Multi-target unmanned aerial vehicle path planning method based on improved A satellite and artificial potential field method

Abstract

The invention relates to a multi-target unmanned aerial vehicle path planning method based on an improved satellite A and an artificial potential field method, which comprises the following steps: firstly, determining a starting point and all target points of an unmanned aerial vehicle, then modeling a flight environment of the unmanned aerial vehicle by adopting a grid method to generate a grid map, and then determining an access sequence of all target points of the unmanned aerial vehicle by utilizing a confinement artificial potential field algorithm; an improved A star algorithm is adopted to plan paths between the starting point and the first target point of the unmanned aerial vehicle and between any two adjacent target points to generate a plurality of sub-paths, and finally all the sub-paths are spliced in sequence to obtain a final planned total path of the unmanned aerial vehicle. According to the multi-target unmanned aerial vehicle path planning method based on the improved A star and the artificial potential field method, the improved A star algorithm and the confinement artificial potential field method are fused, and the calculation efficiency and the search efficiency of path search are remarkably improved while the safety and the optimality of the path are ensured.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, and in particular to a multi-objective UAV path planning method based on an improved A-satellite and artificial potential field method. Background Technology

[0002] With the rapid development of drone technology, drones are being applied in increasingly wider fields, including transportation, disaster relief, environmental monitoring, and agriculture. In these applications, achieving efficient and safe multi-objective path planning has become one of the core challenges for autonomous drone flight. Multi-objective path planning aims to enable drones to simultaneously visit multiple target points in complex environments, while ensuring optimal path selection and safety.

[0003] In multi-objective path planning, factors such as the order of visiting multiple target points, path turns, and obstacle avoidance typically need to be considered. Unlike single-objective path planning, multi-objective path planning requires optimizing multiple objectives simultaneously, and the relationships between these objectives may conflict. Therefore, multi-objective path planning algorithms must not only ensure that the UAV can traverse the predetermined target points, but also consider the order of flight of the target points and ensure that the planned path does not overlap with obstacles, thus ensuring the safety and optimality of the path.

[0004] In existing technologies, A-star algorithm and artificial potential field (APF) are two widely used multi-objective path planning methods. A-star is a classic heuristic search algorithm widely applied to path planning problems. By combining the characteristics of breadth-first search and greedy best-first search, A-star can find the optimal path from the starting point to the destination in a given environment, demonstrating strong path search capabilities. However, A-star requires considering all possible paths during the search process, resulting in a large computational load, especially when there are many target points or complex obstacles. Furthermore, it can only plan paths from the starting point to a specific target point and cannot handle cases with multiple target points. Artificial Potential Field (APF) is an algorithm that simulates a physical potential field to guide path planning. APF treats target points as sources of attraction and obstacles as sources of repulsion, generating a force field during path planning where attraction and repulsion occur simultaneously. The UAV moves according to the direction of this force field, thus achieving obstacle avoidance and target guidance. The advantages of APF are its computational simplicity, real-time performance, and suitability for path planning in dynamic environments. However, the APF algorithm is prone to getting trapped in local optima, especially when there are many or complex obstacles. Furthermore, it has weak control over global path planning and struggles to optimize across multiple objectives. Therefore, when faced with multi-objective path planning, the APF algorithm often fails to effectively handle the access order and path optimization problems involving multiple objectives.

[0005] In summary, existing technologies, when using the most widely used A* algorithm and artificial potential field method for multi-objective path planning, especially when there are many target points or complex obstacles, face problems such as high computational complexity or easy getting trapped in local optima, resulting in reduced search efficiency. Summary of the Invention

[0006] Therefore, it is necessary to provide a multi-target UAV path planning method based on the improved A-satellite algorithm and the artificial potential field method to address the above-mentioned technical problems. This method integrates the improved A-satellite algorithm and the confined artificial potential field method, which significantly improves the computational efficiency and search efficiency of path search while ensuring path safety and optimality.

[0007] This invention provides a multi-objective UAV path planning method based on an improved A-satellite and artificial potential field method, comprising the following steps:

[0008] Determine the drone's starting point and all target points;

[0009] The raster method is used to model the flight environment of the UAV and generate a raster map;

[0010] The order in which all target points are visited by the UAV is determined using a confined artificial potential field algorithm.

