A UAV mapping data acquisition and processing system with adaptive trajectory optimization

Abstract

This invention discloses a UAV mapping data acquisition and processing system with adaptive trajectory optimization, relating to the field of UAV trajectory optimization and mapping processing technology. It includes a ground control terminal, an airborne flight control terminal, and an airborne data processing terminal. The ground control terminal receives input mapping task parameters, including mapping area boundaries, accuracy requirements, and airspace control rules, and generates an initial operational trajectory which is then sent to the airborne terminal. The airborne flight control terminal is equipped with a multi-source sensing module that collects real-time terrain data of the operational area, airspace obstacle data, and its own flight status data. It dynamically adjusts the trajectory according to a three-level priority optimization logic, prioritizing ensuring that the mapping overlap rate meets accuracy requirements and that obstacle avoidance distance meets safety standards, before optimizing flight energy consumption. This invention can effectively adapt to the mapping operation needs of complex terrain and complex airspace, dynamically adjusting the trajectory while ensuring stable data acquisition accuracy across the entire area and eliminating collision risks throughout the flight.

Description

Technical Field

[0001] This invention relates to the field of UAV trajectory optimization and mapping processing technology, and in particular to a UAV mapping data acquisition and processing system with adaptive trajectory optimization. Background Technology

[0002] Unmanned aerial vehicle (UAV) mapping is a core technology for geographic information data acquisition, widely used in natural resource surveys, infrastructure exploration, emergency mapping, and urban 3D modeling. With the deepening of industry applications, the demand for mapping operations in complex terrain and airspace environments continues to grow. The industry's requirements for mapping data accuracy, operational safety, and operational efficiency are constantly increasing. The core requirement lies in balancing the three dimensions of achieving mapping accuracy standards, controllable flight safety, and controllable operational costs.

[0003] Currently, the mainstream UAV mapping systems in the industry are mainly divided into two categories. The first category is the fixed pre-planned trajectory mapping system. Before the operation, the operator plans fixed waypoints, flight altitudes, and data collection intervals in advance at the ground based on the target mapping area boundaries, resolution requirements, and equipment parameters. The UAV flies along the preset trajectory to complete data collection throughout the entire process. All collected data is imported into ground software for processing after the UAV returns. This type of system has simple implementation logic and strong trajectory controllability. It is widely used in conventional mapping scenarios with open, flat, and interference-free airspace. It has high technical maturity and low deployment costs. However, this type of system has obvious application limitations: when facing mountainous and hilly areas with large terrain undulations, the fixed flight altitude will result in insufficient overlap and substandard accuracy at higher terrain levels, while excessive overlap at lower terrain levels will waste flight time; it cannot identify and avoid dynamic obstacles, posing a collision safety risk during operation; when encountering temporary restricted no-fly zones or newly added obstacles, it cannot dynamically adjust the trajectory, and the UAV must be recalled and replanned, resulting in extremely low operation efficiency.

[0004] The second type is adaptive trajectory mapping systems with basic obstacle avoidance capabilities. These systems add a simple ranging sensor module to the UAV. When an obstacle is detected during operation, the system automatically executes a pre-set detour logic, returning to the original trajectory to continue operation after the detour. This type of system solves the basic obstacle avoidance problem of fixed trajectory systems and reduces recall rates in simple interference scenarios. However, the trajectory adjustment logic of this type of system only considers obstacle avoidance requirements and does not include mapping overlap rate and endurance energy consumption in the optimization objectives. The data overlap rate collected during the detour segment has too large a deviation, mostly failing to meet the mapping accuracy requirements, necessitating subsequent re-flights, resulting in high operational costs. Furthermore, the lack of onboard real-time data processing capabilities makes it impossible to determine the validity of the collected data during operation, often resulting in the discovery of unqualified data and the need for a re-flight after returning to base, further reducing operational efficiency.

[0005] In summary, neither of the two existing mainstream surveying and mapping systems can simultaneously adapt to the multi-dimensional operational requirements in complex terrain and airspace scenarios. Problems such as poor overlap rate stability, high re-flight rate, high safety risks, and delayed data feedback have long existed, restricting the large-scale application of UAV surveying and mapping in complex scenarios. Corresponding technical solutions are urgently needed to solve the above-mentioned existing problems. Summary of the Invention

[0006] This invention proposes a UAV mapping data acquisition and processing system with adaptive trajectory optimization to solve the problems mentioned in the prior art, such as poor trajectory adaptability, inability to balance mapping accuracy and flight safety, and lagging data validity verification.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a UAV mapping data acquisition and processing system with adaptive trajectory optimization, comprising a trajectory pre-planning module, an airborne environmental perception module, a trajectory adaptive adjustment module, a multi-source data acquisition module, and a mapping data real-time processing module that are sequentially connected via an airborne CAN bus and an Ethernet hybrid link. The trajectory pre-planning module is used to receive the boundary coordinates of the mapping task area, the mapping ground resolution requirements, the UAV calibration endurance parameters, the no-fly zone coordinates, and the mapping reference coordinate system parameters sent by the ground terminal, and generate an initial mapping trajectory covering the entire operation area. The initial mapping trajectory includes the preset waypoint three-dimensional coordinates, cruising flight altitude, cruising flight speed, data acquisition trigger position, and preset no-fly zone detour point. The airborne environmental perception module is used to cover an airspace of 300 meters in front, behind, to the left, right, up, and down of the drone during its entire flight, and to collect dynamic obstacle information, real-time meteorological parameters, and real-time terrain undulation data of the work area in real time. The response latency of the adaptive trajectory adjustment module is no more than 200 milliseconds. It is used to dynamically adjust the initial mapping trajectory in real time to generate the optimal operation trajectory with three optimization objectives: achieving the mapping overlap rate standard, complying with the obstacle avoidance safety distance, and minimizing flight energy consumption. The time synchronization accuracy of the multi-source data acquisition module is no higher than 1 millisecond, and it is used to synchronously trigger the acquisition of visible light images, laser point clouds, and POS positioning and attitude determination data according to the acquisition trigger position corresponding to the optimal operation trajectory. The real-time mapping data processing module uses an airborne edge computing unit with a processing delay of no more than 1 second. It is used to perform spatiotemporal synchronization registration, various distortion corrections, and initial splicing of regions on the collected multi-source data to generate semi-finished data that meets the mapping accuracy requirements.

[0008] Preferably, when the trajectory pre-planning module generates the initial mapping trajectory, it first completes the rasterization of the target mapping area with a resolution consistent with the ground resolution required by the task. Then, combined with the effective acquisition swath of the acquisition equipment carried by the UAV at the pre-calibrated corresponding flight altitude, it calculates the initial parameters of the trajectory's line spacing, lateral overlap rate, and heading overlap rate. The effective acquisition swath is pre-reserved with a 10% redundancy to offset the swath reduction caused by terrain undulations. The line spacing calculation is additionally superimposed with wind resistance yaw margin. Finally, the generated initial trajectory is pre-verified with no-fly zone avoidance verification and endurance verification to ensure that the initial trajectory has no illegal waypoints and that a single flight's endurance can cover the entire operation area.

