vital for accurate range and intensity measurements in aerosol-laden atmospheres. The LiDAR equation used in automotive applications is primarily focused on directly measuring the range and reflectivity of targets to create detailed environmental maps. In meteorology, the emphasis shifts to understanding atmospheric aerosols through the backscattering and extinction coefficients, offering insights into aerosol concentration and distribution. Both applications rely on precise mathematical formulations to optimize performance and interpret the data collected by LiDAR systems. ## Visual perception of wind hazards with LiDAR system In the pursuit of ensuring the utmost safety in UAM systems, the LiDAR system plays a pivotal role. Its ability to detect aerosols and visial perceive low-level wind hazards effectively is crucial. The system’s capability to detect objects approximately 2 km ahead provides ample range for timely obstacle identification and avoidance. High spatial resolution (at least 0.1° horizontally and vertically) is necessary to accurately distinguish between small and large obstacles. Operating in the near-infrared spectrum (905 nm or 1550 nm) ensures an optimal balance between eye safety and atmospheric penetration. The system also requires a fast scanning rate of at least 50 Hz to capture rapid movements of small objects like birds and a high point cloud density of at least 300 points per square meter for detailed obstacle mapping. Incorporating Doppler LiDAR capabilities will enable the visual perception of wind speed and direction by measuring the Doppler shift with a minimum velocity resolution of 0.1 ms^−1^ for accurate aerosol speed detection. Furthermore, the LiDAR system’s role in enhancing UAM safety and efficiency is undeniable. Its robust real-time data processing capabilities, including algorithms for identifying and classifying different obstacles, are instrumental. The system’s ability to calculate aerosols’ backscattering and extinction coefficients provides a comprehensive understanding of atmospheric conditions and potential hazards. Advanced signal processing ensures performance in adverse weather conditions. Given the power constraints of UAM platforms, power efficiency is a critical consideration. Seamless integration with the UAM’s navigation and control systems optimizes decision-making. These specifications significantly enhance UAM safety and efficiency by robustly detecting small, fast-moving obstacles and larger, slower-moving wind hazards. ### Operation strategies of four LiDARs We have evaluated the performance of a groundbreaking cycloidal scanning LiDAR system designed specifically for UAM applications through extensive simulations. This innovative system was compared with the renowned Velodyne VLS-128 automotive LiDAR^110^ and Neuvition’s solid-state LiDAR Titan M1-R^111^, serving as our benchmarks. The Velodyne VLS-128 is known for its high channel count, providing a dense point cloud and exceptional lateral angular resolution, making it a powerful tool for automotive applications. In the context of our study, we made critical adjustments to the specifications of the Velodyne VLS-128 to extend its maximum measurement range to 2 km, aligning it with the requirements of UAM operations. This adaptation, however, presented several challenges. The Velodyne VLS-128 traditionally performs distance measurements sequentially, which works efficiently within the shorter ranges typical of automotive applications. Similarly, the Neuvition Titan M1-R, a solid-state LiDAR, offers a non-mechanical and compact design that ensures durability and robustness in demanding environments. Unlike mechanical systems, the Titan M1-R is capable of wide-area scanning without moving parts, with high resolution and fine angular precision, making it an excellent candidate for applications requiring reliability and minimal maintenance, such as UAM. However, the performance of both the VLS-128 and Titan M1-R is constrained when operating over extended ranges, particularly in maintaining resolution and scanning efficiency at distances up to 2 km. One significant challenge observed with these LiDAR systems is the increase in dwell time-the duration the LiDAR system spends measuring each point. Dwell time is directly proportional to the distance being measured; as the distance increases, the time taken for the laser beam to travel to the object and back also increases, thereby affecting the overall measurement time. The system requires more time to gather sufficient data for accurate distance calculation. The dwell time increases significantly with the extended range of the VLS-128 and Titan M1-R to 2 km. This extended dwell time has a cascading effect on other performance metrics. Specifically, it negatively impacts the lateral angular resolution and range resolution. The lateral angular resolution, which determines the system’s ability to distinguish between two closely spaced objects in the horizontal plane, deteriorates as the system spends more time on each measurement point, reducing the frequency of data collection across the FoV. Similarly, the range resolution, or the ability to distinguish between two objects at different distances, suffers due to the increased time per measurement. Maintaining the same rotations per minute (RPM) while extending the range also presents challenges. The rotating mechanism of the VLS-128 is designed to balance speed and measurement density, but increasing the measurement range means each rotational scan covers a larger area. This results in sparser data collection, which can be problematic for detecting small or fast-moving objects like aerosols in the atmosphere or potential wind hazards. In contrast, the cycloidal scanning LiDAR system utilizes OOFDMA to encode positional information into the laser beam. This technique allows the LiDAR to transmit and receive multiple data streams simultaneously. The OOFDMA approach encodes the measurement location information directly into the laser beam, which is then transmitted. Upon encountering an object, the laser beam is backscattered and received by the LiDAR system, which decodes the positional information to determine the distance and position of the object. The utilization of OOFDMA provides numerous significant benefits. It allows the LiDAR system to maintain consistent performance regardless of the maximum measurement range. Even with an extended range of up to 2 km, the OOFDMA-based LiDAR does not suffer from the same degradation in lateral angular resolution and range resolution observed in traditional sequential measurement systems like the VLS-128 and Titan M1-R. This consistency is achieved because OOFDMA enables the system to conduct simultaneous measurements across multiple locations rather than sequentially-a feature that significantly enhances its performance in UAM applications. Additionally, the cycloidal scanning LiDAR system’s performance was evaluated in two (1) a configuration where multiple units were clustered to focus their FoV on specific regions, and (2) a configuration where a single unit operated solely utilizing non-repetitive scanning patterns. This dual evaluation demonstrated the system’s adaptability to both broad-area environmental monitoring and focused hazard detection. The system’s ability to dynamically adjust its scanning strategy ensures superior performance in detecting small, fast-moving objects, such as birds, as well as slower-moving aerosols. Figures 10, 11, 12 and Table 3 illustrate these results, providing comprehensive insights into the system’s performance in detecting wind hazards and ensuring safety in UAM operations. A detailed comparison with existing automotive LiDAR technology is essential to assess the cycloidal scanning LiDAR system’s performance for UAM applications, particularly for aerosol detection and low-level wind hazard identification. Using the provided Table 2, we evaluate and elucidate the scientific and technical specifications of the cycloidal scanning LiDAR system in contrast to the VLS-128 and Titan M1-R, including their adapted versions for UAM.Table 2Operating characteristics of four LiDARs.ItemVLS-128 for UAM^110^M1-R for UAM^111^Single cycloidal scanning LiDAR^31,32^Proposed four-unit cycloidal scanning LiDAR^31,32^Beam steering and coordination mechanismBody rotationOpticalphased arrayBody rotation with Risley prismBody rotation with Risley prism and cooperative across four unitsRotation per minute1200120012001200Lateral angular resolution5.1168°0.14°0.002°0.002°Dwell time710.6 μs200 μs0.5 μs0.5 μsField-of-view Azimuth40°15°360°360° Elevation40°8°40°40° PropertyFixedFixedVariableVariableLaser beam FiringOne pulseOne pulseOOFDMA coded pulsesOOFDMA coded pulses Wavelength903 nm1550 nm903 nm to 1568.3623 nm903 nm to 1568.3623 nm Channel128700128128 OperationTransmit wait receiveTransmit wait receiveConcurrently transmit and receiveConcurrently transmit and receiveInterferencePresentPresentAbsentAbsentNumber of LiDAR units1114Overlapping scanN/AN/AN/APossible ### Detailed environmental and operational description for LiDAR aerosol detection in UAM applications To ensure the safe and efficient operation of UAM systems, it is vital to utilize a LiDAR system that can effectively detect aerosols and visual perceive low-level wind hazards. The operational environment for these systems is defined by a specific detection volume extending 1660–2020 m forward from the UAM, with a height ranging from 90 m to 810 m and a lateral width of − 540 m to 540 m, as illustrated in Fig. 4. This detection area is designed to cover the UAM’s flight path and potential avoidance maneuvers in the presence of wind hazards. Extensive simulations were conducted to evaluate the cyclodial scanning LiDAR system’s performance compared to the Velodyne VLS-128 and Neuvition Titan M1-R, a well-regarded automotive LiDAR system adapted for UAM applications. The simulations included various atmospheric conditions, from calm weather with no wind hazards to scenarios featuring turbulent eddies, gust, and wind shear. These varied conditions comprehensively assessed each LiDAR system’s performance under different operational environments. The simulations utilized RSoft OptSim^31,32,99^ to model the optical properties of the LiDAR systems, including the waveform, intensity, wavelength, and spacing of the transmitted and received laser beams. MATLAB, integrated with computational fluid dynamics (CFD)^42,112–115^, was employed to simulate the characteristics of aerosols and their interactions with the laser beams, mainly focusing on the effects of backscattering. The standard configuration of the Velodyne VLS-128 utilizes body rotation for beam steering and a single pulse mechanism, operating on a transmit-wait-receive cycle. This approach enables the system to complete a distance measurement at one point before moving to the next. With a maximum detection distance of 150 m, a range resolution of 0.02 m, and a lateral angular resolution of 0.38376°, the VLS-128 is well-suited for close-range obstacle detection in automotive applications. Operating at 1200 RPM with a dwell time of 53.3 μs, it consumes 20 W of power. Adapting the VLS-128 for UAM applications required extending its maximum detection range to 2 km, which led to several performance compromises-the increased dwell time to 710.6 μs notably degraded the lateral angular resolution to 5.1168° and the range resolution to 0.2 m. This longer dwell time reduced the system’s ability to capture high-resolution data quickly, making it less effective for detecting small, fast-moving objects like aerosols and accurately mapping low-level wind hazards. The power consumption surged to 720 W, reflecting the higher energy demands for long-range detection. The Neuvition Titan M1-R, a solid-state LiDAR, was originally designed for shorter-range applications with precise specifications. Before modification, it featured an angular resolution of 0.01° both horizontally and vertically, a FoV of 15° horizontally and 8° vertically, a maximum measurement distance of 300 m, and operated at 60 RPM. To adapt the Titan M1-R for a maximum measurement range of 2 km, significant changes were made to its configuration. The angular resolution was reduced to 0.14° in both horizontal and vertical directions to maintain coverage and scanning efficiency over the extended range. The maximum