[0011] An improved A-star algorithm is used to plan and generate multiple sub-paths between the UAV's starting point and the first target point, as well as between any two adjacent target points;

[0012] By sequentially piecing together all the sub-paths, the final planned total path for the drone is obtained;

[0013] The evaluation function of the improved A* algorithm is:

[0014] f = g + σ h h+σ θ f θ

[0015] In the formula, f represents the evaluation function value of reaching the target state node required by the task from the initial selected state node, g represents the actual cost function value of reaching a certain state in the environment where the UAV is located from the initial selected state node, and h represents the estimated cost function value of the search path from a certain state in the environment where the UAV is located to the target state node required by the task. θ σ represents the corner cost. h σ represents the weighting coefficients for estimating the cost function value. θ The weighting coefficient represents the cornering cost.

[0016] In one embodiment, the corner cost f in the evaluation function of the improved A* algorithm θ The process of determining is as follows:

[0017] Get the vectors of the current node and the previous node;

[0018] Calculate the cosine of the angle between the current node vector and the previous node vector;

[0019] Determine if the cosine of the angle between the current node vector and the previous node vector is less than the cosine of a preset angle threshold. If so, calculate the angle value of the turning angle and determine the turning cost as θ / π; otherwise, determine the turning cost as 0.

[0020] In one embodiment, the order in which the UAV visits all target points is determined using a confined artificial potential field algorithm, including the following steps:

[0021] Calculate the gravitational force exerted on the drone by each target point when the drone is at the starting point;

[0022] Determine the repulsive force exerted by obstacles on the drone within a range of M grid cells at the starting point;

[0023] The total potential field of each target point relative to the drone is obtained by superimposing the repulsive force on the drone at the starting point and the gravitational force on the drone from each target point.

[0024] Sort all target points relative to the total potential field of the UAV from smallest to largest;

[0025] The order in which all target points are accessed by the UAV is obtained by sorting the target points according to their order relative to the total potential field of the UAV.

[0026] In one embodiment, an improved A* algorithm is used to plan and generate multiple sub-paths between the UAV's starting point and the first target point, as well as between any two adjacent target points, including the following steps:

[0027] The initial selection state node of the UAV whose sub-path is currently to be planned is taken as the center point;

[0028] The evaluation function values ​​of the 26 next state nodes of the center point are compared, and the state node with the smallest evaluation function value is selected as the moving target of the UAV.

[0029] The process involves determining whether the drone's moving target is the target point corresponding to the current sub-path to be planned. If so, the initially selected state node and all state nodes with the smallest evaluation function values ​​are sorted to form the current sub-path to be planned. Otherwise, the state node with the smallest evaluation function value is used as the new center point, and the evaluation function values ​​of the 26 next state nodes of the calculated center point are compared.

[0030] In one embodiment, the method further includes sequentially outputting the final planned total path of the UAV in a raster index format or a continuous coordinate format.

[0031] The beneficial effects of this invention are as follows: In the process of planning each sub-path, this invention employs an improved A* algorithm. By adding weights to the actual cost and the heuristically estimated cost to the target in the evaluation function, and by adding corner constraints to the cost function, its search efficiency is improved, and paths with excessively large corners are avoided, thereby improving the efficiency and quality of the path planning algorithm. Furthermore, when using the confined artificial potential field algorithm to determine the access order of all target points for the UAV, this invention treats obstacles within a certain range as repulsive force sources, rather than all obstacles in the entire flight environment. This reduces the possibility of the artificial potential field algorithm getting trapped in local optima while effectively reducing computational complexity. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating a multi-objective UAV path planning method based on an improved A-satellite and artificial potential field method provided in an embodiment of the present invention.

[0033] Figure 2 This is a schematic diagram illustrating the definition of the horizontal turning angle of a drone provided in an embodiment of the present invention;

[0034] Figure 3 This is a schematic diagram illustrating the definition of the pitch angle of a UAV provided in an embodiment of the present invention;

[0035] Figure 4 This is a two-dimensional view of the multi-target point experimental results provided in this embodiment of the invention;

[0036] Figure 5 This is a two-dimensional view of the multi-target point experimental results provided in this embodiment of the invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0038] In one embodiment, such as Figure 1 As shown, Figure 1 This is one of the flowcharts illustrating a multi-target UAV path planning method based on an improved A-satellite and artificial potential field method provided in this invention. When applied to computer equipment, this method includes the following steps:

[0039] S101. Determine the starting point of the drone and all target points.

