Unmanned aerial vehicle variable step size path planning method based on environment complexity
By using the variable step size RRT algorithm, which combines environmental complexity and ellipsoidal sliding window to dynamically adjust the step size, the problem of low efficiency and non-smooth paths in complex environments of traditional path planning algorithms is solved, and a more optimized path planning effect is achieved.
Patent Information
- Application Number
- CN202511712590.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-04-10
AI Technical Summary
Traditional path planning algorithms have poor environmental adaptability in complex obstacle environments, low path planning efficiency, uneven generated paths, and high distance costs, making it difficult to ensure stable flight of UAVs.
The variable step size RRT algorithm is adopted. By introducing the concept of environmental complexity, the genetic algorithm is used to optimize the path planning performance parameters. The step size is dynamically adjusted by combining the ellipsoidal sliding window to construct local environmental perception capability and realize the dynamic adaptability of path planning.
The generated paths are smoother, shorter, more efficient, and consume less energy, thus improving the stability and safety of drone flight.
Smart Images
Figure CN121829532A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of unmanned aerial vehicle autonomous navigation, and particularly relates to an unmanned aerial vehicle variable step path planning method based on environment complexity. BACKGROUND
[0002] Unmanned aerial vehicle technology plays an important role in many industries such as aerial photography, inspection and disaster relief. With the continuous development of unmanned aerial vehicle technology, path planning has become a hot research topic today. Path planning is the core theory of unmanned aerial vehicle autonomous navigation, which mainly studies how to quickly establish a safe and smooth path between the starting point and the ending point in a specific environment. Commonly used path planning algorithms include Rapidly-exploring Random Tree (RRT) algorithm, A-star algorithm, particle swarm algorithm, annealing algorithm and artificial potential field method.
[0003] Traditional path planning algorithms have too simple theoretical ideas, so there are some common problems, such as poor environmental adaptability, easy to enter local optimum in complex obstacle environment; low search efficiency, the generated path is not smooth enough, and it is difficult to ensure the smooth flight of the unmanned aerial vehicle; the path is not the shortest, which increases the energy consumption of the power equipment, etc. RRT algorithm is an incremental global path planning algorithm, which is used to solve the path planning problem with algebraic constraints (obstacle constraints). It has the characteristics of no model, high search efficiency, fast convergence speed, and is suitable for various environments, and is widely used. With further research, scholars have proposed various improved RRT algorithms, such as Bi-RRT, which starts from both the starting point and the ending point for bidirectional path exploration, improving the convergence speed of the algorithm; Dynamic-RRT, which reduces and combines path nodes to remove invalid nodes and reduce the path cost.
[0004] Due to the strong randomness of the RRT algorithm, it also brings some disadvantages to path planning, such as poor environmental adaptability, low efficiency and performance of path planning when encountering complex obstacle regions; the generated path is relatively tortuous, and the path cost is large. SUMMARY
[0005] Therefore, the present application discloses an unmanned aerial vehicle variable step path planning method based on environment complexity.
[0006] The present application adopts the following technical solutions: An unmanned aerial vehicle variable step path planning method based on environment complexity, the method comprising a variable step RRT algorithm, the variable step RRT algorithm comprising: S1, quantizing the environment map into a three-dimensional matrix, adding the path starting point to the random tree path node list, and determining the initial step length; S2, using the RRT algorithm, generating a new node one from the path starting point when the line between the random sampling point and the path starting point does not pass through the obstacle, and adding the new node one to the path node list; S3, constructing an ellipsoid sliding window with the new node one as the center point, and calculating the local environment complexity of the window region; S4, taking the total environment complexity as a demarcation point, when the local environment complexity is less than or equal to the total environment complexity, the step length is not changed; when the local environment complexity is greater than the total environment complexity, the step length is recalculated according to the local environment complexity; S5, marking the new node one as a new starting point, generating a new node two from the new starting point with the unchanged step length or the recalculated step length, and adding the new node two to the random tree list; S6, repeating the above operations until the distance between the generated new node and the end point is less than the step length.
[0007] Further, the step S1 of determining the initial step length comprises: , wherein p0 is the initial step length, and wherein c0 is the total environment complexity.