[0009] Preferably, the airborne environmental perception module includes a millimeter-wave radar, a binocular vision camera, a meteorological sensor, and an RTK positioning unit. The millimeter-wave radar has a detection range of at least 500 meters and a ranging accuracy of at least 0.1 meters, used to identify the three-dimensional coordinates, movement speed, and movement direction of medium to large dynamic obstacles. The binocular vision camera has a frame rate of at least 30 frames per second, used to identify the position of small obstacles, including birds and low-altitude floating objects. The meteorological sensor collects real-time wind speed, wind direction, and air humidity parameters, providing a basis for correcting flight energy consumption calculations. The RTK positioning unit has a planar positioning accuracy of at least 1 centimeter and a heading accuracy of at least 0.1 degrees, used to output real-time positioning and attitude data of the UAV. The update frequency of dynamic obstacle data is at least 10 Hz to ensure the accuracy of obstacle position prediction.

[0010] Preferably, the adaptive trajectory adjustment module has a built-in multi-objective optimization solution unit. The input parameters of the multi-objective optimization solution unit include initial trajectory parameters, obstacle position and movement parameters obtained in real time, terrain undulation parameters, remaining endurance parameters, and flight energy consumption correction coefficients corresponding to meteorological parameters. The output optimal operational trajectory simultaneously satisfies four constraints: the first is that both the lateral overlap rate and the heading overlap rate are not lower than the task requirement threshold; the second is that the distance between the trajectory and all obstacles is not lower than the preset safe distance; the third is that the remaining endurance can cover the remaining operational area's flight energy consumption; and the fourth is that the trajectory's turning radius does not exceed the UAV's maximum maneuvering turning radius, and the altitude difference change rate does not exceed the UAV's maximum climb / descent rate. The task requirement threshold for the overlap rate can be customized according to the surveying task level.

[0011] Preferably, the multi-objective optimization solution unit uses an improved particle swarm optimization algorithm to solve for the optimal trajectory. During the solution process, a fitness function with three priority levels is set, where the overlap rate meets the standard weight by 50%, the obstacle avoidance safety distance weight by 30%, and the energy consumption weight by 20%. During the solution process, all particle solutions that do not meet the overlap rate standard are first eliminated, then solutions with collision risks are eliminated from the remaining particle solutions, and finally the solution with the highest fitness is selected from the remaining valid solutions as the optimal operating trajectory. The initial population size of the particle swarm is set to 30, and the maximum number of iterations does not exceed 20 to ensure that the solution speed meets the real-time requirements of trajectory adjustment.

[0012] Preferably, the acquisition triggering logic of the multi-source data acquisition module is bound to the optimal operational trajectory depth. The preset position deviation threshold is 0.5 meters and the heading angle deviation threshold is 0.5 degrees. When the deviation between the UAV's three-dimensional coordinates returned by RTK in real time and the three-dimensional coordinates of the acquisition trigger position is less than 0.5 meters, and the deviation between the real-time heading angle and the preset acquisition heading angle is less than 0.5 degrees, hardware trigger signals are synchronously output to the visible light camera and the lidar. The time synchronization accuracy of the trigger signal does not exceed 1 microsecond, ensuring that the spatial correspondence error between the visible light image and the lidar point cloud acquired at the same trigger moment does not exceed 0.1 meters, thus meeting the accuracy requirements of subsequent multi-source data fusion.

[0013] Preferably, the real-time mapping data processing module has a built-in spatiotemporal registration unit. The spatiotemporal registration unit first converts the WGS84 coordinates output by the POS data into the local mapping coordinates required by the task through seven pre-imported parameters. Then, using the millisecond-level timestamp of the POS data as a reference, it binds the visible light image and laser point cloud with a timestamp deviation of no more than 1 millisecond with the POS position and attitude data at the corresponding time to complete the unified mapping of spatial coordinates. The spatial coordinate error of the multi-source data after registration does not exceed 0.2 meters.

[0014] Preferably, the real-time mapping data processing module further includes a distortion correction unit and a fast stitching unit. The distortion correction unit pre-stores the factory calibration distortion parameters of the visible light camera, including radial distortion parameters and tangential distortion parameters, which are used to correct edge distortion of the image. The motion distortion correction of the laser point cloud is combined with the real-time flight speed, angular velocity, and attitude change of the UAV and POS data to interpolate and correct the position of each laser point at the time of acquisition. The fast stitching unit uses the SIFT feature matching algorithm to complete the rapid matching of adjacent images and generates a digital surface model (DSM) and a digital orthophoto image (DOM) by combining the elevation data of the laser point cloud. The resolution of the initially stitched DOM fully meets the ground resolution required by the mission.

[0015] Preferably, it also includes a ground monitoring terminal that is connected to the real-time mapping data processing module via 4G / 5G or a data transmission radio, with a communication distance of not less than 10 kilometers. The ground monitoring terminal is used to receive the optimal operation trajectory, flight status parameters, acquisition progress, and low-resolution preview images of semi-finished mapping data transmitted in real time by the UAV. It also supports manual intervention to issue trajectory adjustment commands, including adding detour points, adjusting flight altitude, and adjusting acquisition interval. The commands issued manually have higher priority than the automatic adjustment results of the trajectory adaptive adjustment module, ensuring human controllability in abnormal scenarios.

[0016] Preferably, the optimization triggering conditions of the trajectory adaptive adjustment module include three categories, any one of which can be met to trigger the module: the first category is the detection of dynamic obstacles within 200 meters in front of the aircraft and 100 meters to the side; the second category is the deviation between the real-time collected terrain undulation data and the pre-stored terrain data exceeding 50 meters; and the third category is the current mapping overlap rate calculated by real-time flight altitude, yaw angle, and trajectory offset being lower than the task requirement threshold and the difference exceeding 5%. After triggering, multi-objective optimization is immediately started to generate a new optimal operational trajectory. After each optimization is completed, the original operational trajectory is automatically overwritten, and the new trajectory is simultaneously uploaded to the ground monitoring terminal for record-keeping.

[0017] Compared with existing technologies, the beneficial effects of this invention are: This invention employs a three-level priority multi-objective optimization trajectory solution logic, simultaneously incorporating mapping overlap rate, obstacle avoidance safety distance, and flight energy consumption into the optimization objectives. It prioritizes ensuring mapping accuracy compliance and flight safety while also considering optimal energy efficiency. This effectively solves the problems in existing technologies where fixed trajectories cannot adapt to terrain undulations or avoid dynamic obstacles, and where ordinary obstacle avoidance systems suffer from substandard overlap rates and wasted endurance after detouring. Under the premise of dynamically adjusting the trajectory, it can ensure stable data acquisition accuracy across the entire operational area, meeting mission requirements, eliminating collision risks throughout the entire process, reducing ineffective energy consumption, and minimizing unnecessary re-flight operations.

