Unmanned aerial vehicle path planning method based on fusion A star potential field algorithm

Abstract

The invention provides an unmanned aerial vehicle path planning method based on an improved A star algorithm. The unmanned aerial vehicle path planning method comprises the following steps: S1, constructing a real-time three-dimensional unmanned aerial vehicle flight environment model containing static and dynamic obstacles by using sensor data; s2, determining a starting point and an ending point of the unmanned aerial vehicle, and initializing parameters and constraints of an improved A star algorithm and an improved artificial potential field algorithm; s3, improving an A star algorithm by improving a function weight coefficient and adjusting a dynamic path; s4, improving an artificial potential field algorithm, introducing an angle factor, and controlling the motion of the unmanned aerial vehicle through a combined potential field, thereby solving the oscillation phenomenon of the unmanned aerial vehicle; s5, generating a final optimized path in combination with the improved A star algorithm path and the artificial potential field control information; and S6, the unmanned aerial vehicle flies through the optimized path and the combined potential field control information. According to the method, the improved A star algorithm and the artificial potential field algorithm are fused, real-time environment information and multiple optimization means are utilized, and efficient, smooth, stable and optimal path planning of the unmanned aerial vehicle in a complex environment is achieved.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) path planning, and in particular to an unmanned aerial vehicle (UAV) path planning method based on a fusion A-star potential field algorithm. Background Art

[0002] UAV path planning is an important area in UAV technology research. UAV path planning is to use one or more fusion algorithms to find one or more safe, collision-free paths from the starting position to the end position under the constraints of the UAV's own performance and environment. Since the flight environment and flight scenes of UAVs change rapidly, it is necessary to select a path planning algorithm that is suitable for the current environment.

[0003] At present, path planning algorithms mainly include intelligent algorithms (ant colony algorithm, genetic algorithm, sparrow search algorithm, neural network, artificial potential field, etc.), sampling-based algorithms (fast random tree algorithm, probabilistic road map algorithm, etc.) and graph search-based algorithms (Dijkstra algorithm, A-star algorithm, etc.). Among them, the algorithm mechanism of intelligent algorithms and sampling algorithms leads to a high degree of uncertainty in the obtained solutions, and the optimality of the solutions cannot be guaranteed. Among the algorithms based on graph search, the algorithm is a simple and fast one, and the search has a targeted inspiration, so the A-star algorithm is selected as the basic algorithm for path planning. As the requirements for paths are gradually increasing, it is necessary to improve the A-star algorithm in terms of algorithm to meet the gradually increasing path requirements.

[0004] Therefore, an efficient and fast UAV path planning solution is urgently needed to solve the above problems. The present invention combines the improved A-star algorithm with the improved artificial potential field algorithm to solve the problems that UAVs are prone to fall into local optimality, unreachable targets, and cannot dynamically avoid obstacles. Summary of the invention

[0005] The purpose of the present invention is to provide a UAV path planning method based on the fusion A-star potential field algorithm to address the above-mentioned problems, so as to more efficiently and conveniently carry out path planning of the UAV during flight in a complex environment.

[0006] To this end, the above-mentioned purpose of the present invention is achieved through the following technical solutions:

[0007] Step S1: Using the sensor data of the drone to construct a drone flight environment model through a three-dimensional grid map, including static and dynamic obstacles, and updating the environment model in real time according to the sensor data;

[0008] Step S2: determine the starting point and end point of the UAV, initialize the parameters of the improved A-star algorithm and the improved artificial potential field algorithm, and set the motion constraints and movable directions of the UAV;

[0009] Step S3: Improve the A-star algorithm, including:

[0010] Step S31: Based on the improved function weight coefficient, the heuristic function and the cost function in the A-star algorithm are calculated using the Manhattan distance to generate a preliminary path point set;

[0011] Step S32: dynamically adjusting the planned path by deleting redundant nodes in the path point set generated by the A-star algorithm and smoothing the generated path using the B-spline method;

[0012] Step S4: improving the artificial potential field algorithm, designing new gravitational field and repulsive field functions by introducing angle factors, so that the artificial potential field has vector characteristics of magnitude and direction, so as to use the combined potential field to control the movement of the drone and suppress the phenomenon of the drone generating oscillation;

[0013] Step S5: Fusing the improved A-star algorithm path and the artificial potential field control information to generate the final optimized UAV flight path;

[0014] Step S6: Send the optimized path and resultant potential field control information to the UAV control system to perform actual flight, monitor the flight status in real time during the flight, and adjust the path and control strategy in a timely manner.