[0008] Further, the step S4 of recalculating the determined step length comprises: , wherein p is the optimal step length, and wherein c is the local environment complexity.
[0009] Further, the path planning method further comprises: three-dimensional environment modeling, description of path planning performance parameters, exploration of the relationship between the optimal step length and the environment complexity, and construction of the local environment complexity and the ellipsoid sliding window.
[0010] Further, the three-dimensional environment modeling comprises: gridding the three-dimensional environment and representing the obstacle information with a matrix.
[0011] Further, the description of the path planning performance parameters comprises: after the path planning is completed, introducing the distance, time, and smoothness parameters to evaluate the performance of the planned path; optimizing the parameters to establish the relationship between the optimal step length and the environment complexity in different environments.
[0012] Further, the exploration of the relationship between the optimal step length and the environment complexity comprises: constructing the elementary function: , wherein c is the environment complexity, p is the optimal step length, and α, β, and γ are undetermined coefficients.
[0013] Further, the exploration of the relationship between the optimal step length and the environment complexity further comprises: the functional relationship between the environment complexity and the optimal step length: .
[0014] Further, the local environment complexity and the construction of the ellipsoid sliding window include: adopting the ellipsoid sliding window to roll and calculate the local environment complexity of the current position in the path search, and constructing the surrounding environment area of the current node in the random tree growth process.
[0015] Further, the position and size of the ellipsoid sliding window change with the complexity of the local environment where the unmanned aerial vehicle flies.
[0016] Advantages of the present application: The present application introduces the concept of environment complexity, optimizes the performance parameters of path planning based on the traditional RRT algorithm using genetic algorithm, and establishes a mathematical model of the optimal step length and environment complexity. Further, the sliding window method is applied to construct the ellipsoid sliding window to detect the local environment in path planning, thereby dynamically changing the step length, and a variable step length RRT algorithm is proposed. The algorithm can change the step length in real time according to the local environment in path planning, can better adapt to the environment, and is optimized in the three performance indicators of distance, smoothness and time, improves the efficiency of path planning, and the generated path is smoother, the path has no dead angle, and the distance is shorter, and the path is closer to the global optimum. The planned path has the characteristics of less time, high efficiency and small total cost, and can effectively ensure the stability of the unmanned aerial vehicle flight. BRIEF DESCRIPTION OF DRAWINGS
[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.
[0018] Figure 1 is a schematic diagram of different complexity environments of the present application, wherein (a) is a relatively simple environment, and (b) is a relatively complex environment; Figure 2 is a schematic diagram of the environment and the environment matrix of the present application, wherein (a) is an environment map, and (b) is an environment matrix; Figure 3 is a schematic diagram of the path generated after path planning of the present application; Figure 4 is a schematic diagram of the path smoothness of the present application; Figure 5 is a schematic diagram of the relationship between the optimal step length and the environment complexity of the present application; Figure 6 is a schematic diagram of the sliding window of the present application; Figure 7 is a schematic diagram of the environment map for simulation experiment of the present application; Figure 8Fig. 1 is a schematic diagram of simulation experiment of the RRT algorithm of the present application, wherein (a) is a schematic diagram of experimental results after traditional RRT path planning, and (b) is a schematic diagram of experimental results after variable step RRT path planning; Figure 9 Fig. 2 is a schematic diagram of simulation experiment data and comparison of the RRT algorithm of the present application. DETAILED DESCRIPTION
[0019] In order to better understand the technical solutions of the present application, the embodiments of the present application are described in detail below with reference to the drawings.
[0020] It should be clear that the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0021] Embodiment 1 A variable step path planning method for unmanned aerial vehicles based on environmental complexity, the method comprising a variable step RRT algorithm, the variable step RRT algorithm comprising: S1, quantizing an environmental map into a three-dimensional matrix, adding a path starting point to a random tree path node list, and determining an initial step length; S2, using an RRT algorithm, starting from the path starting point, when a line connecting a randomly sampled point and the path starting point does not pass through an obstacle, generating a new node one with the initial step length, and adding the new node one to the path node list; S3, constructing an ellipsoidal sliding window with the new node one as a center point, and calculating a local environmental complexity of a region within the window; S4, taking an overall environmental complexity as a demarcation point, when the local environmental complexity is less than or equal to the overall environmental complexity, not changing the step length, and when the local environmental complexity is greater than the overall environmental complexity, recalculating and determining a step length according to the local environmental complexity; S5, marking the new node one as a new starting point, starting from the new starting point, generating a new node two with the unchanged step length or the recalculated and determined step length, and adding the new node two to the random tree list; S6, repeating the above operations until the distance between the generated new node and an end point is less than the step length.