[0018] This invention utilizes a real-time mapping data processing link mounted on an airborne edge computing unit to simultaneously complete spatiotemporal registration, distortion correction, and initial regional stitching of multi-source data at the acquisition end. This effectively solves the problem in existing technologies where the validity of collected data can only be verified after the UAV returns, and unqualified data requires a reflight. Data accuracy and compliance can be confirmed during the operation, significantly shortening the overall operation cycle and avoiding the consumption of ineffective operation costs.

[0019] The system architecture of this invention has strong compatibility, supports adaptation to multiple types of UAV flight platforms, and allows for customized settings of parameters such as overlap rate threshold, safety distance, and optimized trigger conditions according to different surveying task levels. It also supports priority control of manual intervention commands, making it suitable for both routine surveying operations in open and flat areas and complex operation scenarios with large terrain undulations and numerous airspace interferences. The generated semi-finished surveying data can be directly connected to existing mainstream surveying post-processing software without additional adaptation or modification. It has a low threshold for industry promotion, and its applicable scenarios cover the vast majority of civilian surveying needs, making it highly valuable for promotion and application. Attached Figure Description

[0020] Figure 1 This is an overall flowchart of a UAV mapping data acquisition and processing system with adaptive trajectory optimization proposed in this invention; Figure 2 Flowchart for initial surveying and mapping trajectory pre-planning and safety verification; Figure 3 Flowchart for adaptive trajectory adjustment triggering and multi-target optimization; Figure 4 Flowchart for triggering control of high-precision synchronous acquisition of multi-source data; Figure 5 Flowchart for real-time registration, stitching, and distortion correction of surveying data. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figures 1 to 5 This invention discloses a UAV mapping data acquisition and processing system with adaptive trajectory optimization, comprising a trajectory pre-planning module, an airborne environmental perception module, a trajectory adaptive adjustment module, a multi-source data acquisition module, and a mapping data real-time processing module, which are sequentially connected via an airborne CAN bus and an Ethernet hybrid link.

[0023] The CAN bus is used to connect low-bandwidth, high-real-time modules such as flight control and positioning, with a communication latency of no more than 10ms; the Ethernet is used to connect high-bandwidth modules such as data acquisition and processing, with a transmission rate of no less than 1Gbps.

[0024] The trajectory pre-planning module is deployed on the ground industrial control terminal and supports importing inputs such as task area boundaries in SHP format, custom ground resolution, UAV battery capacity, and equipment calibration parameters. The generated initial trajectory can be directly exported as a KMZ format file supported by the UAV flight control system.

[0025] All hardware of the airborne environmental perception module is mounted on the top of the drone fuselage via shock-absorbing brackets, providing unobstructed coverage of the drone's omnidirectional airspace.

[0026] The adaptive trajectory adjustment module is deployed on the airborne edge computing unit, using NVIDIA Jetson Xavier NX as the core computing power carrier, ensuring that the trajectory solution response latency is no higher than 200ms. The multi-source data acquisition module achieves full-device time synchronization through the PTP precise time protocol, with a synchronization accuracy of no higher than 1ms. The real-time mapping data processing module is also deployed on the aforementioned edge computing unit, employing a multi-threaded parallel processing architecture. The processing latency for a single frame image + point cloud is no higher than 1s, and the generated semi-finished data can be directly imported into mainstream mapping post-processing software for further processing.

[0027] In this invention, the specific implementation logic of the trajectory pre-planning module for generating the initial trajectory is also disclosed: When the trajectory pre-planning module generates the initial mapping trajectory in advance, it first completes the rasterization of the target mapping area according to the accuracy consistent with the ground resolution required by the task. For example, when the task requires a ground resolution of 5cm, the target area is divided into 5cm×5cm raster units, and each raster corresponds to a collection coverage identifier.

[0028] Combined with the initial parameters for calculating the effective acquisition swath of the acquisition equipment carried by the UAV at the pre-calibrated corresponding flight altitude, the calibration method for the effective acquisition swath is as follows: at the calibration flight altitude, acquire images of the ground calibration board, calculate the resolution deviation between the image edge and center, and reserve 10% redundancy to offset the swath reduction caused by terrain undulations. For example, at a flight altitude of 100 meters, the theoretical swath of the visible light camera is 115m, and after reserving 10% redundancy, 103m is taken as the effective swath.

[0029] The route spacing calculation includes an additional wind resistance yaw margin. When the local maximum crosswind speed is 5m / s, the margin is set to 2m. The final route spacing can ensure that the initial value of the lateral overlap rate meets the mission requirements.

[0030] After generating the initial flight path, two checks are automatically performed: the first is a no-fly zone avoidance check, which uses a polygon collision detection algorithm to compare the coordinates of waypoints with the boundary coordinates of the no-fly zone. If a waypoint falls into the no-fly zone, an alternative flight path is automatically generated. The second is an endurance check, which calculates the total energy consumption of the flight path based on 0.02 kWh of energy consumption per kilometer of flight. The total energy consumption is required to not exceed 80% of the battery's rated capacity, with a 20% emergency endurance redundancy reserved. After the check passes, the final initial flight path is output.

[0031] This invention also discloses the specific hardware configuration and parameters of the airborne environmental perception module: the airborne environmental perception module includes a millimeter-wave radar, a binocular vision camera, a meteorological sensor, and an RTK positioning unit. The millimeter-wave radar is a Texas Instruments IWR6843ISK, with a detection range of at least 500m, a ranging accuracy of at least 0.1m, a data update frequency of 10Hz, and the ability to identify the three-dimensional coordinates, moving speed, and moving direction of dynamic obstacles larger than 0.5m.

[0032] The binocular vision camera uses the ZED 2i, with a frame rate of 30 frames per second and a detection range of at least 200m. It can identify small obstacles larger than 0.1m, including birds, low-flying objects, and kites. The meteorological sensor uses a combination of an MS5611 barometer and a hot-wire anemometer, with a wind speed measurement accuracy of 0.1m / s, a wind direction measurement accuracy of 1°, and an air humidity measurement accuracy of 1%RH. The collected meteorological parameters are used to correct the flight energy consumption calculation coefficient, with a correction coefficient of 1.2 for headwinds and 0.8 for tailwinds.

[0033] The RTK positioning unit uses the Huace P2 module, with a planar positioning accuracy of no less than 1cm, a heading accuracy of no less than 0.1 degrees, and a data update frequency of 20Hz. It can output real-time 3D coordinates and attitude data of the UAV with UTC timestamps. The dynamic obstacle data update frequency is no less than 10Hz, ensuring that the obstacle position prediction lead is no less than 10s.