[0015] Preferably, in step S1, the surrounding environment is scanned in real time by using a lidar sensor carried on the drone to obtain high-precision three-dimensional point cloud data; a detailed three-dimensional grid map is constructed in real time using the three-dimensional point cloud data, and static obstacles and dynamic obstacles are accurately depicted based on the map, wherein static obstacles include buildings and terrain, and dynamic obstacles include other flying objects or movable targets.

[0016] Preferably, in step S31, the algorithm search efficiency is improved and the search time is reduced by improving the function weight coefficient; when the UAV is far away from the target position, the proportion of the actual cost value g(n) in the evaluation function f(n) is increased, and when the UAV is close to the target position, the proportion of the heuristic function h(n) in the evaluation function f(n) is increased to effectively reduce the time of global path planning. The evaluation function f(n) is as follows:

[0017]

[0018] Among them, ρ(x,x g ) represents the Euclidean distance from the current point to the target point, ρ(x s ,x g ) represents the Euclidean distance from the starting point to the target point, g(n) represents the actual cost, and h(n) represents the heuristic function.

[0019] Preferably, in step S31, the heuristic function and the cost function in the A-star algorithm are calculated using the Manhattan distance; where (x n ,y n , z n ) is the current position of the drone in the three-dimensional environment, and the calculation formula of the Manhattan distance is as follows:

[0020] g(n)=|(x n -x s )+(y n -y s) +(z n -z s )|

[0021] h(n)=|(x n -x g )+(y n -y g) +(z n -z g )|

[0022] Preferably, in step S32, a global secondary optimization is performed on the preliminary path point set generated in step S31 to delete redundant nodes. The specific optimization steps include:

[0023] Firstly, the path point set is secondarily differentiated to generate an inflection point set;

[0024] Starting from the starting position, select any three consecutive points a, b, and c in the inflection point set, and check whether there are obstacles on the line between a and c. If there are no obstacles, directly connect the inflection points a and c, and then check whether there are obstacles on the line between the inflection points a, c, and d in the inflection point set. If there are no obstacles, delete the redundant node c, otherwise keep the inflection point c;

[0025] And so on, redundant inflection points are gradually determined and deleted from the starting point to the target point to obtain a final path containing an optimized inflection point set.

[0026] Preferably, in step S32, the path is smoothed using a B-spline method, and the calculation formula of the B-spline curve is:

[0027]

[0028] Where n represents the number of control points, C i represents the coordinates of the control point, B i,p (u) represents the control point C i The spline basis function of .

[0029] Preferably, in step S4, the artificial potential field algorithm is improved, wherein the gravitational field generated by the target position on the drone is:

[0030] U att =k att [ρ(x,x g )] 2 (sinθ(x,x gh ))(sinθ(x,x g ),cosθ(x,x g ))

[0031] In the formula, k att represents the gravitational field gain coefficient, θ(x,x g ) represents the angle between the current position of the drone and the target position, θ(x,x gh ) represents the angle between the current position of the drone and the target position about the XZ axis;

[0032] The repulsive force field generated by obstacles on the drone is:

[0033]

[0034] In the formula, k att represents the repulsive field gain coefficient, ρ 0 Indicates the impact range of obstacles;

[0035] The above-mentioned gravitational force and repulsive force are superimposed to obtain the resultant potential field U of the drone in the motion space: total for:

[0036]

[0037] Preferably, step S5 comprises the following steps:

[0038] Step S51: Initialize the A-star algorithm parameters and set the key parameters in the artificial potential field algorithm, including the gravitational field gain coefficient k att , repulsive field gain coefficient k rep , and the obstacle influence range ρ 0 ;