[0040] S102. Use the grid method to model the flight environment of the UAV and generate a grid map.

[0041] This invention employs a grid method to model the flight environment of unmanned aerial vehicles (UAVs) in order to construct a unified data structure that facilitates path planning calculations. The grid method is a common cell decomposition modeling method. Its basic principle is to divide the three-dimensional space into grid cells of equal volume to accurately describe environmental information, including the location, shape, and spatial distribution characteristics of obstacles.

[0042] In the map construction process, the starting point of the UAV and all target points are first determined, and a raster map is generated based on the three-dimensional spatial information. To mathematically represent the map information, this invention establishes a three-dimensional Cartesian coordinate system, where a point on the map is selected as the origin, and the x, y, and z axes are defined to represent the raster index values ​​of the map in the east-west, north-south, and vertical directions, respectively. Each raster cell corresponds to a fixed-volume spatial element used to store the cell's accessibility, obstacle information, and other necessary environmental parameters.

[0043] Compared to visual graph and topology graph methods, the grid method has the advantages of unified data structure, ease of storage and computation, making it particularly suitable for path planning tasks of UAVs. Therefore, this invention adopts the grid method as the basis for 3D environment modeling, providing support for subsequent path optimization and navigation calculations.

[0044] S103. Use the confined artificial potential field algorithm to determine the order in which all target points are visited by the UAV.

[0045] Specifically, the restricted artificial potential field algorithm is used to determine the order in which all target points are visited by the UAV, including the following steps:

[0046] S1031. Calculate the gravitational force exerted on the drone by each target point when the drone is at the starting point.

[0047] The gravitational potential energy exerted by each target point on the moving drone varies with the distance between the drone and the target point. The magnitude of the gravitational potential energy in the artificial potential field is proportional to the square of the distance between the drone and the target point; the greater the distance, the greater the gravitational potential energy generated by the target point. The gravitational potential field function is expressed as follows:

[0048]

[0049] In the formula, U att (q) represents the gravitational field generated by the target point on the UAV, ε is the gravitational potential gain function greater than 0, and ρ(q,q) goal ) 2 Let be the Euclidean distance between the current position of the drone and the target point. In three-dimensional space, the Euclidean distance calculation follows the formula:

[0050]

[0051] In the formula, q represents the current position of the drone.goal Indicates the target point. Represents the three-dimensional coordinates of the target point, (x q y q , z q () represents the three-dimensional coordinates of the current position of the drone.

[0052] Traditional artificial potential field methods consider all existing obstacles when constructing the repulsive potential field. In this constructed potential field, the farther the drone is from the obstacle, the smaller the repulsive force, and vice versa. However, in grid maps, this method requires calculating the repulsive force of all obstacles, consuming a large amount of computational resources and thus failing to guarantee the efficiency of the algorithm.

[0053] S1032. Determine the repulsive force of obstacles encountered by the UAV within a range of M grids at the starting point.

[0054] In this embodiment, the repulsive force exerted on the drone by obstacles within M grid cells of the drone is only considered in the grid map, which effectively reduces the consumption of computing resources. It should be noted that M can be specifically set according to the specific flight environment and path planning task.

[0055] If the distance between the obstacle and the drone exceeds the defined grid range, the repulsive force of the obstacle is not considered; that is, the repulsive force U of the obstacle... obs (q) = 0.

[0056] S1033. The total potential field of each target point relative to the drone is obtained by superimposing the repulsive force on the drone at the starting point and the gravitational force on the drone at each target point.

[0057] When the drone is currently located within a grid of M grids and there are m obstacles, each obstacle provides a repulsive force of U. obs Then the formula for calculating the total potential field U(q) of the target point relative to the UAV is:

[0058]

[0059] The above formula can be used to calculate the total potential field of each target point relative to the UAV.

[0060] S1034. Sort all target points relative to the total potential field of the UAV from smallest to largest.

[0061] S1035. Sort the target points according to the order of all target points relative to the total potential field of the UAV to obtain the order of all target point visits by the UAV.

[0062] The order in which all target points are visited by the UAV is used to guide the path search direction of the improved A* algorithm. This makes the improved A* algorithm more directional during the search, thereby improving its search efficiency.

[0063] S104. The improved A-star algorithm is used to plan and generate multiple sub-paths between the starting point of the UAV and the first target point, as well as between any two adjacent target points.