[0034] In this invention, the specific constraint rules of the multi-objective optimization solution unit built into the trajectory adaptive adjustment module are also disclosed: the input parameters of the multi-objective optimization solution unit include the initial trajectory parameters, the obstacle position and movement parameters obtained in real time, the terrain undulation parameters, the remaining endurance parameters, and the flight energy consumption correction coefficients corresponding to the meteorological parameters.

[0035] The output optimal operational trajectory simultaneously satisfies four constraints: First, both the lateral overlap rate and the forward overlap rate must not be lower than the task requirement threshold. The overlap rate is calculated as: forward overlap rate = 1 - (speed × acquisition interval) / acquisition swath width along the forward direction; lateral overlap rate = 1 - line spacing / lateral acquisition swath width. The overlap rate threshold for conventional mapping tasks can be set to 20%, while for high-precision mapping it can be set to 30%. Second, the distance between the trajectory and all obstacles must not be lower than the preset safety distance. The safety distance for static obstacles is 10m, and the safety distance for dynamic obstacles is 30m. Third, the remaining flight endurance must cover the remaining operational area's flight energy consumption, with a 10% emergency redundancy added during calculation. Fourth, the turning radius of the trajectory must not exceed the UAV's maximum maneuvering turning radius of 5m, and the altitude difference change rate must not exceed the UAV's maximum climb / descent rate of 3m / s, avoiding trajectory points exceeding the UAV's maneuverability.

[0036] This invention also discloses the specific implementation logic of the improved particle swarm optimization algorithm used in the multi-objective optimization solution unit: the multi-objective optimization solution unit uses the improved particle swarm optimization algorithm to solve for the optimal trajectory, and sets a fitness function with three priority levels during the solution process. The core velocity update formula of the algorithm is as follows: ;in This is the inertia weight, with a value of 0.7, used to balance global search and local search capabilities; , The learning factor is set to 2. , A random number between 0 and 1; This is the optimal trajectory solution for the i-th particle in the k-th iteration; This is the globally optimal trajectory solution at the k-th iteration; This represents the current trajectory solution of the i-th particle in the k-th iteration; Let be the search step size of the i-th particle in the k-th iteration.

[0037] The fitness function formula is: ;in For the overlap rate to meet the standard, the actual overlap rate must be greater than or equal to the required threshold. The value is 1, otherwise = Actual overlap rate / Required threshold; For obstacle avoidance safety, the distance between the flight path and all obstacles must be greater than or equal to the safe distance. The value is 1 if it is set to 1, otherwise the value is 0. For energy efficiency, =Minimum theoretical energy consumption / Actual energy consumption of the current trajectory, with a value ranging from 0 to 1.

[0038] When solving, first eliminate all Collision risk solution, then eliminate If the overlap rate of a solution does not meet the standard, the solution with the highest fitness F among the remaining valid solutions is selected as the optimal operational trajectory. The initial population size of the particle swarm is set to 30, and the maximum number of iterations does not exceed 20, ensuring that the solution time does not exceed 150ms, thus meeting the real-time requirements for trajectory adjustment. The core pseudocode of the algorithm is as follows: / / Improved core logic of particle swarm optimization algorithm init_particles(30); / / Initialize 30 particles, each particle corresponding to a set of trajectory parameters for(int iter=0; iter<20; iter++){ for each (Particle p in particle_list){ F1 = calc_overlap_score(p); / / Calculate the overlap score F2 = calc_obstacle_score(p); / / Calculate the obstacle avoidance score if(F2 == 0) continue; / / Remove particles that pose a collision risk F3 = calc_energy_score(p); / / Calculate the energy score p.fitness = 0.5*F1 + 0.3*F2 + 0.2*F3; update_pbest(p); / / Update the individual best } update_gbest(); / / Update the global best update_velocity_pos(particle_list); / / Update the velocity and position of all particles } export_route(gbest); / / Output the optimal route This invention also discloses a specific implementation method for the acquisition triggering logic of the multi-source data acquisition module: the acquisition triggering logic of the multi-source data acquisition module is bound to the optimal operational trajectory depth, with a pre-set position deviation threshold of 0.5m and a heading angle deviation threshold of 0.5°. The trigger control unit is implemented using FPGA, which acquires the current three-dimensional coordinates and heading angle of the UAV output by RTK in real time, calculates the Euclidean distance between the current coordinates and the acquisition trigger position, and the distance calculation formula is: ; in The current coordinates of the drone. The coordinates of the preset acquisition trigger point. When Furthermore, when the deviation between the current heading angle and the preset acquisition heading angle is less than 0.5°, the FPGA synchronously outputs hardware trigger pulses to the visible light camera and lidar. The rising edge error of the trigger pulse does not exceed 1μs, which can ensure that the spatial correspondence error between the visible light image and the lidar point cloud acquired at the same trigger moment does not exceed 0.1m, fully meeting the accuracy requirements of subsequent multi-source data fusion.

[0039] This invention also discloses the specific implementation logic of the spatiotemporal registration unit built into the real-time mapping data processing module: The spatiotemporal registration unit first converts the WGS84 coordinate system coordinates output from the POS data into the local mapping coordinate system coordinates required by the task through pre-imported Bursa seven parameters. The seven-parameter conversion formula is as follows: ; in Three-dimensional coordinates in the local surveying coordinate system; These are three-dimensional coordinates in the WGS84 coordinate system. Scale factor; It is a 3×3 rotation matrix, calculated from three Euler angles; The translation amounts in three directions are given. The seven parameters mentioned above can be obtained by calibration using control points with at least three known coordinates within the survey area. After the conversion is completed, the visible light image and laser point cloud with a timestamp deviation of no more than 1ms are bound to the POS position and attitude data at the corresponding time to complete the unified mapping of spatial coordinates. The spatial coordinate error of the registered multi-source data does not exceed 0.2m.

[0040] This invention also discloses the specific implementation logic of distortion correction and rapid stitching in the real-time mapping data processing module: the real-time mapping data processing module further includes a distortion correction unit and a rapid stitching unit. The distortion correction unit pre-stores the factory calibration distortion parameters of the visible light camera, including three radial distortion parameters. and 2 tangential distortion parameters The formula for image distortion correction is: ; ;in These are the pixel coordinates of the distorted image. The distance from the pixel to the image center is used; correction eliminates stretching distortion at the image edges. Motion distortion correction for laser point clouds uses a linear interpolation method, with the acquisition time for each laser point... Time between two adjacent POS points , When the time interval is between, the corrected coordinates of the point are It can eliminate point cloud shifts caused by attitude changes during drone flight.