measurement range was increased from 300 meter to 2 km, necessitating a substantial boost in rotational speed to 1800 RPM to ensure sufficient point density despite the extended range. While the field of view remained the same at 15° horizontally and 8° vertically, these adjustments enabled the Titan M1-R to operate effectively at 2 km. However, these changes also introduced trade-offs, including a sparser point cloud and reduced fine detail, particularly for small or fast-moving objects. Despite these limitations, the adapted Titan M1-R provides a valuable benchmark for evaluating the cycloidal scanning LiDAR system in UAM-specific scenarios. The cyclodial scanning LiDAR system utilizes OOFDMA and a Risley prism for beam steering. This innovative approach allows the LiDAR to embed positional information directly into the laser beam, facilitating simultaneous transmission and reception of multiple data streams. This concurrent measurement capability significantly enhances the system’s efficiency and accuracy. The cyclodial scanning LiDAR maintains a maximum detection distance of 2 km while preserving high-resolution performance. It offers a range resolution of 0.1 m and an exceptionally refined lateral angular resolution of 0.002°. These specifications ensure detailed and accurate detection of aerosols and low-level wind hazards, which are crucial for UAM operations. The dwell time is minimal at 0.5 μs, enabling rapid and precise measurements across the FoV. Operating at 1200 RPM, the system ensures comprehensive coverage and real-time data acquisition while consuming only 50 W of power. The cyclodial scanning LiDAR system incorporates advanced signal processing capabilities, including real-time data processing and robust obstacle detection and visual perception algorithms. These features allow the system to quickly interpret data, providing timely and actionable information for hazard avoidance and navigation adjustments. A novel method proposed for this system involves the use of four cyclodial LiDAR units overlapping their scanning FoV on a specific region to achieve highly focused and dense environmental monitoring. This configuration ensures maximum detail and reliability in detecting critical hazards, such as aerosols and low-level wind patterns, which are essential for safe UAM operations. Additionally, the cyclodial scanning LiDAR system’s performance was compared against several configurations to evaluate its adaptability and efficiency. These included a single cyclodial LiDAR unit operating independently with non-repetitive scanning patterns, the modified Velodyne VLS-128, and the Neuvition Titan M1-R. The single cyclodial LiDAR unit demonstrated its capability to provide wide-area coverage, while the VLS-128 contributed high-precision scanning for shorter ranges. The Titan M1-R, adapted for a maximum measurement range of 2 km, provided a benchmark for extended-range applications, complementing the cyclodial system’s versatility and highlighting its superiority in handling dynamic and complex UAM environments. Another significant advantage of the cyclodial scanning LiDAR system is its ability to mitigate interference through advanced signal processing techniques, ensuring reliable data even in challenging atmospheric conditions. Its power efficiency and high-resolution performance position it as a superior solution to traditional automotive LiDAR systems, making it ideally suited for the dynamic and complex environment of UAM operations. The detailed simulation and analysis highlight the critical advantages of the proposed OOFDMA-based LiDAR system with Risley prisms over the single-unit cyclodial LiDAR configuration, the modified Velodyne VLS-128, and the Titan M1-R. Its ability to maintain high-resolution performance, efficient power usage, and advanced data processing capabilities makes it an ideal solution for detecting aerosols and identifying low-level wind hazards in UAM applications. This ensures robust and reliable performance, enhancing the safety and operational efficiency of UAM systems. By leveraging its proposed method of using four units with overlapping FoVs alongside complementary benchmarking with single-unit operations, VLS-128, and Titan M1-R, the cyclodial scanning LiDAR system provides a scalable and efficient solution for UAM. These combined capabilities enable safer and more effective navigation in complex and rapidly changing environments, establishing it as a benchmark for LiDAR technology in urban air mobility. The detection area for the Velodyne VLS-128, Titan M1-R and the cyclodial scanning LiDAR system covers a range of 2 km forward, 90 m to 810 m in height, and 540 m to either side. This extensive detection volume is critical for effectively monitoring the flight path of the UAM and identifying potential hazards. The process begins by emitting laser beams from the LiDAR systems towards the designated detection area. As these laser beams travel through the atmosphere, they encounter aerosols. When a laser beam strikes an aerosol, backscattering occurs, which involves the reflection of the laser light back toward the LiDAR receiver. This backscattering phenomenon is essential for detecting the presence of aerosols. The intensity of the backscattered laser beam varies depending on the composition and size of the aerosols. Different aerosol particles reflect laser light with different intensities, allowing for the distinction between different aerosols. By analyzing the backscattered signals, the LiDAR system can deduce the characteristics of the aerosols. Using the previously discussed LiDAR equations, the received power of the backscattered laser beam is used to calculate the size of the aerosols. The formula considers factors such as the transmitted power, the target cross-sectional area (reflectivity of the aerosols), and the target range. By inputting the intensity of the received laser beam into these equations, the LiDAR system can accurately determine the size of the detected aerosols. ### Comparison of scanning patterns across four LiDAR systems Figure 5 compares the scanning patterns of the Velodyne VLS-128, Titan M1-R, and the cyclodial scanning LiDAR system, highlighting their effectiveness in detecting aerosols and identifying potential low-level wind hazards in UAM operations. The black dots in the figure represent the scanning positions of each LiDAR system, illustrating the spatial coverage and density of their respective scanning patterns. Figure 5a shows the scanning pattern of the Velodyne VLS-128. Its sequential scanning mechanism, achieved through 128 vertically distributed channels, produces a relatively sparse and uniform grid of measurement points. While this pattern provides a general overview of the detection area, its lower density limits its ability to detect small-scale aerosols or turbulence. The large angular separation between channels and reliance on sequential measurements also restricts its precision in capturing fine atmospheric details, making it less suitable for dynamic UAM environments where real-time hazard detection is critical. Figure 5b illustrates the Titan M1-R’s scanning pattern. As a solid-state LiDAR, it achieves dense coverage through high-speed, non-mechanical scanning. The pattern appears uniform and detailed, making it effective for medium-range detection of aerosols or objects. However, when adapted for extended ranges of up to 2 km, the pattern becomes less dense, reducing its ability to capture small or fast-moving particles. This limitation impacts its effectiveness in detecting fine-scale wind hazards, such as gusts or eddies, which are vital for UAM safety. Figure 5c represents the scanning pattern of a single cyclodial scanning LiDAR unit operating independently. This unit utilizes OOFDMA-coded pulses and a Risley prism to achieve a non-repetitive, high-density scanning pattern. Compared to the VLS-128 and Titan M1-R, the single cyclodial LiDAR provides better coverage and resolution, even at extended ranges. Its ability to dynamically adjust the field of view ensures flexibility in monitoring changing atmospheric conditions. This pattern allows the system to detect fine-scale aerosols and turbulence more effectively than the VLS-128 or Titan M1-R, although its coverage remains inherently more limited than multi-unit configurations. Figure 5d showcases the scanning pattern of the proposed cyclodial scanning LiDAR system using four overlapping units. Each unit employs OOFDMA-coded pulses and a Risley prism, creating an interwoven and highly dense pattern. This overlapping configuration eliminates gaps in the scanning area, significantly enhancing the measurement density and precision. The high density of scanning points improves the system’s ability to detect small-scale aerosols and identify low-level wind hazards, such as eddies and gusts. This level of detail is critical for UAM operations, where real-time and accurate atmospheric monitoring is necessary to ensure safe navigation. The overlapping pattern also provides redundancy, reducing the likelihood of missed detections, even in challenging environments. The comparison across these systems highlights the distinct advantages of the proposed four-unit cyclodial scanning LiDAR system in terms of resolution, coverage, and adaptability. The Velodyne VLS-128 and Titan M1-R provide basic coverage suitable for general environmental monitoring; however, their scanning densities are limited, making them less effective for detailed atmospheric analysis. The VLS-128, with its fixed vertical channels and relatively large angular separation, creates a sparse grid of data points that struggles to capture small-scale aerosols or subtle wind patterns. Similarly, the Titan M1-R, while offering a more uniform and denser coverage than the VLS-128 due to its solid-state scanning mechanism, still lacks the resolution and adaptability required for fine-grained detection, especially at extended ranges of up to 2 km. In contrast, the single cyclodial LiDAR unit demonstrates a significant improvement by achieving higher density and flexibility in its scanning pattern. Its ability to dynamically adjust its field of view and utilize non-repetitive scanning patterns allows it to detect fine aerosols and subtle atmospheric turbulence with greater accuracy. However, its single-unit configuration inherently limits its coverage, leaving gaps in the scanning area, which can reduce the system’s reliability in highly dynamic UAM environments. The proposed cyclodial scanning LiDAR system builds on the strengths of the single unit and eliminates its limitations by overlapping the scanning fields of all four units. This configuration produces an interwoven and ultra-dense scanning pattern, ensuring comprehensive coverage without any gaps. The system’s high resolution enables it to detect and characterize even the smallest aerosols, identify complex wind patterns, and monitor turbulence such as eddies and gusts. This level of detail is crucial for UAM operations, where small-scale atmospheric variations can significantly impact flight safety. Furthermore, the overlapping scanning pattern provides redundancy, enhancing reliability and reducing the chances of missed detections even in challenging atmospheric conditions. This ability to combine unmatched resolution, dense coverage, and adaptability makes it uniquely capable of addressing the challenges posed by dynamic atmospheric conditions in UAM environments. By providing real-time, high-precision data, it ensures enhanced safety, efficient hazard avoidance, and improved operational efficiency, setting a new standard for LiDAR technology in aerial mobility applications. ### Aerosol detection results in calm atmospheric conditions Figure 6 compares the aerosol detection results obtained from the Velodyne VLS-128, the Neuvition Titan M1-R, a single cyclodial scanning LiDAR system, and the proposed cyclodial scanning LiDAR system under calm atmospheric conditions, devoid of low-level wind disturbances. The detection area spans a forward distance of 2 km, with a vertical range of 90 m to 150 m and a width of 540 m on each side, providing comprehensive coverage of the UAM operational environment. Figure 6a illustrates the aerosol distribution in calm atmospheric conditions, where the absence of wind ensures the aerosols are uniformly dispersed without forming any discernible patterns