[0064] In this embodiment, the improved A* algorithm is used to plan and generate multiple sub-paths between the UAV's starting point and the first target point, as well as between any two adjacent target points. This includes the following steps:

[0065] S1041. Take the initial selection state node of the UAV with the current sub-path to be planned as the center point.

[0066] S1042. Compare the evaluation function values ​​of the 26 next state nodes of the center point, and select the state node with the smallest evaluation function value as the moving target of the UAV.

[0067] S1043. Determine whether the moving target of the UAV is the target point corresponding to the current sub-path to be planned. If so, sort the initially selected state nodes and all state nodes with the smallest evaluation function values ​​to form the current sub-path to be planned. Otherwise, take the state node with the smallest evaluation function value as the new center point and return the evaluation function values ​​of the 26 next state nodes of the center point for comparison.

[0068] S105. Sequentially connect all sub-paths to obtain the final planned total path for the UAV.

[0069] It should be noted that multiple sub-paths include sub-paths between the starting point and the first target point, sub-paths between the first target point and the second target point, sub-paths between the second target point and the third target point, etc.

[0070] The evaluation function of the improved A* algorithm is:

[0071] f = g + σ h h+σ θ f θ (4)

[0072] In the formula, f represents the evaluation function value of reaching the target state node required by the task from the initial selected state node, g represents the actual cost function value of reaching a certain state in the environment where the UAV is located from the initial selected state node, and h represents the estimated cost function value of the search path from a certain state in the environment where the UAV is located to the target state node required by the task. θ σ represents the corner cost. h σ represents the weighting coefficients for estimating the cost function value. θ The weighting coefficient represents the cornering cost.

[0073] This invention improves the search efficiency of the path planning algorithm and avoids planning paths with excessively large turns by adding weights to the actual cost and the heuristic estimated cost to the target in the evaluation function and adding corner constraints to the cost function.

[0074] In one embodiment, the turning angle of the UAV is divided into a horizontal yaw angle and a vertical pitch angle. The maximum allowable turning angle in the horizontal direction depends on the UAV's own maneuverability. If the planned path has an excessively large turning angle, the actual flight may be constrained by control performance, resulting in a situation where the UAV cannot fly according to the predetermined trajectory. Figure 2 As shown, α is defined as the first segment of the track, α i+1 Let θ be the angle between the two flight paths, and let θ be the angle between the two flight paths. max If the maximum turning angle supported by the drone is given, then the following formula should be satisfied:

[0075]

[0076] θ≤θ max (6);

[0077] like Figure 3 As shown, the analysis shows that the angle between the vectors is positively correlated with the difference in the directions of the two vectors. Therefore, the drone's turning angle limit can effectively constrain the drone's motion behavior.

[0078] The pitch angle is defined as the angle between the fuselage axis along the nose direction and the ground plane. The pitch angle is positive when the UAV is tilting up and negative when it is tilting down. Similar to equation (5), β represents the pitch angle and must satisfy equation (7).

[0079] β≤θ max (7);

[0080] This invention proposes a novel heuristic strategy by introducing the cornering cost f into the evaluation function. θ This is used to implement the angle constraint in the path planning algorithm. To ensure the effectiveness of the algorithm, this study focuses on f. θ A threshold-based calculation method was designed, which involves setting a reasonable angle threshold θ. t This invention effectively balances the straightness of the path and the smoothness of turns by introducing a significant penalty for large turns while maintaining optimal path length. Furthermore, to impose a heavier penalty on larger turns, this invention employs a linearly increasing penalty mechanism for angles exceeding a threshold. Specifically, in the evaluation function of the improved A* algorithm in this embodiment, the turn cost f... θ The process of determining is as follows:

[0081] (1) Obtain the vectors of the current node and the previous node. Initialize the corner cost f. θ .

[0082] (2) Calculate the cosine of the angle between the current node vector and the previous node vector.

[0083] Specifically, the cosine value of the angle between the two vectors, cosθ, is calculated according to equation (5). cost .

[0084] (3) Determine whether the cosine of the angle between the current node vector and the previous node vector is less than the cosine of the preset angle threshold. If so, calculate the angle value of the turning angle and determine the turning cost as θ / π. Otherwise, determine the turning cost as 0.

[0085] Specifically, through θ cost =cos -1 (cosθ cost Calculate the angle value of the rotation.