[0041] The fast stitching unit uses the SIFT feature matching algorithm to extract 128-dimensional feature points from each frame of image. After matching the feature points of adjacent images, the RANSAC algorithm is used to remove mismatched points. Finally, the stitching generates a digital orthophoto image DOM. Combined with the registered laser point cloud, a digital surface model (DSM) is generated. The resolution of the initially stitched DOM fully meets the ground resolution required by the mission.

[0042] This invention also discloses the specific implementation logic of the ground monitoring terminal: the system further includes a ground monitoring terminal connected to the real-time mapping data processing module via 4G / 5G or a data transmission radio, with a communication distance of no less than 10km. The data transmission radio uses DJI Data Transmission 3, supporting low-bandwidth control command transmission within a 15km range; the 5G module uses Huawei MH5000, supporting high-speed data transmission of 100Mbps, and can transmit real-time previews of collected low-resolution semi-finished data. The ground monitoring terminal is equipped with visual host computer software, which can display parameters such as the drone's flight position, optimal operating trajectory, data collection progress, and remaining battery life in real time. The data collection progress is calculated as the proportion of the number of covered grid cells to the total number of grid cells.

[0043] The ground monitoring terminal also supports manual intervention to issue trajectory adjustment commands, including adding detour points, adjusting flight altitude, and adjusting data collection intervals. Manually issued commands have higher priority than the automatic adjustment results of the trajectory adaptive adjustment module. After the command is issued, the current trajectory solving process will be interrupted immediately to execute the manual command, ensuring human controllability in abnormal scenarios.

[0044] This invention also discloses the specific judgment logic of the optimization triggering conditions of the trajectory adaptive adjustment module: the optimization triggering conditions of the trajectory adaptive adjustment module include three categories, and it can be triggered if any one of them is met.

[0045] The first category is when a dynamic obstacle is detected in a fan-shaped airspace 200 meters in front of the aircraft and 100 meters to the side, and the obstacle's moving speed is greater than 0.5 m / s, and a collision risk is predicted. The second category is where the deviation between the ground elevation data obtained by real-time lidar altimetry and the pre-stored DEM digital elevation model data exceeds 50 meters, and the original flight path's altitude parameters are no longer suitable for the terrain. The third category involves situations where the current mapping overlap rate, calculated using real-time flight altitude, yaw angle, and track offset, is lower than the task requirement threshold and the difference exceeds 5%, indicating that the data acquisition accuracy cannot meet the requirements. Upon triggering, multi-objective optimization is immediately initiated to generate a new optimal operational track. After each optimization, the original operational track is automatically overwritten, and the new track is simultaneously uploaded to the ground monitoring terminal for storage and record-keeping, facilitating subsequent traceability.

[0046] Example 1 Application of forest land resource survey and mapping in southwestern mountainous areas This surveying mission is aimed at investigating forest resources in a 10-square-kilometer mountainous area in Southwest China. The mission requires a ground resolution of 10 centimeters and an overlap rate of no less than 20% in both the flight direction and the lateral direction. The terrain of the work area is complex, with a maximum overall topographic relief of 80 meters. There are small dynamic obstacles such as wild birds in the airspace, and there is no public communication base station coverage. Data transmission radios are used to achieve air-to-ground communication.

[0047] The flight path pre-planning module is deployed on the ground control terminal. First, it imports the boundary coordinates of the mission area and the parameters of the mapping reference coordinate system. The target mapping area is then rasterized with a precision of 10 cm × 10 cm, with each grid corresponding to a unique data acquisition coverage identifier. Subsequently, combining the UAV's calibration endurance parameters and the technical specifications of the onboard data acquisition equipment, the effective data acquisition swath width at the corresponding flight altitude is calculated, with a 10% redundancy reserved to offset swath width reduction caused by terrain undulations. When calculating the flight path spacing, a wind resistance yaw margin corresponding to the local maximum crosswind speed is added, ultimately determining the initial flight altitude to be 150 meters above the average ground elevation of the area, the flight path spacing to be 138 meters, and the initial overlap rate to be set at 20%. After the initial flight path is generated, the system automatically performs no-fly zone avoidance verification and endurance verification. It uses a polygon collision detection algorithm to compare waypoints with the pre-stored no-fly zone boundary coordinates and automatically generates detour paths for waypoints that fall into the no-fly zone. The total flight path energy consumption is calculated based on the energy consumption per unit distance of flight to ensure that the total energy consumption does not exceed 80% of the battery's rated capacity and reserves 20% of emergency endurance redundancy. After the verification is passed, the initial flight path is sent to the airborne terminal.

[0048] All hardware of the airborne environmental perception module is mounted on the top of the UAV fuselage via shock-absorbing brackets, providing unobstructed coverage of the airspace within a 300-meter radius in all directions. The millimeter-wave radar detects dynamic obstacles larger than 0.5 meters within a 500-meter range, outputting their 3D coordinates, speed, and direction of movement, with a data update frequency of at least 10 Hz. The binocular vision camera operates at 30 frames per second, identifying small obstacles larger than 0.1 meters within a 200-meter range, including birds and low-flying objects. Weather sensors collect real-time wind speed, direction, and humidity parameters of the operational area, providing a basis for correcting flight energy consumption calculations. The energy consumption correction coefficient is increased in headwinds and decreased in tailwinds. The RTK positioning unit outputs real-time 3D coordinates and attitude data for the UAV, with a planar positioning accuracy of at least 1 cm and a heading accuracy of at least 0.1 degrees, with a data update frequency of 20 Hz. The update frequency of all dynamic obstacle data is at least a preset threshold, ensuring that the lead time for obstacle position prediction meets safety requirements.

[0049] The adaptive trajectory adjustment module is deployed on the airborne edge computing unit, with a response latency of no more than 200 milliseconds. In this mission, the terrain undulation deviation trigger threshold is set to 30 meters, the dynamic obstacle detection range is adjusted to 300 meters in front of the fuselage, and trajectory optimization is automatically triggered every 30 seconds. The module has a built-in multi-objective optimization solution unit. The input parameters include initial trajectory parameters, obstacle parameters obtained from real-time perception, terrain parameters, remaining endurance parameters, and energy consumption correction coefficients corresponding to meteorological parameters. The solution unit adopts an improved particle swarm optimization algorithm, with an initial population size of 30 and a maximum number of iterations not exceeding 20. During the solution process, a three-level priority fitness function is set, with overlap rate meeting the target weight accounting for 50%, obstacle avoidance safety distance accounting for 30%, and energy consumption accounting for 20%. The algorithm first eliminates all trajectory solutions with collision risk, then eliminates solutions with insufficient overlap rate from the remaining solution set, and finally selects the solution with the lowest energy consumption from the solution set that meets the safety and accuracy requirements as the optimal operating trajectory. In this mission, the system prioritizes adjusting the flight altitude to adapt to the terrain undulations, and only adjusts the horizontal track when obstacles are encountered. After each optimization is completed, the original operational track is automatically overwritten and uploaded to the ground monitoring terminal for record-keeping.