or clusters. The colored dots in each subfigure represent the detected aerosols, with the size of the dots proportional to the intensity of the backscattered signal. This intensity generally correlates with the size of the aerosol particles, though clusters of smaller particles in close proximity can also produce larger signal intensities, creating the appearance of larger individual particles. Figure 6b illustrates the aerosol detection results from the Velodyne VLS-128. The sparse and uniform distribution of data points reflects the system’s inherent limitations in dwell time and resolution, which constrain its ability to detect small or closely spaced aerosols. The measurements are evenly scattered across the detection area, consistent with the calm atmospheric conditions, but lack the density required for detailed aerosol characterization. The VLS-128 primarily captures general aerosol distributions but struggles with capturing fine details or subtle variations in aerosol density. Figure 6c presents the results from the Neuvition Titan M1-R. Compared to the VLS-128, the Titan M1-R demonstrates slightly higher measurement density due to its solid-state scanning capabilities. However, its performance is limited at extended ranges, resulting in sparser data points farther away from the sensor. The uniform scattering of points indicates calm conditions, yet the Titan M1-R provides better representation of aerosol clusters than the VLS-128, though still falling short of achieving high-resolution atmospheric analysis. Figure 6d shows the aerosol detection results from a single cyclodial scanning LiDAR system. The system’s non-repetitive scanning pattern, enabled by its advanced beam steering with a Risley prism, creates a unique distribution of data points. While it achieves higher measurement density and improved resolution compared to the VLS-128 and Titan M1-R, its single-unit configuration results in some gaps in coverage. Interestingly, due to the scanning pattern, the detected aerosols sometimes form shapes that can visually resemble the presence of wind patterns, even in calm conditions. This can create a misleading impression of airflow or turbulence when none exists, underscoring the importance of interpreting results with knowledge of the scanning methodology. Figure 6e depicts the aerosol detection results from the proposed cyclodial scanning LiDAR system. By overlapping the fields of view of the four units, this configuration achieves unparalleled measurement density and resolution. The dense and interwoven pattern of data points eliminates gaps and provides a highly detailed representation of aerosol distribution. The system excels in detecting small aerosols and capturing clusters with exceptional accuracy, offering a level of detail crucial for UAM operations. Despite the calm atmospheric conditions, the high density of data points ensures a comprehensive understanding of aerosol characteristics, setting it apart from the other systems. The comparison of aerosol detection results across the four LiDAR systems-Velodyne VLS-128, Neuvition Titan M1-R, single cyclodial scanning LiDAR unit, and the proposed cyclodial scanning LiDAR system-highlights significant differences in their measurement density, resolution, and coverage. The VLS-128 demonstrates sparse data points with low resolution due to its fixed vertical channels and sequential scanning, making it suitable only for general aerosol distribution but inadequate for detailed atmospheric analysis. The Titan M1-R improves upon this with faster solid-state scanning and better density, but its performance declines at extended ranges, limiting its ability to detect fine aerosols. The single cyclodial LiDAR system further enhances measurement density and resolution using non-repetitive scanning patterns enabled by its Risley prism and OOFDMA-coded pulses, capturing smaller aerosols and clusters more effectively, though it lacks complete coverage and can create visual artifacts under calm conditions. In contrast, the proposed cyclodial scanning LiDAR system excels by overlapping the fields of view of its four units, producing an interwoven and highly dense dataset that eliminates gaps, enhances resolution, and provides unparalleled accuracy in detecting fine aerosols and clusters. This comprehensive coverage and high resolution make the proposed system the most effective for UAM operations, enabling precise atmospheric monitoring and ensuring enhanced safety and efficiency in dynamic environments. ### Aerosol detection results in eddy turbulence conditions Fig. 7Detected aerosol distribution in eddy turbulence conditions (**a**) aerosol distribution; (**b**) VLS-128; (**c**) M1-R; (**d**) single cyclodial scanning LiDAR system; and (**e**) proposed cycloidal scanning LiDAR system. Figure 7 presents the aerosol detection results obtained from four different LiDAR systems-Velodyne VLS-128, Neuvition Titan M1-R, a single cyclodial scanning LiDAR unit, and the proposed cyclodial scanning LiDAR system under eddy turbulence conditions^33,116,117^. Figure 7a illustrates the aerosol distribution in an environment characterized by eddy turbulence, a condition where swirling airflow redistributes aerosols unevenly. These evaluations were conducted in a controlled setting to simulate low-level wind disturbances commonly encountered in UAM operations. Under such conditions, aerosols concentrate densely in the central swirling regions of the turbulence and disperse outward to the surrounding areas, forming a distinctive vortex-like pattern. The distribution provides insight into the dynamics of aerosol movement and the spatial variability caused by turbulent airflow. The size of the dots corresponds to the intensity of the backscattered signal, which depends on the size of the aerosol particles and sometimes clusters of smaller particles in close proximity. This detailed visualization serves as a foundation for evaluating the performance of different LiDAR systems in detecting and characterizing such complex atmospheric patterns. Figure 7b shows the results from the Velodyne VLS-128, where the detected aerosols are unevenly distributed across the detection volume. The swirling eddy pattern is weakly captured due to the VLS-128’s fixed vertical channel arrangement and relatively low resolution. The sparse data points fail to delineate the eddy turbulence pattern clearly, and the lack of density in the measurements limits the system’s ability to identify and characterize the turbulence effectively. The results indicate the presence of atmospheric disturbances but do not provide sufficient detail to confirm the precise shape or characteristics of the eddy turbulence. Figure 7c illustrates the aerosol detection pattern of the Neuvition Titan M1-R. The Titan M1-R demonstrates improved measurement density compared to the VLS-128, capturing more aerosol clusters and providing a somewhat clearer indication of swirling motions within the eddy turbulence. However, its detection capability at extended ranges remains limited, and while it captures the general structure of the turbulence better than the VLS-128, it still lacks the resolution and density required for a detailed analysis of the eddy pattern. Figure 7d depicts the results from a single cyclodial scanning LiDAR unit. The system’s non-repetitive scanning pattern captures a higher density of data points than the VLS-128 and Titan M1-R, allowing for better identification of aerosol clusters and partial visualization of the eddy pattern. However, the single-unit configuration introduces certain limitations, such as gaps in the coverage, and the scanning pattern itself can create visual artifacts that resemble swirling motions even in calm conditions. This ambiguity makes it challenging to determine whether the detected pattern is solely due to the turbulence or influenced by the scanning method. Figure 7e showcases the aerosol detection results from the proposed cyclodial scanning LiDAR system. By overlapping the fields of view of the four units, this configuration achieves unmatched measurement density and resolution, capturing the eddy turbulence pattern with remarkable clarity and detail. The detected aerosols form a well-defined swirling pattern consistent with the characteristics of eddy turbulence, providing valuable insights into the distribution and intensity of the aerosols. The system’s advanced scanning mechanism and high-resolution capabilities ensure accurate detection and characterization of the turbulence, making it the most effective tool for monitoring such atmospheric conditions. The comparison of the four LiDAR systems highlights the clear superiority of the proposed cyclodial scanning LiDAR system in detecting and characterizing aerosols under eddy turbulence conditions. While the Velodyne VLS-128 provides basic detection capabilities, its low resolution and sparse data density fail to accurately represent the swirling eddy patterns. The Neuvition Titan M1-R improves on this with slightly better density and aerosol cluster detection but remains limited at extended ranges. The single cyclodial scanning LiDAR unit demonstrates higher density and better visualization of turbulence compared to the VLS-128 and Titan M1-R, yet its coverage gaps and occasional scanning pattern artifacts hinder its effectiveness for comprehensive turbulence analysis. In contrast, the proposed cyclodial scanning LiDAR system excels in all aspects, producing dense, detailed, and accurate data that precisely capture the swirling patterns characteristic of eddy turbulence. This unparalleled resolution and coverage make it an indispensable tool for identifying atmospheric disturbances and ensuring the safety and efficiency of UAM operations, establishing its clear advantage over traditional and single-unit LiDAR configurations. ### Aerosol detection results in gust conditions Fig. 8Detected aerosol distribution in gust conditions (**a**) aerosol distribution; (**b**) VLS-128; (**c**) M1-R; (**d**) single cyclodial scanning LiDAR system; and (**e**) proposed cycloidal scanning LiDAR system. In Fig. 8, the analysis focuses on aerosol detection results under gusty wind conditions^13–15,33^, a scenario where strong, localized airflows redistribute aerosols in distinctive patterns. Gust conditions often cause aerosols to concentrate densely in the central regions of the airflow, forming characteristic shapes while dispersing outward into the surrounding areas, as shown in Fig. 8a These conditions were tested across four LiDAR Velodyne VLS-128, Neuvition Titan M1-R, a single cyclodial scanning LiDAR unit, and the proposed cyclodial scanning LiDAR system, with a detection area spanning 2 km forward, 90 m to 810 m in height, and 540 m to each side. Figure 8b depicts the aerosol detection results from the Velodyne VLS-128. While the VLS-128 captures a general distribution of aerosols influenced by gust turbulence, its data points are sparse and uneven due to the system’s longer dwell time and lower measurement density. The turbulence pattern is weakly represented, with only vague indications of aerosol clustering and redistribution. The low resolution and insufficient density hinder the VLS-128’s ability to capture the detailed structure of gust patterns, making it challenging to differentiate gust turbulence from general atmospheric disturbances. Figure 8c shows the results from the Neuvition Titan M1-R, which provides a slightly improved representation compared to the VLS-128. Its higher measurement density allows for better identification of aerosol clusters and some faint indications of gust patterns. However, the gust shape remains weak, and the data lack the resolution necessary to clearly identify the airflow’s intensity or direction. This limitation makes it difficult to confirm the presence of gust turbulence with confidence. Figure 8d presents the detection results from a single cyclodial scanning LiDAR unit. The system achieves higher data density and better resolution compared to the VLS-128 and Titan M1-R, resulting in a more detailed visualization of aerosol redistribution. However, the pattern detected resembles eddy turbulence rather than gust conditions, with swirling shapes that could potentially lead to misinterpretations. Additionally, the detected pattern can sometimes appear even under calm conditions, suggesting that some of the observed