A Method for Pattern Decision-Making and Path Planning in Unmanned Aerial Vehicle Mapping of Complex Areas

Abstract

This invention provides a mode decision-making and path planning method for UAV mapping in complex areas, relating to the field of UAV technology. This method addresses the shortcomings of traditional UAV mapping in automatically adapting to different terrains with significant elevation variations. Based on an elevation difference threshold algorithm, it determines the UAV's flight mode at different terrain elevations, generating mapping sub-regions using different flight modes. This avoids the complexity of manually setting flight modes and delineating mapping areas, ensuring that each mapping sub-region adopts the optimal mapping path based on its terrain characteristics. Path planning is performed for each mapping sub-region, recording the coordinates of the entry / exit points. A mathematical model for single-UAV multi-mapping sub-region coverage path planning is constructed, using a variable neighborhood search (VNS)-genetic algorithm to obtain the route with the shortest flight distance passing through all task points within the mapping area, improving mapping efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a method for pattern decision-making and path planning in UAV mapping of complex areas. Background Technology

[0002] With the rapid development of drone technology, drones are increasingly widely used in surveying and mapping. High-precision cameras, lidar, and other sensors onboard drones can acquire high-resolution terrain and landform data, which are widely applied in geographic information systems, urban planning, and environmental monitoring. However, traditional drone surveying methods have limitations in complex terrain, especially in mountainous areas with significant elevation differences. Typically, operators need to manually plan the drone's flight path beforehand based on terrain features, dividing the survey area into contour flight zones and ground-hugging flight zones. Contour flight is suitable for flat areas, while ground-hugging flight requires the drone to maintain a constant altitude close to the ground surface to ensure surveying accuracy. Because areas with significant elevation differences have high requirements for flight paths, this planning process often requires considerable manpower and time. Furthermore, manually planned paths may lack sufficient accuracy, leading to duplicate or missed areas, affecting the overall efficiency of the surveying. Summary of the Invention

[0003] The technical problem to be solved by this invention is to address the shortcomings of the prior art by providing an algorithm for automatically dividing the surveying area based on terrain. Based on the known boundary and terrain information of the area to be surveyed, the algorithm sets different flight modes according to the changes in terrain height of the area to be surveyed, and automatically divides the area to be surveyed into surveying sub-areas suitable for different flight modes. Furthermore, the invention provides a shortest path planning algorithm for each surveying sub-area to generate the shortest flight path connecting all surveying sub-areas, thereby ensuring the surveying accuracy under different terrain conditions.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: The present invention provides a method for mode decision-making and path planning for UAV mapping in complex areas, specifically including the following steps:

[0005] Step 1: Establish a mapping mode decision algorithm based on altitude difference threshold to determine the flight mode of the UAV at different terrain altitudes and generate mapping sub-regions using different flight modes;

[0006] Step 1.1: Set different flight modes according to the terrain of the area to be surveyed;

[0007] Step 1.2: Obtain the data required by the mapping mode decision algorithm based on the height difference threshold, including the length and width of the area covered by the mapping image captured by the UAV in a single shot, the acceptable height difference Δh_threshold of the UAV, and the digital elevation model (DEM) data of the area to be mapped, where each DEM data point represents the altitude value at that location.

[0008] Step 1.3: Based on the maximum and minimum elevation values ​​of all DEM data points within the area to be surveyed, calculate the overall elevation difference ΔH within the area to be surveyed, as shown in the following formula:

[0009] ΔH=H max -H min (1)

[0010] Among them, H max and H min These are the elevations of the highest and lowest points within the area to be surveyed;

[0011] Step 1.4: Based on the acceptable altitude difference Δh_threshold for the UAV, divide the overall altitude difference ΔH of the area to be mapped into several equal-altitude flight zones of different altitude levels. The altitude difference between two adjacent altitude levels shall not exceed the acceptable altitude difference Δh_threshold for the UAV.

[0012] Step 1.5: Define a sliding window whose size matches the area covered by the mapping image captured by the UAV in a single shot. Use this sliding window to traverse the area to be mapped. Based on the altitude difference within the current sliding window and the acceptable altitude difference Δh_threshold for the UAV, determine the flight mode of the UAV within the current sliding window area.

[0013] Define the elevation difference Δh within the current sliding window as:

[0014] Δh=h max -h min (2)

[0015] Among them, h max and h min These are the altitudes of the highest and lowest points within the current sliding window, respectively.

[0016] The flight mode of the UAV within the current sliding window is determined based on the altitude difference Δh within the current sliding window. If Δh > Δh_threshold, the flight mode of the area covered by the window is set to ground-hugging flight mode. If Δh ≤ Δh_threshold, the flight mode is set to equal altitude flight mode for each altitude level based on the highest and lowest altitudes of the window. When the UAV is in ground-hugging flight mode, the corresponding altitude control system is used to control the flight altitude.

[0017] Step 1.6: For the sliding window region where the flight mode is determined, obtain the flight modes of its adjacent sliding window regions. Based on the principles of adjacency and feature consistency, merge adjacent sliding window regions with the same flight mode, record the boundary point coordinates of the merged region, and generate the mapping sub-regions in the area to be mapped that use different flight modes.

[0018] Step 2: Determine the minimum number of task points within each surveying sub-region by overlapping the drone's shooting area, perform path planning for each surveying sub-region, and record the coordinates of the entry / exit points of each surveying sub-region.

[0019] Step 2.1: Draw several task windows with length L and width W to cover the survey sub-area. Take the center point of each task window as the flight task point of the UAV. Here, L is the length of the area covered by the survey image taken by the UAV in a single shot, and W is the width of the area covered by the survey image taken by the UAV in a single shot.