[0086] Including corner costs in the original evaluation function for sub-path node selection can be problematic. Since corner costs are determined by the vector between the previous node and the child node and are not affected by map size, in large map scenarios, the corner costs may be too small compared to the estimated and actual costs, insufficient to have a significant impact on the evaluation function, thus weakening the corner constraint. Therefore, this invention introduces a weighting coefficient in equation (4) to constrain large-angle turns by adjusting the corner costs according to the map size; and by introducing a weighting coefficient into the estimated cost, it balances the effectiveness and reliability of the algorithm.

[0087] This invention employs a segmented path search method for path planning, calculating each sub-path separately to ensure the optimality and feasibility of the overall path. The specific steps are as follows:

[0088] (1) Planning from the starting point to the first target point: The improved A* algorithm is used to calculate the path from the UAV starting point S to the first target point k1.

[0089] (2) Plan the sub-paths between each target point in sequence: For each pair of adjacent target points k i and k i+1 Path between Perform a path search until the last target point is reached.

[0090] (3) Calculation of the final path: Planning from the last target point k m Path to destination G

[0091] (4) Path merging: All sub-paths are sequentially concatenated to form a complete flight path. The merged path is shown below:

[0092]

[0093] In one embodiment, the method of the present invention further includes sequentially outputting the final planned total path of the UAV in a raster index format or a continuous coordinate format. The output path point set P consists of multiple three-dimensional coordinate points (x, y, z).

[0094] The methods for outputting path points include, but are not limited to:

[0095] (1) Raster index format: Path points are represented by raster numbers, which facilitates path tracing on discrete maps;

[0096] (2) Continuous coordinate format: The path is output in three-dimensional coordinates (x,y,z), which is suitable for the navigation and control system of UAVs;

[0097] (3) File storage and transmission: Supports storing path point data as standard format files, which facilitates data transmission and subsequent path tracking algorithms.

[0098] Furthermore, the path output process of the present invention can flexibly adjust the granularity of the output path according to different flight mission requirements. For example, the path point density can be reduced to reduce computational overhead, or the path point density can be increased to improve path tracking accuracy, so as to adapt to the needs of different application scenarios.

[0099] In one specific embodiment, simulation experiments were conducted to further verify the beneficial effects of the method of the present invention. Specifically, as shown in Table 1, this table contains basic information about the simulation starting point, target point, and map information used in the verification process. These simulation data were used to experimentally demonstrate the effectiveness and accuracy of the present invention.

[0100] Table 1. Simulation starting point, target point, and map information used in the verification process.

[0101] Starting point (150,100,10) Target Point 1 (300,300,400) Target Point 2 (250,200,60) Target Point 3 (400,450,30) Target point 4 (400,250,90) Number of obstacles 10 Map size 500*500*200

[0102] This experiment simulates urban building clusters, adding a degree of randomness to the distribution of buildings in the real physical world. A 500m x 500m test area and a 200m high airspace were used to verify the effectiveness and superiority of the invention. The methods for setting the start and end points in the experiment are shown in Table 1, and the simulation graphics are shown below. Figure 4 and Figure 5 As shown in Tables 2 and 3, the data is presented in Table 2.

[0103] like Figure 4 and Figure 5As shown, in the preset path planning scenario, the set starting position is marked as the starting point, the remaining task points are target points, and the last target point to be reached is marked as the endpoint. The line segments in the figure intuitively illustrate the flight path generated by this invention, with the path direction pointing sequentially from the starting point to the target point and finally to the endpoint. Both 2D and 3D perspectives demonstrate that the fusion algorithm proposed in this invention can effectively avoid obstacles in the environment, producing a coherent, collision-free path that conforms to the actual flight constraints of the aircraft, proving that the algorithm possesses excellent path generation and obstacle avoidance performance in complex 3D space.

[0104] Table 2 Simulation results of different A* algorithms

[0105] Traditional A* algorithm planning time / s 0.337 Improved A* algorithm planning time / s 0.166 Percentage increase in efficiency 50.74 Whether the planning was successful yes

[0106] Table 3 Simulation results of confined APF

[0107]

[0108] As can be seen from the experimental data in Table 2, this invention significantly reduces the search time by optimizing the traditional A* algorithm and introducing heuristic function weighting and corner constraint strategies without affecting path quality. The average search time of the A* algorithm without optimization was 0.337 seconds, while the search time after optimization decreased to 0.166 seconds, resulting in an overall search efficiency improvement of 50.74%. This fully verifies the effectiveness and feasibility of the proposed improved A* algorithm in improving computational efficiency.