[0050] The multi-source data acquisition module achieves full-device time synchronization through a precise time protocol, with a synchronization accuracy of no more than 1 millisecond. The acquisition trigger logic is bound to the optimal operational flight path depth, with pre-set position deviation thresholds of 0.5 meters and heading angle deviation thresholds of 0.5 degrees. The trigger control unit is implemented using a field-programmable gate array (FPGA), which reads the UAV's 3D coordinates and heading angle from the RTK output in real time and calculates the spatial distance between the current position and the preset acquisition trigger point. When both position deviation and heading angle deviation meet the threshold requirements, a hardware trigger pulse is simultaneously output to the visible light camera and LiDAR. The rise time error of the trigger pulse does not exceed 1 microsecond, ensuring that the spatial correspondence error between the visible light image and the LiDAR point cloud acquired at the same time does not exceed 0.1 meters. Simultaneously, the corresponding POS positioning and attitude data are acquired.

[0051] The real-time mapping data processing module adopts a multi-threaded parallel processing architecture, with a processing latency of no more than 1 second for a single frame of imagery and point cloud. The spatiotemporal registration unit first converts the globally universal coordinate system output from the POS data into the local mapping coordinate system required by the task using pre-imported seven parameters. Then, using the millisecond-level timestamps of the POS data as a benchmark, it binds visible light images and laser point clouds with timestamp deviations of no more than 1 millisecond to the corresponding position and attitude data, completing a unified spatial coordinate mapping. The spatial coordinate error of the registered multi-source data does not exceed 0.2 meters. The distortion correction unit calls the pre-stored factory calibration parameters of the visible light camera to correct radial and tangential distortions in the images. For motion distortion in the laser point cloud, it interpolates and corrects the acquisition position of each laser point by combining the real-time flight speed, angular velocity, and attitude changes of the UAV. The rapid stitching unit extracts feature points from each frame of imagery, matches feature points from adjacent images, and removes mismatched points, completing the rapid stitching of adjacent images. Combined with the registered laser point cloud, it generates a digital orthophoto and a digital surface model. The resolution of the initial stitched result fully meets the ground resolution required by the task.

[0052] The ground monitoring terminal establishes a communication connection with the airborne terminal via a data transmission radio, with a communication distance of no less than 10 kilometers. The monitoring terminal is equipped with visual host computer software that displays the UAV's flight position, optimal operational trajectory, flight status parameters, data acquisition progress, and remaining battery power in real time. The data acquisition progress is calculated based on the proportion of covered grid cells to the total number of grid cells. Simultaneously, it receives low-resolution preview images of the semi-finished mapping data transmitted from the airborne terminal. It supports manual intervention commands from operators, such as adding detour points, adjusting flight altitude, and adjusting data acquisition intervals. Manual commands have higher priority than the automatic adjustment results from the trajectory adaptive adjustment module, ensuring operational controllability in abnormal scenarios.

[0053] In this embodiment, the original fixed-track mapping scheme suffers from severe imbalances in overlap rate due to terrain undulations. The overlap rate in valley areas is only 12%, failing to meet accuracy requirements, while the overlap rate in mountain peak areas reaches as high as 40%, resulting in wasted flight time and posing a risk of collision. Furthermore, the coverage area per sortie is limited, requiring multiple re-flights. With this system, the flight track can adaptively adjust its altitude in real time according to the terrain, maintaining a stable overlap rate of 19% to 22% across the entire operational area, fully meeting the mission accuracy requirements. The system can identify and avoid dynamic obstacles such as wild birds in real time, eliminating collision safety hazards throughout the process. The coverage area per sortie is increased by 25%, eliminating the need for re-flights. The onboard real-time data processing capability verifies data validity during operation, avoiding the discovery of substandard data upon return. The overall operation cycle is shortened by 30%, significantly reducing the operational cost and safety risks of mapping complex terrain.

[0054] Example 2 Application of preliminary surveying and mapping in new infrastructure construction in urban suburbs This surveying mission serves as a preliminary investigation for a new infrastructure project covering 5 square kilometers in the suburbs of a city in the Yangtze River Delta. The mission requires a ground resolution of 5 centimeters and an overlap rate of no less than 30% in both the forward and lateral directions. The work area is located at the border between the city and the suburbs, and the airspace contains other dynamic obstacles such as low-altitude drones and kites, and there may be temporary no-fly zones. High-speed air-to-ground data transmission will be achieved using a 5G communication link.

[0055] The trajectory pre-planning module imports the task area boundary file in SHP format and rasterizes the target mapping area into a 5cm x 5cm grid. Combining the UAV's endurance parameters and the data acquisition equipment calibration data, the initial flight altitude is determined to be 100 meters. The effective data acquisition swath at this altitude is calculated, with a 10% terrain redundancy allowance. Crosswind yaw margin is added during the flight path spacing calculation, ultimately determining a flight path spacing of 72 meters and an initial overlap rate of 30%. After the initial trajectory is generated, the system automatically performs no-fly zone avoidance and endurance checks, generating preset detour points for pre-stored permanent no-fly zones. The total trajectory energy consumption is calculated based on energy consumption per unit distance, ensuring that the total energy consumption does not exceed 80% of the battery's rated capacity, and reserving sufficient emergency endurance. After successful verification, the initial trajectory is exported as a standard format file supported by the UAV flight control system and sent to the onboard flight control terminal.

[0056] The airborne environmental perception module provides omnidirectional coverage of the airspace surrounding the aircraft. Millimeter-wave radar detects medium-sized dynamic obstacles within a 500-meter range, outputting their motion parameters and 3D coordinates, with a data update frequency of 10 Hz. A binocular vision camera identifies small low-altitude obstacles within a 200-meter range, including kites, balloons, and small low-flying drones. Meteorological sensors collect real-time wind speed, wind direction, and air humidity data, correcting flight energy consumption calculation coefficients in real time and providing accurate energy consumption data for trajectory optimization. The RTK positioning unit provides centimeter-level real-time positioning and attitude determination data, with a planar positioning accuracy of no less than 1 cm and a heading accuracy of no less than 0.1 degrees, with a data update frequency of 20 Hz, providing a high-precision spatiotemporal reference for trajectory adjustment and data acquisition. In this mission, the lateral detection trigger threshold for dynamic obstacles was adjusted to 150 meters to improve the coverage of airspace perception and predict potential collision risks in advance.