[0020] Step 2.2: Determine the minimum number of task points within each surveying sub-region by overlapping the UAV shooting areas, and extract the center point of each task window covering the surveying sub-region;

[0021] Step 2.2.1: Set the scene overlap in the UAV's heading and lateral directions to ensure that the boundary of the surveyed sub-region is within the coverage area of ​​the task window, and obtain the number of task windows covering the surveyed sub-region in the UAV's heading and lateral directions;

[0022] Set the scene overlap in the drone's flight direction The distance s between any two adjacent mission points along the flight path is calculated using the following formula:

[0023]

[0024] Based on the total span of the area to be mapped along the UAV's heading and the distance between two adjacent task points, the number of task windows for the UAV in the heading and lateral directions is obtained;

[0025] The number of task windows n in the drone's flight path is:

[0026]

[0027] Where l represents the total span of the surveyed sub-region along the UAV's flight path. This is for rounding up;

[0028] Set the side-up view overlap of the drone If the value is 0, then the distance between any two adjacent task points in the lateral direction is W. The number of task windows m in the lateral direction for the UAV is calculated as follows:

[0029]

[0030] Where w is the total span of the surveyed sub-region in the direction of the UAV;

[0031] Step 2.2.2: Adjust the relative position of the side-to-side task windows of the UAV, determine the minimum number of task windows covering the survey sub-area, and extract the center point of each task window at this time;

[0032] The difference Δw between the total lateral window span and the total lateral span of the surveyed sub-region is calculated as follows:

[0033] Δw=m*Ww (6)

[0034] Divide Δw into k equal parts, and successively translate Δw / k distance in the lateral direction to generate k different combinations of task window positions. Iterate through each combination of task window relative positions to obtain the minimum number of task windows covering the mapping sub-region, and extract the center point coordinates of the UAV heading and lateral task windows.

[0035] Step 2.3: Based on the "ox-plowing" surveying method, perform path planning for each surveying sub-region to obtain fixed paths. After completing the path planning for a surveying sub-region, record the coordinates of the entry / exit points of the fixed paths for each surveying sub-region.

[0036] Step 3: Construct a mathematical model for coverage path planning of multiple sub-regions of a single UAV, and obtain the route with the shortest flight distance that passes through all task points in the area to be mapped based on the variable neighborhood search VNS-genetic algorithm;

[0037] Step 3.1: Construct a mathematical model for coverage path planning of multiple survey sub-regions using a single UAV, and set constraints based on the surveying task;

[0038] After path planning for all surveying sub-regions, multiple candidate entry / exit points are generated for each surveying sub-region. The set of candidate entry / exit points for all surveying sub-regions is set as V = {v1, v2, ..., v...}. N}, i∈[1,N], where v i Let N be the number of candidate entry / exit points; set the path set W = {w ij |vi ,v j Let w ∈V, j∈[1,N],j≠i} represent the set of all paths, where w ij The distance from the i-th candidate entry / exit point to the j-th candidate entry / exit point; generate a point group for all candidate entry / exit points belonging to the same mapping sub-region. If there are a total of M mapping sub-regions, then generate M point groups.

[0039] To ensure that the UAV maps all sub-regions without repeatedly mapping any particular sub-region, the following objective function model is established with the shortest distance required for the UAV to complete the mapping task of all sub-regions as the objective:

[0040]

[0041] Among them, D m Let D be the distance that the UAV needs to fly to perform a mapping task on the m-th sub-region, where D = {D1, D2, ..., D...} m ,…,D M} represents the set of distances that a UAV needs to fly to perform a mapping task on a sub-region. r is the total flight distance required for the UAV to connect all mapping sub-areas. ij It is a 0 / 1 variable, r ij =1 indicates that the drone has arrived at the j-th entry / exit point from the i-th entry / exit point; otherwise, r = 1. ij =0; This is the sum of the distances the UAV needs to fly to perform mapping tasks in each of the various mapping sub-regions, calculated by the distances between the UAVs in each mapping sub-region. Minimize, thus optimizing the objective function;

[0042] Step 3.2: Optimize the objective function based on VNS variable neighborhood search-genetic algorithm to obtain the total path that passes through all task points in the area to be mapped and has the shortest flight distance;

[0043] (1) Initialize the genetic algorithm. For the entry / exit points of the fixed path in each mapping sub-region and the task points outside the mapping sub-region, arbitrarily select a fixed road segment in the mapping sub-region as the starting road segment. Starting from an entry / exit point of the starting road segment, use a greedy algorithm to select the point that is closest to the current entry / exit point and has not yet reached the task point as the next point to be visited, and obtain S candidate paths as S initial individuals of the population.

[0044] (2) Perform crossover and mutation operations on the remaining task points of the S initial individuals of the population after removing the fixed path to generate new offspring individuals;

[0045] (3) Calculate the fitness of each new offspring individual and select an optimal path solution set from the current population using the "roulette wheel" algorithm;

[0046] For the UAV path planning problem, a fitness function is set to evaluate the performance of each offspring individual. The fitness function is shown in the following formula:

[0047]

[0048] (4) Set the number of iterations of the VNS algorithm and set its initial number of iterations to zero. Use the 2-opt swap strategy to adjust the path order, that is, select any two edges from the path and swap their connection to form a new path, thereby reducing the path length or optimizing other objective function values. Each time, select two points from the remaining task points excluding the fixed path and swap their order in the original path, and calculate the fitness of the new path solution set. If the optimal path solution is obtained before the number of iterations, proceed to step (5); otherwise, repeat this step until the preset number of VNS algorithm iterations is reached.

[0049] (5) Update the optimal path and proceed to the next iteration of the genetic algorithm;

[0050] (6) Repeat steps (2) to (5) until the preset number of iterations of the genetic algorithm is reached. During the iteration process, individuals in the population will continuously evolve and gradually approach the optimal path scheme.