structures may be artifacts of the scanning mechanism rather than actual atmospheric features. Figure 8e illustrates the results obtained from the proposed cyclodial scanning LiDAR system. This configuration provides the most detailed and accurate representation of gust turbulence, with aerosols densely clustered in the central regions of the gust and dispersed outward into the periphery. The high resolution and overlapping fields of view eliminate gaps in the detection area, creating a dense and comprehensive dataset. The distinct gust-shaped pattern is clearly visible, allowing for precise identification and characterization of the airflow. This capability highlights the system’s superior performance in detecting complex atmospheric disturbances, making it particularly well-suited for UAM applications. The comparison of the four LiDAR systems under gusty conditions reveals significant differences in their performance. The Velodyne VLS-128 captures a basic aerosol distribution but suffers from sparse data points and low resolution, making it incapable of reliably identifying gust patterns. The Neuvition Titan M1-R offers slightly better density and clustering capabilities but still lacks sufficient resolution to clearly define gust turbulence. The single cyclodial scanning LiDAR unit improves further with higher resolution and density, yet its results occasionally include artifacts resembling turbulence patterns, complicating interpretation and limiting accuracy. In contrast, the proposed cyclodial scanning LiDAR system excels in all aspects, producing dense, high-resolution data that clearly delineates the gust’s central airflow and peripheral dispersion. This system provides the most accurate and detailed representation of gust patterns, making it uniquely suited for detecting and analyzing complex atmospheric disturbances critical to UAM safety and operations. ### Particle image density and its Use in detecting aerosols in low-level wind hazards The concept of PID^118–121^ refers to the concentration and spatial distribution of particles, such as aerosols, within a specific volume of space. LiDAR technology enables the quantification of aerosol particles detected per unit area or volume by converting the backscattered light signals into PID metrics. This approach simplifies the complex point cloud data generated by LiDAR systems, which, while visually intuitive for humans, requires significant computational resources for machine processing. By transforming the data into PID, the information becomes highly simplified, reducing computational load and enabling real-time processing. This is particularly important for UAM applications, where rapid detection and response to atmospheric changes are critical for safe operations. The assessment of PID involves several steps. The LiDAR system emits laser pulses into the atmosphere, where they interact with aerosol particles, scattering the light. The LiDAR receiver captures the backscattered signal, and the intensity and distribution of this signal are analyzed to determine the concentration and density of aerosol particles within the detection volume. High-intensity backscatter corresponds to a higher particle image density, whereas low-intensity backscatter reflects a lower density. By mapping PID across the detection volume, LiDAR systems can identify distinct patterns associated with various wind hazards. For instance, swirling patterns with fluctuating densities signify eddy turbulence, concentrated bursts of high density moving in a single direction indicate gusts, and abrupt changes in density over short distances reveal wind shear. In practical applications, PID proves indispensable for UAM systems, offering a simplified and computationally efficient way to quantify aerosol distribution by counting particles within grid cells. This method enables real-time analysis while preserving essential details of aerosol concentration, making it particularly beneficial for UAM operations where timely and accurate detection of atmospheric conditions is critical for safe navigation. Unlike raw point cloud data, which visually represents aerosol positions but requires extensive computational resources for processing, PID significantly reduces data complexity, allowing UAM systems to make timely and precise decisions. This includes real-time adjustments to flight paths to avoid turbulent or hazardous areas, ensuring smooth and safe navigation. The simplified nature of PID also facilitates the creation of detailed yet computationally efficient maps of aerosol concentration and movement over time, providing critical insights into atmospheric conditions. These maps enable UAM systems to identify potential risks, plan optimal routes, and enhance overall operational safety. By leveraging PID analysis, UAM systems achieve an optimal balance between data processing efficiency and situational awareness, making it a key enabler of safe, effective, and dynamic aerial mobility. #### Particle image density results in calm atmospheric conditions Fig. 9Particle image density in calm atmospheric conditions (**a**) VLS-128; (**b**) M1-R; (**c**) single cyclodial scanning LiDAR system; and (**d**) proposed cycloidal scanning LiDAR system. In Fig. 9, the PID analysis under calm atmospheric conditions provides a clear comparison of the aerosol detection capabilities of four LiDAR Velodyne VLS-128, Neuvition Titan M1-R, single cyclodial scanning LiDAR, and the proposed four-unit cyclodial scanning LiDAR system. Figure 9a presents the PID results obtained using the Velodyne VLS-128, where particle counts remain low, ranging between 0 and 5 across most grid cells. The uniform distribution of counts reflects the stable atmospheric conditions without significant clustering, yet it also underscores the limitations of the VLS-128 in resolution and measurement density. While it captures the general aerosol distribution, the system struggles to provide the granularity necessary for detailed atmospheric analysis. Similarly, Fig. 9b, showing PID results from the Neuvition Titan M1-R, reveals a slightly improved distribution with particle counts exceeding 10 in some regions, benefiting from its solid-state scanning mechanism. However, the Titan M1-R’s detection density declines at greater distances, leading to a less comprehensive representation of aerosol concentrations. The PID results from the single cyclodial scanning LiDAR, illustrated in Fig. 9c, show a significant improvement in both density and resolution. Particle counts reach up to 70 in certain regions, indicating the system’s ability to detect more aerosols and provide greater detail than the VLS-128 and Titan M1-R. However, due to its non-repetitive scanning pattern, the single cyclodial LiDAR leaves some areas underrepresented, as its measurements are confined to the specific paths of its scanning mechanism. This limitation affects its ability to provide full spatial coverage, reducing its reliability in representing the entire detection volume. In contrast, the proposed four-unit cyclodial scanning LiDAR system, shown in Fig. 9d, achieves unparalleled performance. By overlapping the fields of view of its four units, the system produces significantly higher PID values, with counts exceeding 160 in densely populated areas, and ensures uniform coverage across the grid. This dense and detailed representation of aerosol concentrations provides a comprehensive understanding of atmospheric conditions, making it an ideal tool for real-time monitoring and hazard detection. The advantages of PID become even more apparent when compared to the point cloud data shown in Fig. 6. While point cloud visualizations provide a detailed spatial representation of aerosols, they require significant computational resources for processing and analysis, making them less suitable for real-time applications. PID, in contrast, condenses this information into simple numerical values within grid cells, dramatically reducing computational demand while preserving critical insights into aerosol distribution. For instance, the PID data from the proposed four-unit cyclodial scanning LiDAR system highlights not only the uniformity of aerosol dispersion in calm conditions but also its superior resolution and measurement density compared to the other systems. This allows for faster and more accurate analysis of atmospheric conditions, enabling UAM systems to make informed decisions in real time. The proposed four-unit cyclodial scanning LiDAR system stands out as the most effective tool for detecting and analyzing aerosol distributions under calm atmospheric conditions. Unlike the Velodyne VLS-128, which offers only a basic overview of aerosol dispersion, or the Titan M1-R, which provides modest improvements but lacks the resolution for detailed analysis, the four-unit system delivers unmatched performance. The single cyclodial LiDAR improves upon these systems with higher density and resolution, but its incomplete coverage limits its applicability for comprehensive analysis. By contrast, the four-unit cyclodial system achieves full spatial coverage with exceptional resolution, as reflected in its PID values, which exceed those of the other systems by a significant margin. This system not only captures a more detailed and accurate representation of aerosol distributions but also ensures reliability and consistency across the detection volume. The use of PID enables real-time analysis, making it possible to quickly identify changes in atmospheric conditions or potential hazards without the computational burden associated with point cloud data. This efficiency and precision are critical for UAM operations, where safe navigation depends on timely and accurate environmental monitoring. The proposed system’s capabilities mark a significant advancement in LiDAR technology, positioning it as an indispensable tool for modern aerial mobility systems. By combining dense, high-resolution measurements with real-time processing efficiency, the four-unit cyclodial scanning LiDAR system sets a new standard for atmospheric monitoring and situational awareness in UAM operations. #### Particle image density results in wind hazard conditions The PID analysis across four LiDAR systems-Velodyne VLS-128, Neuvition Titan M1-R, single cyclodial scanning LiDAR, and the proposed four-unit cyclodial scanning LiDAR system-offers a detailed examination of their ability to detect aerosol distributions under eddy and gust turbulence conditions as shown in Figs. 10 and 11. In eddy turbulence conditions, depicted in Fig. 10, aerosols form distinct swirling patterns, with dense clusters near the vortex core dispersing outward into lower-density regions. The PID results highlight stark contrasts in performance among the systems. Figure 10a, illustrating the results for the Velodyne VLS-128, shows sparse and uneven aerosol distributions, offering limited information on the eddy pattern due to the system’s restricted resolution and data density. The swirling patterns, though faintly visible, lack clarity, making it difficult to accurately pinpoint the turbulence characteristics. Figure 10b, from the Titan M1-R, demonstrates moderately improved density and the ability to detect some aerosol clusters near the vortex center. However, the resolution remains insufficient for capturing the detailed structure of the eddy. Figure 10c, representing the single cyclodial scanning LiDAR, achieves higher resolution and measurement density, partially revealing the swirling aerosol distribution. However, its non-repetitive scanning pattern occasionally introduces artifacts resembling eddy turbulence, even under calm conditions, complicating interpretation. Figure 10d, showcasing the proposed four-unit cyclodial scanning LiDAR system, stands out with its ability to generate a dense and precise PID dataset. The overlapping fields of view from its four units eliminate coverage gaps, producing a distinct and highly detailed representation of the eddy. The high PID values near the vortex center and their gradual outward dispersion offer a comprehensive depiction of the turbulence dynamics, enabling precise hazard identification and characterization. In gust turbulence conditions, shown in Fig. 10, aerosols redistribute into linear patterns along the gust’s airflow, with dense clusters near the core and gradual dispersion toward the periphery. The PID