[0057] The adaptive trajectory adjustment module has three optimization objectives: achieving the required mapping overlap rate, complying with obstacle avoidance safety distance, and minimizing flight energy consumption, with a response latency of no more than 200 milliseconds. The module incorporates a multi-objective optimization solution unit. Input parameters include initial trajectory parameters, real-time obstacle information, terrain data, remaining flight endurance, and weather correction coefficients. The output optimal trajectory must simultaneously satisfy four conditions: achieving the required overlap rate, eliminating collision risk, ensuring sufficient flight endurance, and meeting UAV maneuverability constraints. The solution unit employs an improved particle swarm optimization algorithm, initializing 30 particles to represent different trajectory solutions, with a maximum iteration count of 20. The solution process strictly follows a three-level priority logic: first, all collision solutions with obstacle distances less than the safe distance are eliminated (static obstacle safety distance is set to 10 meters, dynamic obstacle safety distance to 30 meters); second, solutions with overlap rates below the task requirement threshold are eliminated; finally, the solution with the lowest energy consumption among the remaining valid solutions is selected as the optimal operational trajectory. In this mission, the system set temporary no-fly zone avoidance as the highest priority. Upon receiving the temporary no-fly zone instruction from the ground, it immediately triggered trajectory optimization and generated a new detour trajectory within 200 milliseconds.

[0058] The multi-source data acquisition module achieves high-precision synchronous acquisition from multiple sensors, with a time synchronization accuracy of no more than 1 millisecond. The acquisition trigger logic is bound to the optimal operational trajectory depth. When the deviation between the UAV's real-time 3D coordinates and the preset acquisition trigger point is less than 0.5 meters, and the deviation between the real-time heading angle and the preset acquisition heading angle is less than 0.5 degrees, the trigger control unit synchronously outputs a hardware trigger pulse, triggering the visible light camera and LiDAR to simultaneously complete data acquisition and record the corresponding POS positioning and attitude data. The time error of the trigger pulse does not exceed 1 microsecond, ensuring that the spatial correspondence accuracy of the multi-source data meets the requirements of subsequent fusion processing.

[0059] The real-time mapping data processing module relies on the airborne edge computing unit to process data in real time. The spatiotemporal registration unit first performs coordinate system transformation, converting POS data from the globally universal coordinate system to local mapping coordinates. Then, using millisecond-level timestamps as a benchmark, it precisely binds visible light imagery, laser point clouds, and position and attitude data, achieving spatiotemporal unification of multi-source data. The distortion correction unit performs radial and tangential distortion correction on the visible light imagery, eliminating stretching deformation at image edges; it also performs motion distortion correction on the laser point cloud, using interpolation methods to correct point cloud position shifts caused by changes in UAV flight attitude. The rapid stitching unit uses a feature matching algorithm to stitch adjacent images together, removing mismatched points to generate a digital orthophoto image. Combined with the laser point cloud data, it generates a digital surface model. The initial stitched result achieves a ground resolution of 5 centimeters, fully meeting the mission accuracy requirements.

[0060] The ground monitoring terminal connects to the airborne terminal via a 5G communication link with a communication distance of no less than 10 kilometers, supporting high-speed data transmission of 100Mbps. The monitoring terminal receives in real time the optimal operational trajectory, flight status, data acquisition progress, and low-resolution preview images of the semi-finished mapping data transmitted back by the UAV. Operators can monitor the operation process and data quality in real time. The system allows for the highest level of manual intervention, supporting the issuance of commands such as adding detour points, adjusting flight altitude, and adjusting data acquisition intervals. Once a manual command is issued, the current trajectory solving process is immediately interrupted, and the manual operation is executed first, ensuring operational safety and controllability in abnormal scenarios such as temporary control measures.

[0061] In this embodiment, the original fixed-track mapping system could not identify dynamic obstacles, resulting in collisions with other low-altitude drones. Furthermore, when encountering temporary no-fly zones, the system required emergency recalls of drones to replan their routes, causing delays of up to several hours and severely impacting operational progress. With this new system, dynamic obstacles within a 150-meter range can be detected in real time, and the system automatically adjusts its trajectory to avoid them, ensuring no safety incidents occur throughout the process. Upon receiving a temporary no-fly zone instruction, the system can generate a new detour trajectory within milliseconds, eliminating the need to recall drones and improving operational efficiency by 40%. The onboard real-time data processing capability allows for data verification during operations, avoiding situations where data is found to be unqualified upon return, requiring a re-flight. The system supports priority manual intervention, enabling flexible responses to various emergencies in complex urban airspace, significantly improving the safety and efficiency of urban area mapping operations.

[0062] Reference Figure 1 This diagram macroscopically illustrates the space-ground collaborative architecture and full lifecycle data flow process of the UAV mapping system. The system is divided into two parts: the ground end and the airborne end. The mission begins with the trajectory pre-planning module deployed on the ground industrial control terminal, which generates an initial flight path by combining the survey area boundary and hardware parameters. After the UAV takes off, the airborne environmental perception module on the top of the fuselage uses millimeter-wave radar and a visual camera to detect surrounding dynamic obstacles and meteorological information in all directions. This real-time perception data is quickly sent to the trajectory adaptive adjustment module deployed on the airborne edge computing unit. Once a sudden change in terrain or obstacle is encountered, the system will immediately calculate the optimal obstacle avoidance trajectory within milliseconds. While ensuring safe flight, the multi-source data acquisition module relies on a precise time protocol to achieve high-precision synchronous exposure of visible light and lidar. Finally, the acquired raw data flows directly into the mapping data real-time processing module for registration and stitching. The semi-finished image can be transmitted back to the ground monitoring terminal in real time via a communication link for manual inspection or intervention, forming a complete automated closed loop.

[0063] Reference Figure 2This diagram details the scientific calculations and dual safety blocking mechanisms for the initial flight path before mission execution. The pre-planning process first divides the target mapping area into finely divided grids according to the ground resolution specified for the mission. Then, the system calculates the effective acquisition swath width of the camera based on flight altitude and mandates a redundancy reserve to compensate for image reduction caused by terrain undulations. When calculating flight path spacing, additional wind resistance and yaw margins are added to ensure lateral overlap meets standards. After the initial flight path parameters are formed, two rigorous automated checks must be performed: the first is a no-fly zone avoidance check, using polygon collision detection to ensure all waypoints are far from controlled airspace; otherwise, an alternative flight path is automatically generated. The second is an endurance check, where the system estimates energy consumption based on total mileage and mandates the retention of emergency endurance redundancy. Only paths that pass both spatial and energy checks are allowed to be output and sent to the UAV flight control system.

[0064] Reference Figure 3 This diagram focuses on the trajectory reconstruction logic of a UAV encountering unexpected situations in a complex low-altitude environment. The adaptive optimization engine is always on standby, with three parallel monitoring optimization trigger conditions: the appearance of a fast-moving dynamic obstacle ahead, a significant deviation between the measured elevation by LiDAR and the pre-stored terrain model, or strong winds causing the actual mapping overlap rate to drop significantly below the threshold. Upon triggering any condition, the system immediately activates the multi-objective optimization solution unit. This unit employs an improved particle swarm optimization algorithm, setting overlap rate compliance, obstacle avoidance safety distance, and theoretical flight energy consumption as three-level fitness functions. Within a very short iteration cycle, the algorithm first ruthlessly eliminates all invalid particles with collision risks and insufficient overlap rates, then selects the path with the highest comprehensive fitness score from the remaining safe solutions as the optimal operational trajectory. Once the new trajectory is generated, it instantly takes over flight control and is simultaneously backed up to the ground monitoring terminal.