[0022] Further, the step S1 of determining the initial step length comprises: , wherein p0 is the initial step length, and wherein c0 is the overall environmental complexity.
[0023] Further, the step S4 of recalculating and determining the step length comprises: , wherein p is the optimal step length, and wherein c is the local environmental complexity.
[0024] Furthermore, the path planning method also includes: three-dimensional environment modeling, description of path planning performance parameters, exploration of the relationship between optimal step size and environment complexity, and construction of local environment complexity and ellipsoidal sliding window.
[0025] Furthermore, the three-dimensional environment modeling includes: rasterizing the three-dimensional environment map and representing obstacle information using a matrix.
[0026] Furthermore, the description of the path planning performance parameters includes: after the path planning is completed, introducing distance, time, and smoothness parameters to evaluate the performance of the planned path; optimizing the parameters and establishing the relationship between the optimal step size and environmental complexity under different environments.
[0027] Furthermore, the exploration of the relationship between the optimal step size and environmental complexity includes: constructing an elementary function: In the formula, c is the environmental complexity, p is the optimal step size, and α, β, and γ are undetermined coefficients.
[0028] Furthermore, the exploration of the relationship between the optimal step size and environmental complexity also includes: a functional relationship between environmental complexity and the optimal step size: .
[0029] Furthermore, the construction of the local environment complexity and the ellipsoidal sliding window includes: using the ellipsoidal sliding window method to calculate the local environment complexity of the current position in the path search, and constructing the surrounding environment region of the current node during the random tree growth process.
[0030] Furthermore, the position and size of the ellipsoidal sliding window change with the complexity of the local environment in which the UAV is flying.
[0031] Example 2 A variable step size path planning method for unmanned aerial vehicles (UAVs) based on environmental complexity.
[0032] The path planning method includes the following five steps: Step 1, 3D environment modeling; Step 2, describing the path planning performance parameters; Step 3, exploring the relationship between the optimal step size and environment complexity; Step 4, constructing the local environment complexity and sliding window; Step 5, variable step size RRT algorithm.
[0033] The following is a step-by-step explanation of the method: Step 1, 3D environment modeling. Details are as follows: Description of environmental complexity. Low-flying drones encounter varying degrees of obstacle complexity in different environments. Diagrams illustrating different environments are shown below. Figure 1 As shown, Figure 1 (b) compared to Figure 1(a) Obstacles are more, and the environment is more complex. The complexity of different environments is determined by the size of the space and the position, shape, size of obstacles in the space, etc. The complexity of different environments can be quantitatively represented by the Hamming distance in the environment matrix.
[0034] As shown in Figure 2 , the obstacles and spaces in the environment where the UAV is located are as shown in Figure 2 . The environment area is equally lengthed in X, Y, and Z directions, and if there is an obstacle in the grid, it is recorded as 1; otherwise, it is recorded as 0. It is represented by a three-dimensional matrix, and the environment matrix is established as shown in Figure 2 (b). For convenience of calculation, the grid length is taken as 1 meter; in reality, the grid size can be adjusted appropriately according to the situation.
[0035] Based on the environment matrix, the complexity of obstacles in the environment is quantitatively represented by the "Hamming distance" of the logic string.
[0036] Let A and B be two logic strings with equal number of bits, pp represent the number of corresponding bits that are both 1, pq represent the number of corresponding bits that A is 1 and B is 0, qp represent the number of corresponding bits that A is 0 and B is 1, and qq represent the number of corresponding bits that A and B are both 0. Then the Hamming distance between the logic strings A and B is represented as: (1) If the logic string A is "0 1 0 1 0" and B is "0 1 1 0 0", then pp=1, pq=1, qp=1, and qq=2. The Hamming distance between A and B is represented as .