[0039] Step S52: after obtaining the prior environmental information, perform global path planning through the improved A-star algorithm in step S3 and output a global path containing a set of inflection points, use the current point in the inflection point set as the starting point of the improved artificial potential field algorithm, and set the next inflection point in the inflection point set as the target point for local path planning; during the planning process, check whether the UAV enters the influence range of the dynamic obstacle, if so, proceed to step S53 for dynamic obstacle avoidance, if not, determine whether the UAV has reached the end point, if not, repeat step S52, if so, proceed to step S54;

[0040] Step S53: When the drone enters the dynamic obstacle influence range ρ 0When , the current position is set as the virtual starting point, the next inflection point of the current point in the inflection point set is set as the virtual target point, and a feasible path between the virtual starting point and the next inflection point in the inflection point set is planned by improving the artificial potential field algorithm. When the position of the drone coincides with the virtual target point, this step ends and returns to step S52;

[0041] Step S54: Output the final global path.

[0042] The beneficial effects of the present invention are:

[0043] 1. Ensure that the UAV can plan its path based on the latest environmental information during flight, improve the accuracy and reliability of planning, and significantly improve the calculation speed of path planning while ensuring path optimization, especially in complex environments.

[0044] 2. The improved A-star algorithm and artificial potential field algorithm are integrated, the Manhattan distance is used to optimize the heuristic function and cost function, and the path is dynamically adjusted and smoothed using B-spline, which significantly improves the efficiency and accuracy of path planning, reduces redundant nodes in the path, and ensures the smoothness of the path.

[0045] 3. The introduction of angle factors and the design of new gravitational field and repulsive field functions solve the oscillation phenomenon existing in the traditional artificial potential field algorithm, making the UAV have higher motion stability and more precise control during flight.

[0046] 4. By combining the heuristic search of the A-star algorithm and the local obstacle avoidance function of the potential field algorithm, an optimized and safe path can be obtained, ensuring that the drone chooses the optimal path during flight and reducing flight time and energy consumption.

[0047] 5. The final optimized flight path is generated by integrating the improved A-star algorithm path with the artificial potential field control information. Combining the advantages of the two algorithms, a path planning that is more adapted to the actual flight environment is achieved, which is suitable for various complex environments and mission requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below.

[0049] Figure 1 This is a flow chart of the overall method of the UAV path planning method based on the fusion A-star potential field algorithm of this application;

[0050] Figure 2 This is the smoothing path node process of step S32 in the UAV path planning method based on the fusion A-star potential field algorithm of this application. DETAILED DESCRIPTION

[0051] The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

[0052] like Figure 1 As shown, this embodiment is a UAV path planning method based on the fusion A-star potential field algorithm, which specifically includes the following steps:

[0053] Step S1: Use the sensor data of the drone to build a drone flight environment model through a three-dimensional grid map, including static and dynamic obstacles, and update the environment model in real time according to the sensor data.

[0054] Furthermore, by using a lidar sensor mounted on a drone to scan the surrounding environment in real time, high-precision three-dimensional point cloud data is obtained; a detailed three-dimensional grid map is constructed in real time using the three-dimensional point cloud data, and static obstacles and dynamic obstacles are accurately depicted based on the map, wherein static obstacles include buildings and terrain, and dynamic obstacles include other flying objects or movable targets.

[0055] Step S2: Determine the starting point and end point of the UAV, initialize the parameters of the improved A-star algorithm and the improved artificial potential field algorithm, and set the motion constraints and movable directions of the UAV.

[0056] Step S3: Improve the A-star algorithm, including:

[0057] Step S31: Based on the improved function weight coefficient, the heuristic function and the cost function in the A-star algorithm are calculated using the Manhattan distance to generate a preliminary path point set;

[0058] By improving the function weight coefficient, the algorithm search efficiency is improved and the search time is reduced; when the UAV is far away from the target position, the proportion of the actual cost value g(n) in the evaluation function f(n) is increased; when the UAV is close to the target position, the proportion of the heuristic function h(n) in the evaluation function f(n) is increased to effectively reduce the time of global path planning. The evaluation function f(n) is as follows:

[0059]