[0065] Reference Figure 4 This figure illustrates the precise, hardware-level collaborative logic of the mapping payload during high-speed flight. To ensure perfect alignment between the point cloud and the imagery, data acquisition actions and the optimal operational trajectory are deeply spatially bound. The system's trigger control core is implemented using a field-programmable gate array (FPGA) chip, providing extremely high real-time performance. This unit receives real-time 3D coordinates and heading angle data output from the positioning module at extremely high frequency and calculates the spatial Euclidean distance between the current fuselage position and the preset ideal acquisition trigger point in real time. Only when both the position deviation and heading angle deviation are reduced to the minimum threshold allowed by the system, the control chip instantly and synchronously sends hardware-level trigger pulses to the visible light camera and lidar. This hard synchronization mechanism with microsecond-level errors eliminates the time difference caused by multiple sensors operating independently at the physical source, ensuring extremely high spatial correspondence accuracy.

[0066] Reference Figure 5This image depicts the real-time processing pipeline of massive amounts of raw data acquired by an airborne edge computing unit. The first step is spatiotemporal benchmark unification: the system extracts pre-stored transformation parameters, accurately maps the globally universal coordinate system to the local mapping coordinate system specified for the task, and uses millisecond-level timestamps to firmly bind position and attitude data with images and point clouds. Next, the core distortion correction unit corrects the radial and tangential stretching caused by lenses in visible light images using factory calibration parameters; for laser point clouds, linear interpolation is used to offset motion offsets caused by UAV maneuvers. After image cleansing, the rapid stitching unit takes over, using feature matching algorithms to extract image feature points, eliminating mismatches, and quickly stitching them together. Finally, a digital orthophoto and a semi-finished digital surface model that meet the resolution requirements are directly generated in the air, greatly reducing the post-processing cycle.

[0067] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A UAV mapping data acquisition and processing system with adaptive trajectory optimization, characterized in that, It includes a trajectory pre-planning module, an airborne environment perception module, a trajectory adaptive adjustment module, a multi-source data acquisition module, and a surveying and mapping data real-time processing module, which are connected in sequence via communication. The trajectory pre-planning module is used to generate an initial mapping trajectory based on the area of ​​the mapping task, the mapping resolution requirements, and the UAV's endurance parameters. The initial mapping trajectory includes preset waypoint coordinates, flight altitude, flight speed, and data acquisition trigger position. The airborne environmental perception module is used to collect dynamic obstacle information, meteorological parameters, and terrain undulation data of the operational airspace in real time during the flight of the UAV. The adaptive trajectory adjustment module is used to adjust the initial mapping trajectory in real time to generate the optimal operation trajectory with the optimization goals of achieving the mapping overlap rate, avoiding obstacles, and minimizing energy consumption. The multi-source data acquisition module is used to synchronously acquire visible light images, laser point clouds, and POS positioning and attitude determination data according to the acquisition trigger position corresponding to the optimal operation trajectory. The real-time mapping data processing module is used to synchronously register, correct distortion, and initially stitch together the collected multi-source data to generate semi-finished mapping data.

2. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 1, characterized in that, When the trajectory pre-planning module generates the initial mapping trajectory, it first rasterizes the target mapping area, and then calculates the initial parameters of the flight path spacing, lateral overlap rate, and heading overlap rate of the trajectory by combining the effective acquisition swath width of a single UAV, and finally generates an initial trajectory covering the entire target mapping area.

3. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 1, characterized in that, The airborne environmental perception module includes a millimeter-wave radar, a binocular vision camera, a meteorological sensor, and an RTK positioning unit, which are used to collect the three-dimensional coordinates, movement speed, wind direction and speed of dynamic obstacles, and real-time positioning data of the UAV, respectively. The update frequency of the dynamic obstacle data is not less than a preset threshold.

4. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 1, characterized in that, The adaptive trajectory adjustment module has a built-in multi-objective optimization solution unit. The input parameters of the multi-objective optimization solution unit include initial trajectory parameters, obstacle parameters obtained in real time, terrain parameters, and remaining endurance parameters. The output optimal operation trajectory satisfies the constraints that the lateral overlap rate and the heading overlap rate are not lower than the task requirement threshold, there is no collision risk, and the remaining endurance can cover the remaining operation area.

5. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 4, characterized in that, The multi-objective optimization solution unit uses an improved particle swarm optimization algorithm to solve for the optimal trajectory. During the solution process, the trajectory solutions that meet the mapping overlap rate requirements are retained first. Then, the solution with the lowest energy consumption and no collision risk is selected from the solution set that meets the overlap rate requirements as the final optimal operation trajectory.

6. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 1, characterized in that, The acquisition triggering logic of the multi-source data acquisition module is bound to the optimal operation track. When the deviation between the real-time POS data of the UAV and the preset acquisition trigger position in the optimal operation track is less than a preset threshold, the visible light camera and lidar are simultaneously triggered to complete the data acquisition.

7. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 1, characterized in that, The real-time mapping data processing module has a built-in spatiotemporal registration unit. The spatiotemporal registration unit uses the timestamp of the POS data as a reference to uniformly transform the spatial coordinates of visible light images and laser point clouds collected at the same time into the mapping coordinate system, thus completing the initial registration of multi-source data.

8. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 7, characterized in that, The real-time mapping data processing module also includes a distortion correction unit and a fast stitching unit. The distortion correction unit is used to correct lens distortion of visible light images and motion distortion of laser point clouds. The fast stitching unit is used to quickly generate the initial DOM and DSM of the mapping area based on the registered multi-source data.

9. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 1, characterized in that, It also includes a ground monitoring terminal that is connected to the real-time mapping data processing module. The ground monitoring terminal is used to receive the optimal operation trajectory, data acquisition progress, and semi-finished mapping data transmitted in real time. It also supports manual intervention to adjust trajectory parameters and data acquisition parameters.

10. The UAV mapping data acquisition and processing system with adaptive trajectory optimization according to claim 5, characterized in that, The optimization triggering conditions of the adaptive trajectory adjustment module include any one of the following: sensing the presence of dynamic obstacles within a preset range, terrain undulation deviation exceeding a preset threshold, or mapping overlap rate under the current flight parameters being lower than the task requirement threshold. Once triggered, multi-objective optimization solution is immediately started to generate a new optimal operational trajectory.

Images