[0051] (7) Change the order of the fixed path candidate entry / exit points in turn, and repeat steps (1) to (6) to obtain multiple path planning schemes for all entry / exit points. Select the path planning scheme with the highest fitness as the total path that passes through all task points in the area to be mapped and has the shortest flight distance.

[0052] The ground-flying mode corresponds to an altitude control system, which specifically includes: a lidar, a data processing module, and a flight control module.

[0053] LiDAR acquires the real-time altitude data of a drone by emitting a laser beam toward the surface of the area to be mapped and receiving the signal reflected from the surface of the area at a certain sampling frequency. The drone's altitude data is then transmitted to the data processing module.

[0054] The data processing module receives real-time altitude data of the UAV sent by the lidar, filters and denoises the data to improve its accuracy, and generates corresponding control commands based on the real-time altitude data of the UAV and transmits them to the flight control module.

[0055] The specific method by which the data processing module generates the corresponding control commands is as follows:

[0056] Calculate the actual height H of the UAV relative to the surface of the area to be mapped. 实际 :

[0057]

[0058] Where c is the speed of light, t is the round-trip time of the laser, and λ is the wavelength of the laser.

[0059] The distance between the drone and the surface of the area to be mapped is compared with the preset expected altitude to calculate the current altitude deviation e(t):

[0060] e(t) = H 期望 -H 实际 (10)

[0061] Based on the altitude deviation, the data processing module uses a PID control algorithm to control the drone's flight altitude. The PID control algorithm is shown in the following formula:

[0062]

[0063] Where u1(t) is the proportional control part, K p The proportional gain is u2(t), the integral control section, used to eliminate steady-state error, and K. i It is the integral gain; u3(t) is the differential control part, used to predict the changing trend of altitude error and make corrections, K d is the differential gain; u(t) is the final control signal used to adjust the altitude of the UAV. If the distance between the UAV and the ground surface of the survey area is higher than the desired altitude, a command to decrease altitude is generated; if the distance between the UAV and the ground surface of the survey area is lower than the desired altitude, a command to increase altitude is generated.

[0064] The flight control module receives control commands from the data processing module and adjusts the UAV's flight attitude and speed to achieve precise control of flight altitude.

[0065] The beneficial effects of adopting the above technical solution are as follows: The present invention provides a mode decision-making and path planning method for UAV mapping in complex areas. This method, after knowing the boundary and altitude information of the mapping area, automatically divides the area to be mapped into different mapping sub-regions. Based on the altitude fluctuations within each sub-region, it automatically determines which flight mode to use for mapping, avoiding the complexity of manually setting flight modes and delineating mapping areas. This ensures that each sub-region can adopt the optimal mapping path according to its terrain characteristics. Simultaneously, within each sub-region, the method can also generate flight paths adapted to that sub-region based on an optimized path planning algorithm, thereby significantly reducing the workload of manual flight path planning, improving mapping efficiency and accuracy, saving labor costs, and enhancing the autonomous mapping capabilities of UAVs in complex terrain, demonstrating significant application prospects. Attached Figure Description

[0066] Figure 1 A flowchart of a mapping mode decision-making algorithm based on a height difference threshold provided in an embodiment of the present invention;

[0067] Figure 2 A heading overlap map provided for an embodiment of the present invention, wherein the scene overlap of the UAV heading is set to 66%;

[0068] Figure 3 This is a schematic diagram of a single mapping sub-region path planning method provided in an embodiment of the present invention, wherein (a) is a schematic diagram of a task window combination that covers the mapping sub-region with the minimum number of task windows, (b) is a schematic diagram of a task point obtained by extracting the center point coordinates of the UAV heading and lateral task windows, and (c) is a schematic diagram of a fixed path generated by path planning of the mapping sub-region based on the "ox-plowing" mapping method.

[0069] Figure 4 This is a schematic diagram of the path planning of the area to be surveyed provided in an embodiment of the present invention. (a) is a schematic diagram of a fixed path obtained by path planning of the survey sub-area based on the "ox-plowing" surveying method. A, B, C, and D are candidate entry / exit points. (b) is a schematic diagram of the task points of the area to be surveyed, which includes multiple independent survey sub-areas. (c) is a schematic diagram of the area to be surveyed after path planning of multiple survey sub-areas. A, B, C, and D are candidate entry / exit points belonging to the same survey sub-area. E, F, G, and H are candidate entry / exit points belonging to another survey sub-area.

[0070] Figure 5 This is a schematic diagram of the 2-opt switching strategy provided in an embodiment of the present invention;

[0071] Figure 6 A flowchart illustrating the operation of an altitude control system in a ground-hugging flight mode, provided as an embodiment of the present invention. Detailed Implementation

[0072] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0073] This embodiment of a method for mode decision-making and path planning in UAV mapping of complex areas specifically includes the following steps:

[0074] Step 1: Establish a mapping mode decision algorithm based on altitude difference threshold to determine the flight mode of the UAV at different terrain altitudes and generate mapping sub-regions using different flight modes;

[0075] When conducting surveying tasks across a region, the diversity of terrain places different demands on the flight strategies of unmanned aerial vehicles (UAVs). In relatively flat areas with little terrain variation, UAVs can fly at a fixed altitude, maintaining consistent image resolution and reducing travel and time costs. However, in areas with significant terrain undulations and elevation differences, maintaining a fixed altitude may lead to inconsistent image resolution, affecting the accuracy of the surveying results. To minimize travel and time costs for UAVs in completing surveying tasks, it is necessary to flexibly adjust the UAV's surveying mode based on terrain changes. In this embodiment, a surveying mode decision algorithm based on elevation difference thresholds is as follows: Figure 1 As shown.