results again reveal substantial differences among the systems. Figure 10a, from the Velodyne VLS-128, captures only a basic aerosol distribution, with sparse data points that fail to accurately delineate the gust pattern. The lack of resolution limits its ability to differentiate gust turbulence from general atmospheric conditions. Figure 10b, from the Titan M1-R, provides slightly better measurement density, capturing some aerosol clustering near the gust core. However, it still lacks the resolution and density required for a detailed analysis of gust turbulence. Figure 10c, representing the single cyclodial scanning LiDAR, provides higher data density and better resolution, revealing parts of the gust pattern. However, its coverage gaps and occasional scanning artifacts hinder its ability to reliably represent the airflow dynamics. In stark contrast, Figure 10d, from the proposed cyclodial scanning LiDAR system, exhibits unmatched accuracy and resolution. The overlapping fields of view of its four units produce a dense and comprehensive PID dataset that clearly captures the gust’s central airflow and its surrounding dispersion, offering a detailed and precise representation of the turbulence structure. The comparison between PID and point cloud data, as seen in Figs. 7 and 8, underscores the practical advantages of using PID for real-time atmospheric analysis. Point cloud data, while visually comprehensive, demands significant computational resources for processing and interpretation, making it less suitable for real-time decision-making in UAM scenarios. PID, on the other hand, simplifies data into actionable insights, reducing computational complexity without sacrificing critical information. This simplification is particularly advantageous for UAM systems, where timely and accurate atmospheric analysis is essential for ensuring safety and operational efficiency. The PID results for the proposed four-unit cyclodial scanning LiDAR system consistently exhibit higher resolution and density compared to the other systems, enabling it to accurately capture and characterize complex atmospheric disturbances. The proposed four-unit cyclodial scanning LiDAR system emerges as the most effective tool for aerosol detection and turbulence characterization under both eddy and gust conditions. Its ability to provide unparalleled resolution and measurement density sets it apart from conventional LiDAR systems like the Velodyne VLS-128 and Titan M1-R. Unlike the VLS-128, which offers only basic aerosol detection with sparse and low-resolution data, and the Titan M1-R, which provides modest improvements but struggles at extended ranges, the cyclodial system achieves exceptional precision. The single cyclodial LiDAR unit enhances density and resolution compared to these systems, but its inherent limitations, such as coverage gaps and scanning artifacts, reduce its reliability for comprehensive turbulence analysis. In contrast, the proposed system leverages its overlapping fields of view and advanced beam-steering capabilities to produce dense, uniform PID datasets that capture the fine details of aerosol distribution and airflow dynamics. These datasets not only delineate the swirling patterns of eddy turbulence but also accurately represent the linear flows and clustering characteristic of gust turbulence. The high-resolution PID data from the proposed system provide several practical benefits. First, it enables real-time monitoring and precise identification of atmospheric disturbances, facilitating timely decision-making and hazard mitigation. Second, its comprehensive coverage eliminates gaps in data, ensuring a full and accurate representation of the detection area. Third, its computational efficiency allows for rapid analysis, making it an ideal solution for dynamic and time-critical UAM operations. The ability to generate actionable insights into atmospheric conditions enhances situational awareness, enabling UAM systems to optimize flight paths, avoid hazards, and maintain operational safety. This unparalleled performance underscores the proposed cyclodial scanning LiDAR system’s role as a transformative technology for UAM applications. By combining high-resolution aerosol detection with real-time processing efficiency, it addresses the limitations of traditional LiDAR systems, offering a robust and reliable solution for navigating complex atmospheric environments. Its advanced capabilities position it as an essential tool for ensuring the safety and efficiency of UAM operations, paving the way for its widespread adoption in modern aerial mobility systems. This innovation not only sets a new benchmark for LiDAR performance but also redefines the standards for atmospheric monitoring in the aviation industry.Fig. 10Particle image density in eddy turbulence conditions (**a**) VLS-128; (**b**) M1-R; (**c**) single cyclodial scanning LiDAR system; and (**d**) proposed cycloidal scanning LiDAR system.Fig. 11Particle image density in gust turbulence conditions (**a**) VLS-128; (**b**) M1-R; (**c**) single cyclodial scanning LiDAR system; and (**d**) proposed cycloidal scanning LiDAR system.Fig. 12Detected aerosol distribution in eddy turbulence conditions visualized as a point cloud for (**a**) 1st frame, (**b**) 4th frame, (**c**) 7th frame, and (**d**) 10th frame. The size of the points represents the size of the detected particles, while the color indicates the distance, with red corresponding to closer particles and blue to farther particles. Changes in wind movement and shape over time are evident. ### Estimating wind dynamics via particle image density #### Normalized particle detection ratio The calculation of the normalized particle detection ratio (NPDR) is a fundamental preprocessing step to ensure the consistency, accuracy, and reliability of PID analysis. Due to the unique measurement patterns of the LiDAR systems and the overlapping detection regions created by the four-unit configuration, the maximum measurable particle count varies significantly across different grid cells. The matrix provided in the Eq. (5) represents the maximum measurable particle count for each grid cell within the LiDAR system’s sensing area. This matrix is derived from the LiDAR’s scanning pattern and the overlapping fields of view provided by the multiple LiDAR units employed in the system. Each entry in the matrix specifies the maximum number of particles that can be detected in a corresponding grid cell under optimal conditions. This value accounts for spatial variations in the LiDAR’s detection sensitivity caused by both the scanning geometry and the overlapping of multiple LiDAR beams. The central regions of the matrix typically exhibit higher values, reflecting areas with significant beam overlap and enhanced detection capabilities. Conversely, lower values near the periphery indicate regions where fewer beams overlap, resulting in reduced sensitivity. This inherent variability introduces potential biases into the raw PID values, which must be corrected to enable meaningful spatial and temporal comparisons of wind dynamics. By normalizing the particle counts in each grid cell, we eliminate systematic errors introduced by the differences in detection patterns, allowing the PID to accurately represent the aerosol distribution within the monitored area.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \begin{bmatrix} 38 & 283 & 440 & 497 & 535 & 604 & 351 & 282 & 38\\ 232 & 336 & 388 & 489 & 716 & 504 & 371 & 329 & 209\\ 256 & 265 & 221 & 249 & 475 & 392 & 391 & 468 & 416\\ 256 & 265 & 221 & 249 & 351 & 196 & 195 & 228 & 218\\ 232 & 336 & 388 & 489 & 716 & 504 & 371 & 329 & 209\\ 38 & 283 & 440 & 497 & 535 & 604 & 351 & 282 & 38 \end{bmatrix} \end{aligned} $$\end{document}To achieve this, the NPDR for each grid cell is computed by dividing the observed particle count by the maximum measurable particle count for that specific cell. The maximum measurable particle count is a function of both the LiDAR measurement pattern and the degree of overlap among the four LiDAR units. Specifically, the overlapping detection regions created by multiple LiDAR beams result in varying sensitivity across the grid, necessitating a correction to account for this heterogeneity. By applying the NPDR normalization, we ensure that the resulting PID values are spatially unbiased and represent the true relative particle distribution. For a given grid cell (*i*, *j*), the NPDR is defined mathematically 6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \text {NPDR}_{i,j} = \frac{\text {Observed particle count}_{i,j}}{\text {Maximum measurable particle count}_{i,j}}, \end{aligned} $$\end{document}where: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \text {Observed particle count}_{i,j} $$\end{document} is the number of aerosol particles detected by the LiDAR system within the (*i*, *j*)-th grid cell. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \text {Maximum measurable particle count}_{i,j} $$\end{document} represents the upper bound of particles that the LiDAR system can detect within the (*i*, *j*)-th grid cell. This is determined by the LiDAR measurement pattern and the cumulative effects of overlapping beams from the four-unit configuration. The NPDR values are dimensionless and range from 0 to 1, providing a normalized metric for comparing aerosol concentrations across grid cells, irrespective of their varying detection capacities. This normalized representation is critical for downstream analyses, as it eliminates distortions caused by the inherent asymmetry in LiDAR detection patterns. As a result, NPDR enables consistent and robust characterization of aerosol distributions, even in complex environments with dynamic atmospheric conditions. Once the normalized PID matrix, denoted as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \{\text {NPDR}_{i,j}\} $$\end{document}, is computed, it serves as the foundation for advanced analytical techniques to estimate various wind-related dynamics. These dynamics include, but are not limited *Wind shape visualization*: Highlighting and delineating regions of significant wind activity by identifying contiguous clusters of above-threshold NPDR values.*Localized wind intensity estimation*: Quantifying the strength of wind activity in specific regions by comparing the NPDR values within the wind shape against the global average.*Weighted centroid estimation*: Determining the geometric center of wind activity using NPDR values as weights to emphasize regions of higher aerosol concentration.*Area change analysis*: Measuring the expansion or contraction of wind-affected regions over consecutive time frames to evaluate dynamic changes in atmospheric conditions.*Directional analysis*: Analyzing the preferential direction of wind movement based on asymmetries in the spatial distribution of NPDR values.*Intensity uniformity assessment*: Evaluating the standard deviation of NPDR values within the wind shape to assess whether the wind strength is uniformly distributed or exhibits sharp gradients.