[0060] Among them, ρ(x,x g ) represents the Euclidean distance from the current point to the target point, ρ(x s ,x g) represents the Euclidean distance from the starting point to the target point, g(n) represents the actual cost value, and h(n) represents the heuristic function. As can be seen from the formula, when the drone moves from the starting position to the target position, the proportion of g(n) in f(n) gradually decreases, allowing the drone to quickly move away from the starting position; while when the drone moves from the starting position to the target position, the proportion of h(n) in f(n) gradually increases, allowing the drone to quickly approach the target position

[0061] The heuristic function and cost function in the A-star algorithm are calculated using Manhattan distance; where (x n ,y n , z n ) is the current position of the drone in the three-dimensional environment, and the calculation formula of the Manhattan distance is as follows:

[0062] g(n)=|(x n -x s )+(y n -y s) +(z n -z s )|

[0063] h(n)=|(x n -x g )+(y n -y g) +(z n -z g )|

[0064] Step S32: dynamically adjusting the planned path by deleting redundant nodes in the path point set generated by the A-star algorithm and smoothing the generated path using the B-spline method;

[0065] Due to the different weights of the heuristic function and the cost function in the adaptive variable weight algorithm, the adaptive variable weight A-star algorithm generates more redundant nodes than the A-star algorithm. A global secondary optimization is performed on the preliminary path point set generated according to step S31 to delete redundant nodes. The specific optimization steps include:

[0066] Firstly, the path point set is secondarily differentiated to generate an inflection point set;

[0067] Starting from the starting position, select any three consecutive points a, b, and c in the inflection point set, and check whether there are obstacles on the line between a and c. If there are no obstacles, directly connect the inflection points a and c, and then check whether there are obstacles on the line between the inflection points a, c, and d in the inflection point set. If there are no obstacles, delete the redundant node c, otherwise keep the inflection point c;

[0068] And so on, redundant inflection points are gradually determined and deleted from the starting point to the target point to obtain a final path containing an optimized inflection point set.

[0069] The B-spline method is used to smooth the path. The calculation formula of the B-spline curve is:

[0070]

[0071] Where n represents the number of control points, C i represents the coordinates of the control point, B i,p (u) represents the control point C i The spline basis function of .

[0072] Figure 2 To smooth the path node process, Figure 2 It can be seen that the four points C i , C i+1 , C i+2 and C i+3 Smoothing inflection point n i , adjacent coordinates P ni-1 , P ni and P ni+1 is the center point of three adjacent key inflection points, and the following formula is used to iteratively solve it:

[0073]

[0074] Where u is the knots node, its number is N, usually determined by the order of the spline function and the number of control points n:

[0075] N=d+n

[0076] In order to meet the UAV dynamic constraints, the spline order is selected as 3 and the number of control points is 4. Therefore, the control point set is

[0077] Finally, through the control point C i Corresponding spline basis function B i,d (u) The B-Spline can be obtained as:

[0078]

[0079] Among them, [B i,3 (u)B i+1,3 (u)B i+2,3 (u)B i+3,3 (u)] T It is a set of control points. After connecting the path set, we get the complete path planning process.

[0080] Step S4: Improve the artificial potential field algorithm by introducing angle factors to design new gravitational field and repulsive field functions so that the artificial potential field has vector characteristics of magnitude and direction, so as to utilize the combined potential field to control the movement of the drone and suppress the oscillation of the drone.

[0081] The artificial potential field algorithm is improved, where the gravitational field generated by the target position on the drone is:

[0082] U att =k att [ρ(x,x g )] 2 (sinθ(x,x gh ))(sinθ(x,x g ),cosθ(x,x g ))

[0083] In the formula, k att represents the gravitational field gain coefficient, θ(x,x g ) represents the angle between the current position of the drone and the target position, θ(x,x gh ) represents the angle between the current position of the drone and the target position about the XZ axis;

[0084] The repulsive force field generated by obstacles on the drone is:

[0085]

[0086] In the formula, k att represents the repulsive field gain coefficient, ρ 0 Indicates the impact range of obstacles;

[0087] The above-mentioned gravitational force and repulsive force are superimposed to obtain the resultant potential field U of the drone in the motion space: total for:

[0088]

[0089] Step S5: Fuse the improved A-star algorithm path and the artificial potential field control information to generate the final optimized UAV flight path. The specific steps are as follows:

[0090] Step S51: Initialize the A-star algorithm parameters and set the key parameters in the artificial potential field algorithm, including the gravitational field gain coefficient k att , repulsive field gain coefficient k rep , and the obstacle influence range ρ 0 ;

[0091] Step S52: after obtaining the prior environmental information, perform global path planning through the improved A-star algorithm in step S3 and output a global path containing a set of inflection points, use the current point in the inflection point set as the starting point of the improved artificial potential field algorithm, and set the next inflection point in the inflection point set as the target point for local path planning; during the planning process, check whether the UAV enters the influence range of the dynamic obstacle, if so, proceed to step S53 for dynamic obstacle avoidance, if not, determine whether the UAV has reached the end point, if not, repeat step S52, if so, proceed to step S54;

[0092] Step S53: When the drone enters the dynamic obstacle influence range ρ 0 When , the current position is set as the virtual starting point, the next inflection point of the current point in the inflection point set is set as the virtual target point, and a feasible path between the virtual starting point and the next inflection point in the inflection point set is planned by improving the artificial potential field algorithm. When the position of the drone coincides with the virtual target point, this step ends and returns to step S52;

[0093] Step S54: Output the final global path.

[0094] Step S6: Send the optimized path and resultant potential field control information to the UAV control system to perform actual flight, monitor the flight status in real time during the flight, and adjust the path and control strategy in a timely manner.

[0095] The present application discloses a method for UAV path planning based on the fusion A potential field algorithm, which can realize efficient and accurate path planning and dynamic obstacle avoidance functions. Through the improvement and fusion of the A algorithm and the artificial potential field algorithm, the method can quickly calculate the optimal path in a complex environment, reduce redundant paths, and effectively avoid static and dynamic obstacles, ensuring the safety and stability of the UAV during flight. In addition, real-time environmental modeling and high-precision scanning of the lidar sensor further improve the accuracy and adaptability of path planning, allowing the UAV to fly freely in a changing environment.

[0096] This system can be widely used in various mission scenarios, including but not limited to drone aerial photography, logistics distribution, environmental monitoring, search and rescue, etc. It can provide users with reliable and efficient drone path planning solutions, improve the success rate and efficiency of mission execution, and reduce the risks and losses caused by inaccurate path planning. In complex environments, the application of this system can significantly improve the autonomous flight capability and mission execution capability of drones, enabling them to perform well in various practical applications.

[0097] The above embodiments are merely illustrative of the concept and efficacy of the present invention, and are not intended to limit the present invention. Any person skilled in the art may modify and alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be as set forth in the claims.

Claims

1. A UAV path planning method based on the fusion A-star potential field algorithm, characterized in that: The following steps are involved: Step S1: Using the sensor data of the drone to construct a drone flight environment model through a three-dimensional grid map, including static and dynamic obstacles, and updating the environment model in real time according to the sensor data; Step S2: determine the starting point and end point of the UAV, initialize the parameters of the improved A-star algorithm and the improved artificial potential field algorithm, and set the motion constraints and movable directions of the UAV; Step S3: Improve the A-star algorithm, including: Step S31: Based on the improved function weight coefficient, the heuristic function and the cost function in the A-star algorithm are calculated using the Manhattan distance to generate a preliminary path point set; Step S32: dynamically adjusting the planned path by deleting redundant nodes in the path point set generated by the A-star algorithm and smoothing the generated path using the B-spline method; Step S4: improving the artificial potential field algorithm, designing new gravitational field and repulsive field functions by introducing angle factors, so that the artificial potential field has vector characteristics of magnitude and direction, so as to use the combined potential field to control the movement of the drone and suppress the phenomenon of the drone generating oscillation; Step S5: Fusing the improved A-star algorithm path and the artificial potential field control information to generate the final optimized UAV flight path; Step S6: Send the optimized path and resultant potential field control information to the UAV control system to perform actual flight, monitor the flight status in real time during the flight, and adjust the path and control strategy in a timely manner.

2. The method for UAV path planning based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: In step S1, the surrounding environment is scanned in real time by using a lidar sensor mounted on the drone to obtain high-precision three-dimensional point cloud data; a detailed three-dimensional grid map is constructed in real time using the three-dimensional point cloud data, and static obstacles and dynamic obstacles are accurately depicted based on the map, wherein static obstacles include buildings and terrain, and dynamic obstacles include other flying objects or movable targets.