[0076] Step 1.1: Set different flight modes according to the terrain of the area to be surveyed;

[0077] In this embodiment, the flight modes include: ground-hugging flight mode, single-altitude level contour flight mode, and multi-altitude level contour flight mode. The ground-hugging flight mode is suitable for terrains with large undulations, such as hills and cliffs; the single-altitude level contour flight mode is suitable for flat areas, such as plains and sea level; and the multi-altitude level contour flight mode is suitable for flat areas containing multiple different altitudes, such as terraced fields.

[0078] Step 1.2: Obtain the data required by the mapping mode decision algorithm based on the height difference threshold, including the length and width of the area covered by the mapping image captured by the UAV in a single shot, the acceptable height difference Δh_threshold of the UAV, and the digital elevation model (DEM) data of the area to be mapped, where each DEM data point represents the altitude value at that location.

[0079] Step 1.3: Determine the maximum and minimum elevation values ​​of all DEM data points within the area to be surveyed, and calculate the overall elevation difference ΔH within the area to be surveyed, as shown in the following formula:

[0080] ΔH=H max -H min (1)

[0081] Among them, H max and H min These are the elevations of the highest and lowest points within the area to be surveyed;

[0082] Step 1.4: Based on the acceptable altitude difference Δh_threshold for the UAV, divide the overall altitude difference ΔH of the area to be mapped into several equal-altitude flight zones of different altitude levels. The altitude difference between two adjacent altitude levels shall not exceed the acceptable altitude difference Δh_threshold for the UAV.

[0083] Step 1.5: Define a sliding window whose size matches the area covered by the mapping image captured by the UAV in a single shot. Use this sliding window to traverse the area to be mapped. Based on the altitude difference within the current sliding window and the acceptable altitude difference Δh_threshold for the UAV, determine the flight mode of the UAV within the current sliding window area.

[0084] Define the elevation difference Δh within the current sliding window as:

[0085] Δh=h max -h min (2)

[0086] Among them, h max and h min These are the altitudes of the highest and lowest points within the current sliding window, respectively.

[0087] The flight mode of the UAV within the current sliding window is determined based on the altitude difference Δh within the current sliding window. If Δh > Δh_threshold, the flight mode of the area covered by the window is set to ground-hugging flight mode. If Δh ≤ Δh_threshold, the flight mode is set to equal altitude flight mode for each altitude level based on the highest and lowest altitudes of the window. When the UAV is in ground-hugging flight mode, the corresponding altitude control system is used to control the flight altitude.

[0088] Step 1.6: For the sliding window region where the flight mode is determined, obtain the flight modes of its adjacent sliding window regions. Based on the principles of adjacency and feature consistency, merge adjacent sliding window regions with the same flight mode, record the boundary point coordinates of the merged region, and generate the mapping sub-regions in the area to be mapped that use different flight modes.

[0089] For a sliding window region divided into different altitude flight modes, it is merged with an adjacent sliding window region of the same altitude flight mode, and the coordinates of the boundary points of the merged region are recorded to obtain multiple altitude flight mode regions of different altitude levels; for a sliding window region divided into a ground-level flight mode, if its adjacent sliding window region is a ground-level flight mode sliding window region or a continuous sliding window region of different altitude flight modes, these sliding windows are merged, and the coordinates of the boundary points of the merged region are recorded to obtain multiple ground-level flight mode regions.

[0090] Step 2: Determine the minimum number of task points within each surveying sub-region by overlapping the drone's shooting area, perform path planning for each surveying sub-region, and record the coordinates of the entry / exit points of each surveying sub-region.

[0091] Step 2.1: Draw several task windows with length L and width W to cover the survey sub-area. Take the center point of each task window as the flight task point of the UAV. Here, L is the length of the area covered by the survey image taken by the UAV in a single shot, and W is the width of the area covered by the survey image taken by the UAV in a single shot.

[0092] Step 2.2: Determine the minimum number of task points within each surveying sub-region by overlapping the UAV shooting areas, and extract the center point of each task window covering the surveying sub-region;

[0093] Step 2.2.1: Set the scene overlap in the UAV's heading and lateral directions to ensure that the boundary of the surveyed sub-region is within the coverage area of ​​the task window, and obtain the number of task windows covering the surveyed sub-region in the UAV's heading and lateral directions;

[0094] To improve the efficiency of UAV mapping tasks and complete the mapping task with the fewest UAV flight points, redundant data is defined as follows: if a mapping image in the UAV mapping image dataset is deleted, and all points within the coverage area of ​​that mapping image can be observed by at least 3 other mapping images, then that mapping image is redundant data.

[0095] To ensure that the mapping images captured by the UAV do not contain redundant data, i.e., that all points in any mapping image coverage area can be observed by at least 3 mapping images, in this embodiment, a scene overlap degree is set along the UAV's flight path. Side-view image overlap of the drone The distance s between every two mission points along the flight path is calculated using the following formula:

[0096]

[0097] The distance s between two adjacent mission points along the flight path is L / 3. The UAV region is then shifted sequentially along the flight path by L / 3 to obtain the flight path overlap map, as shown below. Figure 2 As shown, the points in the three-times-observed portion of the heading overlap map are all observed in at least three other mapping images. Therefore, as long as the boundary of the sub-region to be mapped is within the three-times-observed portion of the heading overlap map, accurate observation of the mapping sub-region can be achieved with the minimum number of task points of the UAV.