*Temporal stability analysis*: Monitoring the consistency of wind patterns across time by comparing successive NPDR matrices and analyzing changes in their spatial configuration.The NPDR not only provides a consistent and normalized representation of aerosol concentrations but also enables the real-time estimation of wind dynamics with minimal computational overhead. This is particularly advantageous for UAM applications, where rapid decision-making and real-time data processing are critical for operational safety. By leveraging the NPDR, computational complexity is significantly reduced compared to traditional approaches, allowing for faster and more efficient analysis. This makes the proposed method highly suitable for real-time monitoring and control in dynamic urban environments. Furthermore, the adoption of NPDR as a standardized metric enhances the robustness of wind dynamic estimations by ensuring that variations in LiDAR detection capabilities do not affect the integrity of the results. This approach not only improves the interpretability of PID data but also enables a wide range of applications, including but not limited to dynamic route optimization, hazard identification, and environmental monitoring. As such, the NPDR serves as a pivotal component in the analytical framework for wind dynamics estimation, enabling accurate, efficient, and actionable insights into complex atmospheric phenomena. To enhance the clarity and understanding of the NPDR methodology proposed in this study, we incorporated a series of figures derived from experimental data. Specifically, the analysis targets two types of wind phenomena-eddy and gust. The proposed cycloidal scanning LiDAR system was used to acquire data over 10 consecutive frames within a 0.5 s interval, capturing the dynamic behavior of aerosols in these turbulence conditions. Out of the 10 frames, four representative frames (1st, 4th, 7th, and 10th) were selected for detailed visualization and analysis. The visualization pipeline includes three raw point cloud representation of particle density, PID grids, and NPDR calculations normalized by the maximum measurable particle counts of each grid cell. This comprehensive approach ensures the precise characterization of wind structures. Figures 12 and 13 show the raw point cloud distribution of aerosols detected under eddy and gust turbulence conditions, highlighting the spatial dispersion and density of particles. These raw distributions were subsequently processed into PID grids as presented in Figs. 14 and 15, which quantify the particle counts detected in each grid cell, providing a structured and interpretable view of the wind dynamics. Figs. 16 and 17 represent the normalized particle detection ratios, calculated by dividing the observed particle count in each cell by the maximum measurable count for that cell. This normalization step corrects for biases introduced by the LiDAR’s scanning pattern and overlapping measurement regions, ensuring consistent and reliable data interpretation. The NPDR visualizations, as demonstrated in the final set of figures, provide a significantly enhanced perspective compared to the raw point cloud and PID representations. In particular, the NPDR results emphasize regions of high wind activity, making wind structures, such as eddies and gusts, visually more prominent and distinct. The normalization process effectively minimizes noise and measurement inconsistencies, which can obscure key features in the raw and PID representations. As a result, the NPDR visualization highlights the core dynamics of wind patterns with greater clarity, enabling easier identification of regions of interest and more accurate analysis of wind behavior. These regions of high NPDR values correspond to areas of intense particle activity and provide critical insights into the spatial and temporal variations in wind phenomena. By leveraging this enhanced visualization capability, the study demonstrates the superiority of NPDR as a tool for detecting and characterizing wind dynamics in a precise and computationally efficient manner.Fig. 13Detected aerosol distribution in gust turbulence conditions visualized as a point cloud for (**a**) 1st frame, (**b**) 4th frame, (**c**) 7th frame, and (**d**) 10th frame. The size of the points represents the size of the detected particles, while the color indicates the distance, with red corresponding to closer particles and blue to farther particles. Over time, as observed in frames (**a**) to (**d**), slight shifts in the location and shape of the wind flow are noticeable, capturing dynamic changes in the gust turbulence.Fig. 14Particle image density in eddy turbulence conditions visualized for (**a**) 1st frame, (**b**) 4th frame, (**c**) 7th frame, and (**d**) 10th frame. The numeric values inside the cells represent the number of detected particles, where lower values are depicted in darker blue, and higher values transition towards intense red shades.Fig. 15Particle image density in gust turbulence conditions visualized for (**a**) 1st frame, (**b**) 4th frame, (**c**) 7th frame, and (**d**) 10th frame. The numeric values inside the cells represent the number of detected particles, where lower values are depicted in darker blue, and higher values transition towards intense red shades.Fig. 16Normalized particle detection ratios in eddy turbulence conditions visualized for (**a**) 1st frame, (**b**) 4th frame, (**c**) 7th frame, and (**d**) 10th frame. The numeric values inside the cells represent the normalized particle detection ratios, where lower values are depicted in darker blue, and higher values transition towards intence red shades.Fig. 17Normalized particle detection ratios in gust turbulence conditions visualized for (**a**) 1st frame, (**b**) 4th frame, (**c**) 7th frame, and (**d**) 10th frame. The numeric values inside the cells represent the normalized particle detection ratios, where lower values are depicted in darker blue, and higher values transition towards intence red shades.Fig. 18Wind region shapes in eddy turbulence conditions represented using convex hulls based on normalized particle detection ratios for frames (**a**) 1st, (**b**) 4th, (**c**) 7th, and (**d**) 10th. The red lines represents the boundaries of the wind regions, highlighting dynamic changes in wind structure and intensity over time.Fig. 19Wind region shapes in gust turbulence conditions represented using convex hulls based on normalized particle detection ratios for frames (**a**) 1st, (**b**) 4th, (**c**) 7th, and (**d**) 10th. The red lines represents the boundaries of the wind regions, highlighting dynamic changes in wind structure and intensity over time.Fig. 20Weighted centroids of wind regions in eddy turbulence conditions, visualized using normalized particle detection ratios and convex hulls for frames (**a**) 1st, (**b**) 4th, (**c**) 7th, and (**d**) 10th. The red lines represent the wind region shapes, while the white circles indicate the weighted centroids, demonstrating the dynamic shifts in central wind activity over time.Fig. 21Weighted centroids of wind regions in gust turbulence conditions, visualized using normalized particle detection ratios and convex hulls for frames (**a**) 1st, (**b**) 4th, (**c**) 7th, and (**d**) 10th. The red lines represent the wind region shapes, while the white circles indicate the weighted centroids, demonstrating the dynamic shifts in central wind activity over time. #### Wind shape visualization The convex hull algorithm^122–124^ is utilized to represent the spatial structure of wind formations by encapsulating ROIs within a detection grid. These ROIs correspond to grid cells with significant wind activity, identified by NPDRs exceeding a predefined threshold. In this context, the grid cells already contain NPDRs, which were computed in a prior step. These ratios represent the proportion of detected particles relative to the maximum detection capacity, and they serve as the basis for identifying critical regions. To determine the region of interest, the threshold value is calculated as the mean of all normalized values stored in the grid cells. Denoted by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \overline{x} $$\end{document}, the threshold value is calculated as the mean of all normalized values stored in the grid cells. Denoted by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \overline{x} $$\end{document}, the threshold value is expressed as 7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \overline{x} = \frac{1}{N} \sum _{i=1}^{6}\sum _{j=1}^{9} x_{i,j}, \end{aligned} $$\end{document}where *N* is the total number of cells in the grid. All grid cells satisfying \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ x_{i,j} > \overline{x} $$\end{document} are selected as part of the RIOs, representing areas with intensitifed win activity. These selected cells are then mapped onto a Cartesian coordinate system, where the colum index *j* corresponds to the *x*-coordinate and the row index *i* corresponds to the *y*-coordinate. To construct the convex hull, the monotone chain algorithm is employed, offering computational efficiency and suitability for real-time applications. The process involves the following *Sorting cells* The selected grid cells are sorted lexicographically based on their coordinates, with the *x*-coordinate as the primary key and the *y*-coordinate as the secondary key.*Lower hull construction* Starting from the leftmost cell, the lower hull is iteratively constructed. For these consecutive cells, *A*, *B*, and *C*, convexity is maintained using the cross-product criterion \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ (x_B - x_A)(y_C - y_B) - (y_B - y_A)(x_C - x_B) $$\end{document}. If the cross-product is positive, the cell *B* is part of the lower hull; otherwise, it is removed.*Upper hull construction* Similarly, the upper hull is constructed by traversing the sorted cells in reverse order, maintaining convexity through the same criterion.*Convex hull formation* The lower and upper hulls are combined to form a minimal convex polygon that encloses all selected cells, representing the wind shape within the detection grid.This convex polygon represents the wind’s external boundary, providing a simplified yet robust visualization of its spatial structure. The convex hull offers a concise representation of the wind shape by enclosing regions with elevated NPDRs, highlighting areas of significant atmospheric activity. By leveraging the convex hull method, wind shape visualization is achieved efficiently and with geometric clarity. The boundary of the wind-affected area was determined using the monotone chain algorithm, which efficiently calculates the convex hull of the data points representing the wind dynamics. For clarity and domain-specific relevance, the convex hull constructed from the normalized particle detection ratios is hereafter referred to as the wind region. This terminology emphasizes the physical and aerodynamic significance of the enclosed area as the effective spatial extent of the wind’s influence. The wind region provides a clear and quantifiable boundary for further analysis of wind dynamics, including intensity, uniformity, and directional characteristics. Following the application of the monotone chain algorithm to generate convex hulls, the resulting wind shapes provide an intuitive and accurate depiction of hazardous wind regions. As shown in Figs. 18 and 19, these convex hulls encapsulate regions of high NPDRs, effectively highlighting areas of strong wind activity. This approach not only simplifies the visualization of complex wind patterns but also emphasizes the critical zones that pose significant risks to UAM operations. By focusing on the outermost boundaries of the NPDR data, the convex hulls delineate the spatial extent of turbulent wind regions with high computational efficiency. The convex hull representation underscores key operational insights, particularly for UAM vehicles. In eddy turbulence conditions, illustrated in Fig. 18, the convex hulls exhibit a more confined and oscillatory shape, reflecting localized turbulence with more predictable patterns. This expansion reveals critical areas of severe wind that must be avoided to maintain flight stability and safety. Conversely, in gust turbulence conditions, for example, the convex hulls expand dynamically across sequential frames, as seen in Fig. 19, reflecting the rapidly changing nature of wind intensity and distribution. These contrasting behaviors between gust and eddy turbulence highlight the adaptability of the convex hull method to varying wind dynamics. By utilizing NPDR as the basis for convex hull construction, the proposed visualization method provides enhanced clarity compared to raw aerosol point clouds or PID visualizations. The convex hulls distinctly outline regions of concentrated wind activity, effectively translating numerical data into actionable insights. This enables UAM operators to preemptively adjust flight paths or altitudes to mitigate risks. The visual prominence of hazardous regions, as emphasized by the convex hulls, demonstrates the robustness and utility of this approach in ensuring the safety and reliability of UAM systems in dynamic atmospheric conditions. #### Localized wind intensity estimation Localized wind intensity estimation is an essential component of wind dynamics analysis, providing detailed insights into both regional and localized atmospheric conditions. This estimation process uses the convex hull representation of wind regions, which is constructed from NPDRs stored in a grid matrix. By focusing on the wind region defined by the convex hull, the overall wind strength and its localized variations can be