[0109] Furthermore, as can be seen from the comparative data in Table 3, the performance of the traditional artificial potential field method is significantly improved after the field constraint process is applied. The original APF algorithm has an average time of 87.525 seconds in global multi-object search, while after introducing the field constraint mechanism, this time is reduced to 20.355 seconds, and the search efficiency is improved by 76.74%. This result shows that the field constraint artificial potential field method not only retains the effectiveness of the original algorithm in target guidance, but also significantly improves the running efficiency in complex environments, verifying the superiority of this invention in multi-object ranking and spatial search.

[0110] In summary, the multi-target UAV path planning method proposed in this invention, based on the improved A-satellite and artificial potential field method, significantly improves the efficiency and operability of path search while ensuring path safety and optimality. This method is effectively applicable to complex 3D environments with multiple target points and dense obstacles.

[0111] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A multi-objective unmanned aerial vehicle path planning method based on improved A-star and artificial potential field method, characterized in that, The method comprises the following steps: determining a starting point of the UAV and all target points; modeling a flight environment of the UAV by using a grid method to generate a grid map; determining a visiting sequence of all target points of the UAV by using a limited artificial potential field algorithm; planning a path between the starting point of the UAV and a first target point and between any two adjacent target points by using an improved A-star algorithm to generate a plurality of sub-paths; sequentially splicing all the sub-paths to obtain a final total path of the UAV; wherein an evaluation function of the improved A-star algorithm is: f = g + s h h + s θ f θ In the formula, f represents the evaluation function value from the initial selection state node of the task to the target state node of the task requirement, g represents the actual cost function value from the initial selection state node to a certain state of the environment where the UAV is located, h represents the estimated cost function value of the search path from a certain state of the environment where the UAV is located to the target state node of the task requirement, f θ represents the corner cost, σ h represents the weight coefficient of the estimated cost function value, σ θ represents the weight coefficient of the corner cost.

2. The multi-objective UAV path planning method based on improved A* and artificial potential field method according to claim 1, characterized in that, The improved A-star algorithm's evaluation function corner cost f θ The determination process is as follows: obtaining a vector of a current node and a previous node; calculating a cosine value of an angle between the vector of the current node and the vector of the previous node; determining whether the cosine value of the angle between the vector of the current node and the vector of the previous node is less than a cosine value of a preset angle threshold, if yes, calculating an angle value of a turning angle and determining a turning angle cost as θ / π, otherwise, determining the turning angle cost as 0.

3. The multi-objective UAV path planning method based on improved A-star and artificial potential field method according to claim 2, characterized in that, The method for determining the visiting sequence of all target points of the UAV by using the limited artificial potential field algorithm comprises the following steps: calculating a gravitational force of each target point on the UAV when the UAV is at the starting point; determining a repulsive force of an obstacle within an M-grid range on the UAV at the starting point; superimposing the repulsive force on the UAV at the starting point and the gravitational force of each target point on the UAV to obtain a total potential field of each target point relative to the UAV; sequentially arranging the total potential fields of all target points relative to the UAV from small to large; sequentially arranging the target points based on the sequence of the total potential fields of all target points relative to the UAV to obtain the visiting sequence of all target points of the UAV.

4. The multi-objective UAV path planning method based on improved A-star and artificial potential field method according to claim 3, characterized in that, The method for planning the path between the starting point of the UAV and the first target point and between any two adjacent target points by using the improved A-star algorithm to generate the plurality of sub-paths comprises the following steps: taking an initial selection state node of the UAV at a current sub-path to be planned as a center point; calculating evaluation function values of 26 next state nodes of the center point to compare and select a state node with the minimum evaluation function value as a moving target of the UAV; determining whether the moving target of the UAV is the target point corresponding to the current sub-path to be planned, if yes, sequentially arranging the initial selection state node and all state nodes with the minimum evaluation function value to form the current sub-path to be planned, otherwise, taking the state node with the minimum evaluation function value as a new center point and returning to the step of calculating the evaluation function values of the 26 next state nodes of the center point to compare.

5. The multi-objective UAV path planning method based on improved A* and artificial potential field method according to claim 4, characterized in that, The method further comprises sequentially outputting the final total path of the UAV in a grid index format or a continuous coordinate format.

Images