[0098] Based on the total span of the area to be mapped along the UAV's heading and the distance between two adjacent task points, the number of task windows for the UAV in the heading and lateral directions is obtained;

[0099] The number of task windows n in the drone's flight path is:

[0100]

[0101] Where l represents the total span of the surveyed sub-region along the UAV's flight path. This is for rounding up;

[0102] Set the side-up view overlap of the drone If the value is 0, then the distance between any two adjacent task points in the lateral direction is W. The number of task windows m in the lateral direction for the UAV is calculated as follows:

[0103]

[0104] Where w is the total span of the surveyed sub-region in the direction of the UAV;

[0105] Step 2.2.2: Adjust the relative position of the side-to-side task windows of the UAV, determine the minimum number of task windows covering the survey sub-area, and extract the center point of each task window at this time;

[0106] When the total span of the lateral window m*W is slightly larger than the total lateral span w of the area to be mapped, the task window is continuously shifted to find the case where the number of task points used by the UAV in the lateral direction is minimized. The center point of each task window at this point is extracted for path planning. Specifically, the difference Δw between the total span of the lateral window and the total lateral span of the sub-area is calculated as shown in the following formula:

[0107] Δw=m*Ww (6)

[0108] Divide Δw into k equal parts, and successively shift them upwards by a distance of Δw / k, generating k different combinations of task window positions. Iterate through each combination of task window positions to obtain the minimum number of task windows covering the survey sub-region, such as... Figure 3 As shown in (a), the coordinates of the center point of the UAV's heading and lateral task windows are extracted, as follows: Figure 3As shown in (b), the value of k should be determined appropriately according to the situation. A larger value of k will increase the accuracy, but will increase the calculation time and reduce efficiency.

[0109] Step 2.3: Based on the "ox-plowing" surveying method, perform path planning for each surveying sub-region to obtain fixed paths. After completing the path planning for a surveying sub-region, record the coordinates of the entry / exit points of the fixed paths for each surveying sub-region.

[0110] The fixed paths obtained by path planning for each survey sub-region based on the "ox-plowing" surveying method are as follows: Figure 3 As shown in (c);

[0111] Step 3: Construct a mathematical model for coverage path planning of multiple sub-regions of a single UAV, and obtain the route with the shortest flight distance that passes through all task points in the area to be mapped based on the variable neighborhood search VNS-genetic algorithm;

[0112] For the overall area to be mapped, due to the different flight altitudes of different flight modes, frequent entry and exit from different mapping sub-areas will lead to frequent altitude switching of the UAV, thus affecting the overall efficiency of the mapping. Therefore, for each independent mapping sub-area, this embodiment only sets one pair of entry / exit points. For a certain mapping sub-area in the area to be mapped, the path planning result using the method described in step 2 is as follows: Figure 4 As shown in (a), the mapping sub-region contains four candidate entry / exit points A, B, C, and D. When point A is used as the entry point, point C becomes the exit point; when point C is used as the entry point, point A becomes the exit point, and so on. It can be seen that there is a one-to-one correspondence between the entry and exit points. When one of the candidate entry points is selected as the starting point, the exit point is also determined. However, changes in the entry / exit points do not alter the flight path of the UAV within the mapping sub-region. For a mapping area containing multiple independent mapping sub-regions, the path planning results using the method described in step 2 are as follows: Figure 4 As shown in (b), when connecting other independent mapping sub-regions, the UAV needs to plan the shortest path traversing all regions, thus making the "ox-plowing" route planning method unsuitable; for the entire mapping area, such as Figure 4 As shown in (c), the UAV mapping task can be summarized as follows: the UAV needs to find a route that connects the AB (BA) / CD (DC) route, the EF (FE) / GH (HG) route, and other waypoints to the shortest route.

[0113] Step 3.1: Construct a mathematical model for coverage path planning of multiple survey sub-regions using a single UAV, and set constraints based on the surveying task;

[0114] After path planning for all survey sub-regions, multiple candidate entry / exit points are generated for each survey sub-region. The set of candidate entry / exit points for all survey sub-regions is set as V = {v1, v2, ..., v...}. N}, i∈[1,N], where v i Let N be the number of candidate entry / exit points; set the path set W = {w ij |v i ,v j Let w ∈V, j∈[1,N],j≠i} represent the set of all paths, where w ij The distance from the i-th candidate entry / exit point to the j-th candidate entry / exit point; generate a point group for all candidate entry / exit points belonging to the same mapping sub-region. If there are a total of M mapping sub-regions, then generate M point groups.

[0115] To ensure that the UAV maps all sub-regions without repeatedly mapping any sub-region, the following objective function model is established based on the shortest distance required for the UAV to complete the mapping task of all sub-regions:

[0116]

[0117] Among them, D m Let D be the distance that the UAV needs to fly to perform a mapping task on the m-th sub-region, where D = {D1, D2, ..., D...} m ,…,D M} represents the set of distances that a UAV needs to fly to perform a mapping task on a sub-region. r is the total flight distance required for the UAV to connect all mapping sub-areas. ij It is a 0 / 1 variable, r ij =1 indicates that the drone has arrived at the j-th entry / exit point from the i-th entry / exit point; otherwise, r = 1. ij =0; This is the sum of the distances the UAV needs to fly to perform mapping tasks in each of the various mapping sub-regions. Since the flight paths of the UAV to perform mapping tasks in each mapping sub-region have already been planned, this is achieved by calculating the distances between the UAVs in each mapping sub-region. Minimize, thus optimizing the objective function;

[0118] Step 3.2: Optimize the objective function based on VNS variable neighborhood search-genetic algorithm to obtain the total path that passes through all task points in the area to be mapped and has the shortest flight distance;

[0119] Genetic algorithms, as a heuristic search algorithm, possess powerful global optimization capabilities and are suitable for solving complex path planning problems. First, a population iteration count is set, and a population is randomly generated. The fitness of individuals in the population is evaluated, and individuals are selected using a "roulette wheel" method, where individuals with higher fitness have a greater probability of being selected and have a better chance of entering the next generation for inheritance and evolution, forming a new generation of the population. Then, crossover and mutation operations are performed, and the fitness of each individual in the new population is calculated. However, genetic algorithms are prone to getting stuck in local convergence and may fail to find the optimal solution.