systematically analyzed and quantified. The process begins with a global baseline, which was previously calculated as the mean NPDR across all grid cells, denoted as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \overline{x} $$\end{document}. This value serves as the threshold for defining the ROIs and provides a reference for comparing the intensity of the wind region to the overall grid data. For localized analysis, the convex hull encloses the grid cells with values exceeding this global threshold. Within the convex hull, the wind region’s mean NPDR, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \bar{x}_{wind} $$\end{document}, is calculated to assess the geneeral intensity of the wind region. This mean is calculated as 8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \bar{x}_{wind} = \frac{1}{M} \sum _{(i,j)\in Convex~Hull} x_{i, j}, \end{aligned} $$\end{document}where *M* is the number of cells inside the convex hull, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ x_{i,j} $$\end{document} represents the NPDR of each enclosed cell. The relative intensity of the wind region is then expressed as the wind intensity 9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} R = \frac{\bar{x}_{wind}}{\overline{x}}, \end{aligned} $$\end{document}A ratio \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ R > 1 $$\end{document} indicates that the wind region is more intense than the global average, with higher values of *R* signifying stronger wind activity within the region. To further analyze localized wind variations, the maximum and minimum NPDRs within the convex hull are identified. These values, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ x_{max} $$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ x_{min} $$\end{document}, provide additional information about the intensity and variability of the wind 10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} x_{max} = \max _{(i,j)\in Convex~Hull} x_{i,j},~~~x_{min} = \min _{(i,j)\in Convex~Hull} x_{i,j}. \end{aligned} $$\end{document}The intensity range within the wind region is then calculated as the difference between the maximum and minimum 11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \Delta x = x_{max} - x_{min}. \end{aligned} $$\end{document}where a large \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \Delta x $$\end{document} indicates significant variability in wind intensity across the region, such as localized gusts or high-intensity regions. By combining the wind intensity ratio *R* with the intensity range \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \Delta x $$\end{document}, a comprehensive assessment of localized wind dynamics is achieved, enabling detailed analysis of wind strength and variability within specific regions of interest. The localized wind intensity characteristics, calculated using the proposed cycloidal LiDAR system, provide a detailed analysis of turbulence in both eddy and gust conditions across 10 sequential frames. The analysis, summarized in Table 3, includes the global mean intensity across the entire measurement area, the mean intensity within the convex hull, the maximum and minimum intensities inside the convex hull, the ratio of the global mean to the wind mean, and the intensity range defined as the difference between maximum and minimum values. The wind regions were identified using the monotone chain algorithm-based convex hull method, which accurately delineates the boundaries of high-intensity zones.Table 3Localized wind intensity metrics for eddy and gust turbulence conditions.WindtypePropertyFrame1Frame2Frame3Frame4Frame5Frame6Frame7Frame8Frame9Frame10EddyGlobalmean intensity30.832.432.432.134.133.234.433.833.133.2Wind regionmean intensity41.850.641.455.053.755.151.749.653.755.8Maximum windregion intensity44.564.563.670.969.161.376.362.168.871.2Minimum windregion intensity38.932.531.544.244.744.536.836.639.538.4Wind regionmean to globalmean ratio1.41.61.31.61.61.71.51.51.61.7Wind regionintensity range5.632.032.126.724.416.839.525.529.332.8GustGlobal meanintensity33.934.134.133.934.034.134.133.734.535.3Wind regionmean intensity51.953.751.260.258.159.360.356.157.255.1Maximum windregion intensity70.076.381.678.968.871.168.467.569.873.8Minumum windregion intensity36.736.035.042.043.837.137.142.040.537.8Wind regionmean to globalmean ratio1.51.51.51.81.71.71.81.71.71.6Wind regionintensity range33.340.346.636.925.034.031.325.529.336.0 For eddy turbulence, the convex hull regions demonstrate significant variability in wind intensity, reflecting localized and transient turbulence patterns. As shown in Table 3, the global mean intensity for eddy conditions remains relatively stable, ranging between 30.8 and 34.4 across all frames. However, the mean wind intensity within the convex hull shows considerable variation, peaking at 55.0 in Frame 4 and dipping to 41.4 in Frame 3. The maximum wind intensity reaches a high of 70.9 in Frame 7, whereas the minimum intensity shows fluctuations between 32.5 and 44.2 across the frames. The ratio of global mean to wind mean remains close to 1.6 across all frames, indicating that the localized regions of high wind intensity are proportionally elevated compared to the global field. The intensity range varies significantly, peaking at 39.5 in Frame 6, reflecting the dynamic nature of eddy turbulence and its spatial non-uniformity. For gust turbulence, the convex hulls enclose regions with consistently stronger wind intensities compared to eddy conditions, highlighting the aggressive nature of gusts. As presented in Table 3, the global mean intensity for gust conditions is slightly higher than that of eddy, ranging between 33.9 and 35.3 across all frames. The mean wind intensity within the convex hull is markedly higher, reaching 60.2 in Frame 4 and staying above 51.2 across all frames. The maximum wind intensity demonstrates even more pronounced values, peaking at 81.6 in Frame 3 and maintaining consistently high readings throughout. Similarly, the minimum wind intensity remains higher than the eddy case, with values ranging between 35.0 and 43.8. The ratio of global mean to wind mean remains relatively stable at approximately 1.7, indicating a consistent amplification of intensity within gust regions. The intensity range for gust turbulence exhibits substantial variation, peaking at 46.6 in Frame 3 and reflecting the sharp gradients characteristic of gusts. The comprehensive analysis of localized wind intensities underscores the effectiveness of the convex hull-based monotone chain algorithm in identifying and characterizing critical wind regions. The variability in intensity metrics across frames reflects the dynamic nature of atmospheric turbulence, with gusts exhibiting higher intensity and greater spatial variability compared to eddies. These findings offer valuable insights into turbulence behavior, providing a foundation for risk assessment and real-time decision-making in urban air mobility operations. #### Advanced analysis of wind dynamics using weighted centroid and temporal characteristics Localized wind centroid estimation plays a pivotal role in understanding and predicting wind dynamics by providing a precise assessment of the central point of activity within a defined wind region and offering insights into its directional movement. By leveraging NPDRs as weights, the weighted centroid calculation ensures computational efficiency and accuracy in localizing the wind region’s center. The convex hull, encapsulating ROIs, focuses the analysis on relevant cells, enhancing precision while excluding extraneous data. Additionally, tracking the centroid’s position over successive time frames enables the determination of the directional vector of the wind region’s movement, which is critical for analyzing dynamic atmospheric conditions. This comprehensive analytical approach, utilizing convex hull-based metrics derived from NPDRs, facilitates the evaluation of wind behavior, including spatial distribution, intensity uniformity, and temporal stability. Its computational efficiency makes it an effective tool for real-time atmospheric monitoring and operational decision-making in safety-critical domains. The weighted centroid is calculated by treating each grid cell within the convex hull as a point in a 2D coordinate system, with its NPDR serving as the weight for that point. The coordinates of the centroid, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ (C_{x}, C_{y}) $$\end{document}, are then determined using the following weighted average 12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} C_{x} = \frac{\sum _{(i,j)\in Convex~Hull} x_{i,j} \cdot j}{\sum _{(i,j)\in Convex~Hull} x_{i,j}},~~~C_{y} = \frac{\sum _{(i,j)\in Convex~Hull} x_{i,j} \cdot i}{\sum _{(i,j)\in Convex~Hull} x_{i,j}}, \end{aligned} $$\end{document}Here, *i* and *j* represent the row and column indices of each grid cell, respectively, which are mapped to the Cartesian coordinates of the grid. The numerator in each equation represents the weighted sum of the coordinates, while the denominator ensures normalization by the total weight (i.e., the sum of all NPDRs within the convex hull). This approach ensures that grid cells with higher NPDRs have a greater influence on the calculated centroid, effectively biasing the centroid toward regions of higher wind activity. Such weighting is critical for accurately reflecting the distribution of wind intensity within the ROI, particularly in scenarios where wind activity is unevenly distributed. The computed centroid provides a single, concise point that represents the geometric and intensity-weighted center of the wind region. This point is particularly useful for tracking the movement of wind regions over time. By computing centroids for successive time frames, the direction and velocity of wind movement can be estimated. The velocity of the centroid, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ v_{centroid} $$\end{document}, can be expressed 13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} v_{centroid} = \sqrt{\left( C_{x}^{t+1} - C_{x}^{t}\right) ^{2} + \left( C_{y}^{t+1} - C_{y}^{t}\right) ^{2}}, \end{aligned} $$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ C_{x}^{t} $$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ C_{y}^{t} $$\end{document} represent the centroid coordinates at time *t*, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ C_{x}^{t+1} $$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ C_{y}^{t+1} $$\end{document} represent the centroid coordinates at the subsequent time frame. The directional vector, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \vec {d} $$\end{document}, describing the movement of the centroid is given 14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \vec {d} = \left( C_{x}^{t+1} - C_{x}^{t}, C_{y}^{t+1} - C_{y}^{t}\right) . \end{aligned} $$\end{document}This vector indicates the change in position of the wind region’s centroid over time. The angle of the directional vector, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \theta $$\end{document}, with respective to the horizontal axis (e.g., eastward direction) can be calculated 15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \theta = \arctan \left( \frac{C_{y}^{t+1} - C_{y}^{t}}{C_{x}^{t+1} - C_{x}^{t}}\right) , \end{aligned} $$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \theta $$\end{document} is expressed in degrees. This angle provodes the precise orientation of thw wind’s movement. This simple calculation provides actionable insight into the dynamics of wind regions, allowing for real-time monitoring and analysis of wind movement patterns. The next step involves calculating the area of the wind region enclosed by the convex hull. The convex hull, constructed from grid cells with NPDRs exceeding a predefined threshold, defines the wind region of interest. The area, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ A^{t} $$\end{document}, is proportional to the number of grid cells enclosed by the convex hull at time *t*. This can be calculated geometrically from the convex hull vertices by directly counting the enclosed grid cells. By comparing the area across successtive time frames, the rate of change in area, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \Delta A / \Delta t $$\end{document}, is 16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \frac{\Delta