[0037] As shown in Figure 2 (b), the three-dimensional matrix C is divided into . The Hamming distance of the matrix C is defined as: ( ) (2) The Hamming distance of the z direction of the matrix C is: (3) Similarly, the Hamming distance of the x direction of the matrix C is , and the Hamming distance of the y direction of the matrix C is . Then the Hamming distance of the matrix C is: (4) The total Hamming distance in the environment space is represented as: (5) In formula (5), V represents the total volume of the three-dimensional environment space. As shown in Figure 2 a cubic environment, the total Hamming distance of the environment is
[0038] Formula (5) is defined as "environment complexity" for quantitatively representing the complexity of obstacles in the environment. It is easy to know that when there are more obstacles in the environment area, the environment is more complex, and the Hamming distance of the environment matrix is larger, and the environment complexity is higher.
[0039] The RRT random tree expands a node outward each time, and has a specified length, which is the distance from the new node to the current node, called step length. The size of the step length will affect the shape of the random tree and the performance of the algorithm; when the step length increases, the search efficiency improves, but the obstacle avoidance ability decreases; when the step length decreases, the obstacle avoidance ability improves, but the algorithm efficiency decreases. Generally speaking, the more complex the obstacles in the environment, the greater the environment complexity, and the smaller the step length should be, and vice versa. According to different environments, the optimal step length is selected to make the algorithm performance the highest.
[0040] Step 2, describe the path planning performance parameters. Specifically as follows: After a path planning is completed, in order to measure the advantages and disadvantages of the generated path, the following parameters are introduced to realize the performance evaluation of the planned path. By optimizing these parameters, the relationship between the optimal step length and the environment complexity under different environments is explored.
[0041] Time T: the total time consumed for a complete path planning, with the unit of seconds.
[0042] Distance X: the total length of the generated path, that is, the sum of the Euclidean distances between adjacent two points in all path nodes, as shown by the red curve in Figure 3 , with the unit of meters.
[0043] Smoothness S: a parameter for measuring the smoothness of the path, with the unit of radians. It is defined as the average value of the included angle (in radian system) formed by three consecutive points in all path nodes from the starting point to the ending point, as shown in Figure 4 . Wherein A, B and C are three consecutive points in the path from front to back, and the angle θ is the included angle between vectors AB and BC.
[0044] The path smoothness S is defined as: (6) In formula (6), n is the number of all path nodes, The included angle formed by three points in the path from front to back in succession.
[0045] Total cost F: In order to reasonably allocate the weight of each parameter and facilitate subsequent optimization design, according to Figure 3 The simulation data of path planning is shown in Table 1.
[0046] Table 1 Simulation data of path planning According to the data shown in Table 1, it is easy to know that the order of magnitude of the "distance" parameter is close to 10 2 , the order of magnitude of the "smoothness" parameter is close to 10 0 , and the order of magnitude of the "time" parameter is close to 10 -1 . In order to unify the order of magnitude, the total cost F is defined as: (7) Step 3: Optimal step length and environment complexity relationship exploration. Specifically as follows: The traditional RRT algorithm usually plans the path with the same step length. Generally speaking, the step length should be smaller in the environment with higher complexity. How to choose the optimal step length is particularly important. For environments with different obstacle complexities, if the appropriate step length cannot be selected, it will lead to problems such as reduced path search efficiency, poor stability, and easy to enter local optimum. The present application establishes a relationship model between the optimal step length and the environment complexity, and explores the optimal step length in different environments.
[0047] The total cost function of formula (7) is defined as the fitness function, and the computer is used to randomly generate environments with different complexities. According to the fitness function, the step length of path planning is optimized according to the genetic algorithm (Genetic algorithm), and the optimized step length and environment complexity data are obtained. Part of the data is shown in Table 2.
[0048] Table 2 Optimal step length and environment complexity relationship (partial data) The scatter plot obtained by preprocessing the experimental data such as standardization and normalization is shown in Figure 5 .