[0120] The Variable Neighborhood Search (VNS) algorithm is an improved local search algorithm that achieves a good balance between concentration and dispersion by alternating searches using different neighborhood structures. Starting from an initial solution, the VNS algorithm continuously searches for better solutions in the neighborhood of the current solution using the neighborhood structure. If a better solution is found, the current solution is updated; this process is repeated iteratively until a termination condition is met. By changing the neighborhood structure, the VNS algorithm can better escape local optima and find approximate optimal solutions to problems more quickly.

[0121] Based on VNS variable neighborhood search—genetic algorithm, better solutions can be found in different neighborhood structures, improving the performance of local search methods. The alternation of population iteration in the genetic algorithm and local optimization in the variable neighborhood search achieves a balance between global exploration and refined local optimization. Through the alternating execution and data interaction of the genetic algorithm and the variable neighborhood search algorithm, complementary advantages are formed, ensuring that the UAV mapping path covers all path points and is the shortest, improving mapping efficiency and data accuracy.

[0122] This embodiment utilizes a VNS-based variable neighborhood search-genetic algorithm for path optimization. In the initial stage, the genetic algorithm generates a population of individuals through crossover and mutation operations, and uses a "roulette wheel" method to select individuals with higher fitness, ensuring population diversity and the global nature of the evolutionary direction. Subsequently, a variable neighborhood search algorithm is applied to the non-fixed path portions of the population individuals to optimize the quality of the path solution using a local search method. Specifically, this includes the following steps:

[0123] (1) Initialize the genetic algorithm. For the entry / exit points of the fixed path in each mapping sub-region and the task points outside the mapping sub-region, arbitrarily select a fixed road segment in the mapping sub-region as the starting road segment. Starting from an entry / exit point of the starting road segment, use a greedy algorithm to select the point that is closest to the current entry / exit point and has not yet reached the task point as the next point to be visited, and obtain S candidate paths as S initial individuals of the population.

[0124] (2) Perform crossover and mutation operations on the remaining task points of the S initial individuals of the population after removing the fixed path to generate new offspring individuals;

[0125] (3) Calculate the fitness of each new offspring individual and select an optimal path solution set from the current population using the "roulette wheel" algorithm;

[0126] To address the UAV path planning problem, a fitness function is set to evaluate the performance of each offspring. In this embodiment, the goal is to obtain the total path that passes through all task points within the mapping area with the shortest flight distance. The fitness function of the offspring is defined as the reciprocal of the distance between the UAV and each mapping sub-region, as shown in the following formula. The shorter the distance between the UAV and each mapping sub-region, the higher the fitness of the offspring.

[0127]

[0128] (4) Set the number of iterations for the VNS algorithm, and set its initial number of iterations to zero. Use a 2-opt swapping strategy to adjust the path order, such as... Figure 5 As shown, any two edges are selected from the path, and their connection is swapped to form a new path, thereby reducing the path length or optimizing other objective function values; each time, two points are selected from the remaining task points excluding the fixed path and their order in the original path is swapped, and the fitness of the new path solution set is calculated; if the optimal path solution is obtained before the number of iterations, then proceed to step (5), otherwise repeat this step until the preset number of VNS algorithm iterations is reached.

[0129] (5) Update the optimal path and proceed to the next iteration of the genetic algorithm;

[0130] (6) Repeat steps (2) to (5) until the preset number of iterations of the genetic algorithm is reached. During the iteration process, individuals in the population will continue to evolve and gradually approach the optimal path scheme.

[0131] (7) Change the order of the fixed path candidate entry / exit points in turn, and repeat steps (1) to (6) to obtain multiple path planning schemes for all entry / exit points. Select the path planning scheme with the highest fitness as the total path that passes through all task points in the area to be mapped and has the shortest flight distance.

[0132] The altitude control system corresponding to the ground-hugging flight mode, such as... Figure 6 As shown, this is used to control the flight altitude of the UAV in ground-hugging flight mode during surveying tasks. This ensures that the UAV maintains a certain flight altitude above the ground surface in terrains with significant elevation changes, thereby guaranteeing the accuracy and reliability of the surveying results and the pixel stability of the surveyed images. The altitude control system in ground-hugging flight mode is implemented based on real-time measurement of the altitude between the UAV and the surface of the surveyed area using LiDAR technology.

[0133] The ground-flying mode corresponds to an altitude control system, which specifically includes: a lidar, a data processing module, and a flight control module.

[0134] LiDAR acquires the real-time altitude data of a drone by emitting a laser beam toward the surface of the area to be mapped and receiving the signal reflected from the surface of the area at a certain sampling frequency. The drone's altitude data is then transmitted to the data processing module.

[0135] The data processing module receives real-time altitude data of the UAV sent by the lidar, filters and denoises the data to improve its accuracy, and generates corresponding control commands based on the real-time altitude data of the UAV and transmits them to the flight control module.

[0136] The specific method by which the data processing module generates the corresponding control commands is as follows:

[0137] Calculate the actual height H of the UAV relative to the surface of the surveyed area. 实际 :

[0138]

[0139] Where c is the speed of light, t is the round-trip time of the laser, and λ is the wavelength of the laser.