A}{\Delta t} = \frac{A^{t+1} - A^{t}}{\Delta t}, \end{aligned} $$\end{document}A positive rate indicates the wind region is expanding or diffusion, while a negative rate implies contraction or concentration. This measure is closely related to wind velocity gradients and turbulence, providing early warnings for hazardous atmospheric conditions such as rapidly intensifying gusts or dissipating eddies. The eigenvector corresponding to the largest eigenvalue of *C* determines the principal axis of the convex hull, indicating the predominant direction of wind activity. This directional analysis is particularly useful for distinguishing between gusts, which often align along a single axis, and shears, which exhibit multi-directional or asymmetric characteristics. The uniformity of wind intensity within the convex hull is assessed using the standard deviation of the NPDRs, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \sigma _{x} $$\end{document}, with in the convex 17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} \sigma _{x} = \sqrt{\frac{1}{M} \sum _{(i,j)\in Convex~Hull} \left( x_{i,j} - \bar{x}_{wind}\right) ^2}, \end{aligned} $$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \bar{x}_{wind} $$\end{document} is the mean value of the enclosed grid cells. A low standard deviation indicates uniform wind intensity, while a high standard deviation suggests significant variability, often associated with localized gusts or microbursts. These variations are critical for understanding the potential impact and risk posed by uneven wind activity. defined as the total duration for which the convex hull remains consistent in shape and orientation within a predefined Finally, the temporal stability of the wind region is analyzed by tracking the persistence of the convex hull’s pattern over consecutive time frames. A stability metric, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ T_{s} $$\end{document}, is defined as the total duration for which the convex hull remains consistent in shape and orientation within a predefined 18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \begin{aligned} T_{s} = \sum _{t=1}^{T} \delta \left( h^{t}, h^{t-1}\right) , \end{aligned} $$\end{document}where T is the total number of time frames, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ h^{t} $$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ h^{t-1} $$\end{document} represent the convex hull at time *t* and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ t-1 $$\end{document}, respectively, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \delta $$\end{document} is a similarity metric that compares the convex hulls at adjacent time frames. This metric provides insights into the persistence and predictability of wind patterns, facilitating the assessment of long-term trends and the identification of potential hazards. The value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \delta $$\end{document} is set to 1 if the convex hull at time *t* and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ t-1 $$\end{document} are sufficently similar, and 0 otherwise. Similarity is determined based on criteria such as overlapping area, hull orientation, or other shape descriptors. Stable wind regions, characterized by long durations of consistent hulls, are typically more predictable and pose lower operational risks. Conversely, intermittent or chaotic hull patterns suggest rapidly evolving wind conditions, which are more challenging to forecast and may necessitate immediate mitigation strategies. Each of these metrics-area change, intensity uniformity, and temporal stability-offers a distinct perspective on wind dynamics. The integration of these metrics into a cohesive analytical framework enables detailed characterization of wind behavior in real-time. This framework leverages the convex hull to focus on regions of interest, ensuring computational efficiency and scalability. By combining mathematical rigor with practical applicability, this methodology enhances the detection, analysis, and prediction of wind patterns, supporting safe and efficient operations in dynamic atmospheric environments. The proposed approach is particularly relevant for applications such as urban air mobility, where rapid and accurate assessments of wind dynamics are critical for ensuring operational safety and success. The proposed cycloidal LiDAR system was utilized to continuously measure wind dynamics across 10 frames within a 0.5-s interval. The results, as shown in Table 4, demonstrate key characteristics of the wind region, specifically targeting eddy and gust turbulence. The computed properties include weighted centroid, velocity, area of the wind region, moving direction, uniformity of the wind region, and temporal stability. These features provide comprehensive insights into wind behavior, enabling a detailed analysis of its spatial and temporal variations. For eddy turbulence shown in Fig. 20, the weighted centroid of the wind region exhibited notable fluctuations across the frames, particularly between frames 4 and 7, indicating dynamic changes in wind concentration. The velocity values, peaking at 104.0 ms^−1^ in frame 5, demonstrate significant variability, reflecting the transient nature of eddy turbulence. The area of the wind region remained relatively consistent at approximately 0.1152 km^2^ but experienced a slight increase in frame 7 to 0.1296 km^2^, suggesting localized expansion. The moving direction displayed rapid shifts, with angles varying from −41.3° to 170.2°, highlighting the instability and erratic nature of eddy turbulence. Uniformity of the wind region varied across frames, with lower values indicating less consistent wind intensity. Temporal stability values between 0.25 and 1.0 further emphasize the dynamic and less predictable behavior of the eddy turbulence. For gust turbulence shown in Fig. 21, the weighted centroid exhibited more stability compared to eddy turbulence, maintaining a relatively consistent trajectory with minor oscillations. The velocity, which peaked at 111.4 ms^−1^ in frame 3, showed a more gradual decline over subsequent frames, reflecting the transient yet less erratic nature of gust turbulence compared to eddy turbulence. The area of the wind region for gust turbulence remained nearly constant at approximately 0.0576 km^2^,, indicating a confined spatial extent. Moving directions showed more gradual changes, with angles ranging from 29.0° to − 163.1°, demonstrating more directional consistency. Uniformity of the wind region exhibited values that gradually increased, suggesting a more uniform wind distribution over time. Temporal stability for gust turbulence, with values mostly at 1.0, indicates a higher level of predictability and steadiness compared to eddy turbulence. These analyses underscore the utility of the proposed LiDAR system in capturing intricate details of wind dynamics, including spatial distribution, intensity variation, and temporal changes. The ability to calculate and analyze weighted centroid, velocity, and other parameters offers a comprehensive understanding of the wind region’s behavior, enabling improved decision-making for applications such as UAM operations. By leveraging this approach, operators can gain actionable insights into wind conditions and mitigate potential risks effectively. The results presented in Table 4 highlight the system’s capability to characterize wind turbulence with high precision and reliability.Table 4Comprehensive wind dynamics analysis for eddy and gust turbulence conditions.WindtypePropertyFrame1Frame2Frame3Frame4Frame5Frame6Frame7Frame8Frame9Frame10EddyWeightedcentroid(5.0,3.5)(4.5,3.8)(4.9,3.3)(5.1,3.1)(4.3,3.3)(4.6,3.7)(4.1,4.8)(3.9,3.8)(3.9,3.7)(3.8,3.4)Velocity(ms^−1^)–74.571.643.3104.051.662.219.913.444.2Area ofwind region(km^2^)0.11520.11520.10080.10080.11520.11520.12960.10080.08640.1008Movingdirection–146.3°− 52.8°− 41.3°164.8°51.9°170.2°152.2°− 109.2°− 94.7°uniformity ofwind region2.112.110.711.39.05.512.49.811.912.3Temporalstability–10.250.570.750.750.770.550.710.71GustWeightedcentroid(4.6,3.5)(5.1,3.8)(6.0,4.1)(5.9,3.5)(5.7,3.2)(6.0,3.3)(5.9,3.2)(5.7,3.1)(5.5,3.5)(5.9,3.7)Velocity(ms^−1^)–63.3111.468.742.632.911.732.757.346.7Area ofwind region(km^2^)0.0720.0720.05760.05760.05760.05760.05760.05760.0720.0864Movingdirection–29.0°17.0°− 100.6°− 130.9°7.9°− 92.0°− 163.1°111.9°28.1°uniformity ofwind region12.217.220.716.910.415.315.511.311.713.9Temporalstability–10.60.7511110.60.83 ## Conclusion The proposed on-board cycloidal scanning LiDAR system represents a groundbreaking advancement in enhancing the visual perception of wind hazards for UAM. This system’s unique design and technological innovations not only detect but also respond to wind-related dangers, significantly impacting the safety and efficiency of UAM operations. This feature provides a sense of reassurance to UAM operators, aviation safety regulators, and technology developers. Through the utilization of PID, the system has demonstrated its ability to estimate various wind dynamics efficiently, including wind shape visualization, localized wind intensity estimation, weighted centroid estimation, area change, intensity uniformity, and temporal stability. The PID-based approach ensures exceptionally low computational requirements, allowing real-time processing and enabling UAM operators to receive actionable insights instantly. This real-time capability ensures a constant stream of updates on atmospheric conditions along the UAM flight path, equipping operators with the necessary information to make prompt and well-informed decisions. The system’s high spatial resolution (0.1°) and rapid scanning rate (50 Hz) guarantee the detection and accurate mapping of even the smallest wind hazards. This detailed visual perception of wind conditions empowers UAM vehicles to take proactive measures, such as adjusting flight paths or altitudes, to steer clear of adverse weather conditions. By employing multiple LiDAR units, the system enhances measurement resolution and range, providing a dense and detailed representation of the environment. The advanced beam steering capabilities of the Risley prism mechanism allow for variable FoV, optimizing the system’s ability to monitor specific regions of interest. This feature minimizes blind spots and ensures full situational awareness, crucial for maintaining operational safety in dynamic urban environments. While the system’s cost is higher than traditional automotive LiDAR systems, it is well-aligned with the price range of advanced aviation radar systems and represents a viable investment for UAM vehicles, which typically cost over $2,000,000 USD. Given its critical role in enhancing UAM safety and reliability, the cycloidal scanning LiDAR system is not only justified but also essential for safe and efficient urban air mobility operations. The emphasis on these capabilities instills confidence in UAM operators, aviation safety regulators, and technology developers. The cycloidal scanning LiDAR system is purpose-built to deliver superior visual perception capabilities that are pivotal in recognizing and mitigating wind hazards like CAT, gust, and wind shear. These hazards pose significant threats to the stability and control of UAM vehicles, particularly during critical flight phases such as takeoff and landing. The system’s high-resolution visual mapping, made possible by integrating Risley prisms and OOFDMA technologies, enables detailed and comprehensive environmental scans with 360° rotational capabilities. Risley prisms are optical devices that allow for precise control of the LiDAR’s scanning pattern, while OOFDMA is a communication technology that enhances the system’s data transmission capabilities. The successful implementation of the cycloidal scanning LiDAR system underscores the importance of advanced visual perception technologies in the evolving landscape of UAM. Future research should continue to refine these technologies, enhancing their accuracy and reliability under diverse atmospheric conditions. Integrating artificial intelligence and machine learning with LiDAR systems can improve dynamic route planning and hazard mitigation, ensuring that UAM operations remain safe and efficient. These technologies can enable the system to adapt to changing conditions and predict potential hazards, further enhancing its safety benefits.