3. The UAV path planning method based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: In step S31, the algorithm search efficiency is improved and the search time is reduced by improving the function weight coefficient; when the UAV is far away from the target position, the proportion of the actual cost value g(n) in the evaluation function f(n) is increased, and when the UAV is close to the target position, the proportion of the heuristic function h(n) in the evaluation function f(n) is increased to effectively reduce the time of global path planning. The evaluation function f(n) is as follows: Among them, ρ(x,x g ) represents the Euclidean distance from the current point to the target point, ρ(x s ,x g ) represents the Euclidean distance from the starting point to the target point, g(n) represents the actual cost, and h(n) represents the heuristic function.

4. The method for UAV path planning based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: In step S31, the heuristic function and cost function in the A-star algorithm are calculated using the Manhattan distance; where (x n ,y n , z n ) is the current position of the drone in the three-dimensional environment, and the calculation formula of the Manhattan distance is as follows: g(n)=|(x n -x s )+(y n -y s) +(z n -z s )| h(n)=|(x n -x g )+(y n -y g) +(z n -z g )| 5. The method for UAV path planning based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: In step S32, a global secondary optimization is performed on the preliminary path point set generated in step S31 to delete redundant nodes. The specific optimization steps include: Firstly, the path point set is secondarily derived to generate an inflection point set; Starting from the starting position, select any three consecutive points a, b, and c in the inflection point set, and check whether there are obstacles on the line between a and c. If there are no obstacles, directly connect the inflection points a and c, and then check whether there are obstacles on the line between the inflection points a, c, and d in the inflection point set. If there are no obstacles, delete the redundant node c, otherwise keep the inflection point c; And so on, redundant inflection points are gradually determined and deleted from the starting point to the target point to obtain a final path containing an optimized inflection point set.

6. The method for UAV path planning based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: In step S32, the path is smoothed using the B-spline method, and the calculation formula of the B-spline curve is: Where n represents the number of control points, C i represents the coordinates of the control point, B i,p (u) represents the control point C i The spline basis function of .

7. The method for UAV path planning based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: In step S4, the artificial potential field algorithm is improved, where the gravitational field generated by the target position on the drone is: U att =k att [ρ(x,x g )] 2 (sinθ(x,x gh ))(sinθ(x,x g ),cosθ(x,z g )) In the formula, k att represents the gravitational field gain coefficient, θ(x,x g ) represents the angle between the current position of the drone and the target position, θ(x,x gh ) represents the angle between the current position of the drone and the target position about the XZ axis; The repulsive force field generated by obstacles on the drone is: In the formula, k att represents the repulsive field gain coefficient, ρ0 represents the obstacle influence range; The above-mentioned gravitational force and repulsive force are superimposed to obtain the resultant potential field U of the drone in the motion space: total for:

8. The method for UAV path planning based on the fusion A-star potential field algorithm according to claim 1 is characterized in that: Step S5 includes the following steps: Step S51: Initialize the A-star algorithm parameters and set the key parameters in the artificial potential field algorithm, including the gravitational field gain coefficient k att , repulsive field gain coefficient k rep , and the obstacle influence range ρ0; Step S52: after obtaining the prior environmental information, perform global path planning through the improved A-star algorithm in step S3 and output a global path containing a set of inflection points, use the current point in the inflection point set as the starting point of the improved artificial potential field algorithm, and set the next inflection point in the inflection point set as the target point for local path planning; during the planning process, check whether the UAV enters the influence range of the dynamic obstacle, if so, proceed to step S53 for dynamic obstacle avoidance, if not, determine whether the UAV has reached the end point, if not, repeat step S52, if so, proceed to step S54; Step S53: When the drone enters the dynamic obstacle influence range ρ0, the current position is set as the virtual starting point, the next inflection point of the current point in the inflection point set is set as the virtual target point, and a feasible path from the virtual starting point to the next inflection point in the inflection point set is planned by using the improved artificial potential field algorithm. When the drone position coincides with the virtual target point, this step is terminated and the process returns to step S52; Step S54: Output the final global path.

Images