[0140] The distance between the drone and the surface of the area to be mapped is compared with the preset expected altitude to calculate the current altitude deviation e(t):

[0141] e(t) = H 期望 -H 实际 (10)

[0142] Based on the altitude deviation, the data processing module uses a PID control algorithm to control the drone's flight altitude. The PID control algorithm is shown in the following formula:

[0143]

[0144] Where u1(t) is the proportional control part, K p The proportional gain is u2(t), the integral control section, used to eliminate steady-state error, and K. i It is the integral gain; u3(t) is the differential control part, used to predict the changing trend of altitude error and make corrections, K d is the differential gain; u(t) is the final control signal used to adjust the altitude of the UAV. If the distance between the UAV and the ground surface of the survey area is higher than the desired altitude, a command to decrease altitude is generated; if the distance between the UAV and the ground surface of the survey area is lower than the desired altitude, a command to increase altitude is generated.

[0145] The flight control module receives control commands from the data processing module and adjusts the UAV's flight attitude and speed to achieve precise control of flight altitude.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A method for pattern decision-making and path planning in unmanned aerial vehicle (UAV) mapping of complex areas, characterized in that: Includes the following steps: Step 1: Establish a mapping mode decision algorithm based on altitude difference threshold to determine the flight mode of the UAV at different terrain altitudes and generate mapping sub-regions using different flight modes; Step 2: Determine the minimum number of task points within each surveying sub-region by overlapping the drone's shooting area, perform path planning for each surveying sub-region, and record the coordinates of the entry / exit points of each surveying sub-region. Step 3: Construct a mathematical model for coverage path planning of multiple sub-regions of a single UAV, and obtain the route with the shortest flight distance that passes through all task points in the area to be mapped based on the variable neighborhood search VNS-genetic algorithm; Step 2 includes the following steps: Step 2.1: Draw several task windows with length L and width W to cover the survey sub-area. Take the center point of each task window as the flight task point of the UAV. Here, L is the length of the area covered by the survey image taken by the UAV in a single shot, and W is the width of the area covered by the survey image taken by the UAV in a single shot. Step 2.2: Determine the minimum number of task points within each surveying sub-region by overlapping the UAV shooting areas, and extract the center point of each task window covering the surveying sub-region; Step 2.3: Based on the ox-plowing surveying method, perform path planning for each surveying sub-region to obtain fixed paths. After completing the path planning for a surveying sub-region, record the coordinates of the entry / exit points of the fixed paths for each surveying sub-region. The specific method for step 2.2 is as follows: Step 2.2.1: Set the scene overlap in the UAV's heading and lateral directions to ensure that the boundary of the surveyed sub-region is within the coverage area of ​​the task window, and obtain the number of task windows covering the surveyed sub-region in the UAV's heading and lateral directions; Set the scene overlap in the drone's flight direction Calculate the distance s between every two adjacent mission points along the flight path, as shown in the following formula: (3) Based on the total span of the area to be mapped along the UAV's heading and the distance between two adjacent task points, the number of task windows for the UAV in the heading and lateral directions is obtained; The number of task windows n in the drone's flight path is: (4) Where l represents the total span of the surveyed sub-region along the UAV's flight path. This is for rounding up; Set the side-up view overlap of the drone If the value is 0, then the distance between any two adjacent task points in the lateral direction is W. The number of task windows m in the lateral direction for the UAV is calculated as follows: (5) Where w is the total span of the surveyed sub-region in the direction of the UAV; Step 2.2.2: Adjust the relative position of the side-to-side task windows of the UAV, determine the minimum number of task windows covering the survey sub-area, and extract the center point of each task window at this time; Calculate the difference between the total span of the lateral window and the total lateral span of the surveyed sub-region. As shown in the formula below: (6) Will Divide into k equal parts, and translate them sequentially upwards from the side. Given the distance, generate k different combinations of task window positions, traverse each combination of task window relative positions, obtain the minimum number of task windows covering the mapping sub-region, and extract the center point coordinates of the UAV heading and side-mounted task windows.

2. The method for pattern decision-making and path planning in complex area UAV mapping according to claim 1, characterized in that: Step 1 includes the following steps: Step 1.1: Set different flight modes according to the terrain of the area to be surveyed; Step 1.2: Obtain the data required by the mapping mode decision algorithm based on altitude difference threshold, including the length and width of the area covered by a single UAV-captured mapping image and the acceptable altitude difference for the UAV. And the digital elevation model (DEM) data of the area to be surveyed, where each DEM data point represents the elevation value at that location; Step 1.3: Based on the maximum and minimum elevation values ​​of all DEM data points within the area to be surveyed, calculate the overall elevation difference within the area to be surveyed. As shown in the following formula: (1) in, and These are the elevations of the highest and lowest points within the area to be surveyed; Step 1.4: Based on the acceptable altitude difference for the drone The overall height difference of the area to be surveyed The area is divided into several altitude levels with equal altitude flight zones, and the altitude difference between two adjacent altitude levels does not exceed the acceptable altitude difference for the UAV. ; Step 1.5: Define a sliding window whose size matches the area covered by the mapping image captured by the UAV in a single shot. Use this sliding window to traverse the area to be mapped, and adjust the elevation difference within the current sliding window according to the acceptable elevation difference of the UAV. Decision on the flight mode of the drone within the current sliding window area; Step 1.6: For the sliding window region where the flight mode is determined, obtain the flight modes of its adjacent sliding window regions. Based on the principles of adjacency and feature consistency, merge adjacent sliding window regions with the same flight mode, record the boundary point coordinates of the merged region, and generate a mapping sub-region in the area to be mapped that uses different flight modes.

3. The method for pattern decision-making and path planning in complex area UAV mapping according to claim 2, characterized in that: The specific method for step 1.5 is as follows: Define the elevation difference within the current sliding window. for: (2) Among them, h max and h min These are the altitudes of the highest and lowest points within the current sliding window, respectively. Based on the current elevation difference within the sliding window Decide on the flight mode of the drone within the current sliding window area. If the window covers the area, the flight mode will be set to ground-hugging flight mode. Based on the highest and lowest altitudes of the window, the flight mode is set to the constant altitude flight mode for each altitude level; when the drone is in ground-hugging flight mode, the corresponding altitude control system is used to control the flight altitude.

4. The method for pattern decision-making and path planning in complex area UAV mapping according to claim 1, characterized in that: Step 3 includes the following steps: Step 3.1: Construct a mathematical model for coverage path planning of multiple survey sub-regions using a single UAV, and set constraints based on the surveying task; Step 3.2: Optimize the objective function based on VNS variable neighborhood search-genetic algorithm to obtain the total path that passes through all task points in the area to be mapped and has the shortest flight distance.

5. The method for pattern decision-making and path planning in complex area UAV mapping according to claim 4, characterized in that: The specific method for step 3.1 is as follows: After path planning for all survey sub-regions, multiple candidate entry / exit points are generated for each survey sub-region, and a set of candidate entry / exit points for all survey sub-regions is set. ,in Let N be the number of candidate entry / exit points; set the path set. Let represent the set of all paths, where The distance from the i-th candidate entry / exit point to the j-th candidate entry / exit point; generate a point group for all candidate entry / exit points belonging to the same mapping sub-region. If there are a total of M mapping sub-regions, then generate M point groups. To ensure that the UAV maps all sub-regions without repeatedly mapping any particular sub-region, the following objective function model is established with the shortest distance required for the UAV to complete the mapping task of all sub-regions as the objective: (7) Among them, D m Let be the distance the UAV needs to fly to perform a mapping task on the m-th mapping sub-region. This represents the set of distances that a drone needs to fly to perform a mapping task over a sub-region. The total flight distance required for the UAV to connect all mapping sub-areas. It is a 0 / 1 variable. This indicates that the drone has arrived at the j-th entry / exit point from the i-th entry / exit point; otherwise... ; This is the sum of the distances the UAV needs to fly to perform mapping tasks in each of the various mapping sub-regions, calculated by the distances between the UAVs in each mapping sub-region. Minimize, thus optimizing the objective function.

6. The method for pattern decision-making and path planning in complex area UAV mapping according to claim 4, characterized in that: The specific method for step 3.2 is as follows: (1) Initialize the genetic algorithm. For the entry / exit points of the fixed path in each mapping sub-region and the task points outside the mapping sub-region, arbitrarily select a fixed road segment of the mapping sub-region as the starting road segment. Starting from an entry / exit point of the starting road segment, use a greedy algorithm to select the point that is closest to the current entry / exit point and has not yet reached the task point as the next point to be visited, and obtain S candidate paths as S initial individuals of the population. (2) Perform crossover and mutation operations on the remaining task points of the S initial individuals of the population, excluding those with fixed paths, to generate new offspring individuals; (3) Calculate the fitness of each new offspring individual and select an optimal path solution set from the current population using the "roulette wheel" algorithm; For the UAV path planning problem, a fitness function is set to evaluate the performance of each offspring individual. The fitness function is shown in the following formula: (8) (4) Set the number of iterations of the VNS algorithm and set its initial number of iterations to zero. Use the 2-opt swap strategy to adjust the path order, that is, select any two edges from the path and swap their connection to form a new path, thereby reducing the path length or optimizing other objective function values. Each time, select two points from the remaining task points excluding the fixed path and swap their order in the original path, and calculate the fitness of the new path solution set. If the optimal path solution is obtained before the number of iterations, proceed to step (5); otherwise, repeat this step until the preset number of VNS algorithm iterations is reached. (5) Update the optimal path and proceed to the next iteration of the genetic algorithm; (6) Repeat steps (2) to (5) until the preset number of iterations of the genetic algorithm is reached. During the iteration process, individuals in the population will continuously evolve and gradually approach the optimal path scheme. (7) Change the order of the fixed path candidate entry / exit points in turn, and repeat steps (1) to (6) to obtain multiple path planning schemes for all entry / exit points. Select the path planning scheme with the highest fitness as the total path that passes through all task points in the area to be mapped and has the shortest flight distance.

7. The method for pattern decision-making and path planning in complex area UAV mapping according to claim 3, characterized in that: The ground-flying mode corresponds to an altitude control system, which specifically includes: a lidar, a data processing module, and a flight control module. LiDAR acquires the real-time altitude data of a drone by emitting a laser beam toward the surface of the area to be mapped and receiving the signal reflected from the surface of the area at a certain sampling frequency. The drone's altitude data is then transmitted to the data processing module. The data processing module receives real-time altitude data of the UAV sent by the lidar, filters and denoises the data to improve its accuracy, and generates corresponding control commands based on the real-time altitude data of the UAV and transmits them to the flight control module. The specific method by which the data processing module generates the corresponding control commands is as follows: Calculate the actual height of the measurement drone relative to the surface of the area to be mapped. : (9) Where c is the speed of light, and t is the round-trip time of the laser beam. The wavelength of the laser is used; the distance between the drone and the surface of the area to be mapped is compared with the preset desired altitude to calculate the current altitude deviation. (10) Based on the altitude deviation, the data processing module uses a PID control algorithm to control the drone's flight altitude. The PID control algorithm is shown in the following formula: (11) in, This is the proportional control section. For proportional gain; The integral control section is used to eliminate steady-state error. It is integral gain; This is the differential control section, used to predict and correct the changing trend of altitude error. This is the differential gain; This is the final control signal used to adjust the drone's altitude. If the distance between the drone and the ground surface of the survey area is higher than the desired altitude, a command to decrease altitude is generated; if the distance between the drone and the ground surface of the survey area is lower than the desired altitude, a command to increase altitude is generated. The flight control module receives control commands from the data processing module and adjusts the UAV's flight attitude and speed to achieve precise control of flight altitude.

Images