[0049] Figure 5 The trend of the medium red data points conforms to the exponential function model, so the elementary function is constructed as: (8) In formula (8), c is the environment complexity, p is the optimal step size, and a, b, and g are to-be-determined coefficients. According to experimental data, the function relationship between the environment complexity c and the optimal step size p is obtained by using the ordinary least squares fitting function: (9) Thus, the fitting model between the optimal step size and the environment complexity is established, and the fitting curve is shown in the blue curve in Figure 5 In the path planning of the unmanned aerial vehicle, based on the model, the optimal step size of the RRT algorithm can be calculated according to different environments.
[0050] Step 4, construction of the local environment complexity and the sliding window. Specifically as follows: In order to make the RRT algorithm have environmental adaptability and be able to dynamically change the step size in the path search according to the environment, the method adopts a sliding window to roll the calculation of the local environment complexity of the current position in the path search, and on this basis, the optimal step size of the next growth of the random tree is calculated by using the model of the environment complexity and the step size in formula (9).
[0051] The sliding window is a method for dynamically calculating the local environment complexity, and the purpose is to construct the environment region around the current node in the growth process of the random tree, and to enhance the local perception ability of the algorithm to the environment. The shape and size of the sliding window can be selected according to different scenes, and in order to adapt to the flight state of the unmanned aerial vehicle, an ellipsoid is adopted as the sliding window.
[0052] An ellipsoid sliding window is constructed with the current node as the center point. In order to construct the local environment region and facilitate calculation, the length of the three semi-axes of the ellipsoid window is three times the current step size p, and the long axis of the ellipsoid is consistent with the direction from the current node to the new node.
[0053] The schematic diagram of the sliding window is shown in Figure 6 . In the figure, the blue ellipsoid region is called the sliding window, and the position and size of the sliding window change with the complexity of the environment in which the unmanned aerial vehicle flies. The environment complexity of the region in the window is called the local environment complexity. Formula (9) is used for calculation.
[0054] Step 5, proposal of the variable step size RRT algorithm. Specifically as follows: The present application proposes a variable step size RRT algorithm based on the sliding window, which can dynamically change the step size according to the local environment in the path planning, so that the RRT algorithm has environmental adaptability and local environment perception ability, and improves the path planning performance.
[0055] First, the environment map is quantized into a three-dimensional matrix, and the starting point Qstart and the end point Q goal , Q start is added to the list of path nodes of the random tree, and an initial step size p0 is given: (10) In formula (10), c0 is the total complexity of the environment. c0 can be calculated by formula (5), and the environment space here is the global environment including the start point and the end point.
[0056] Secondly, Q start is marked as Q near , and the RRT algorithm is used to randomly sample a point Q near from Q rand , to determine whether the line connecting Q near and Q rand passes through the obstacle; if it does, Q rand is regenerated; if it does not, a new node Q new is generated with the current step size p0, Q new is added to the list of path nodes.
[0057] Then, an ellipsoidal sliding window is constructed with Q new as the center point, and the local environmental complexity c of the region within the window can be calculated by formula (5), and the environment space here is the local environment within the ellipsoid. Here, to adapt to the external configuration of the UAV, the x, y, and z semi-axes of the ellipsoid are taken as 2p, p, and p, respectively, and the x-axis of the ellipsoid is consistent with the direction from Q near to Q rand .
[0058] Next, the step size p for generating the next random tree is calculated: (11) Formula (11) gives the expression of the step size p for generating the next random tree. To limit the step size p so that it is not too large, the total complexity c0 of the environment is taken as the demarcation point. When the local complexity c is less than or equal to the total complexity c0, the step size p is a constant value; when the local complexity c is greater than the total complexity c0, the step size p is calculated according to formula (10).
[0059] Then, Q new is marked as Q near . Starting from Q near , a new node Q ’ is generated with the step size p, Q new is added to the list of path nodes, and this point is recorded as Q ’ . new new .
[0060] The above operations are repeated until Q new reaches the end point Qgoal Distance is less than step length p, Q goal is added to the list. Finally, the points in the path list are connected in turn, and a complete path is obtained.
[0061] The variable step length RRT algorithm is as shown in Algorithm 1: To verify the superiority of the variable step length RRT algorithm proposed in the application compared with the traditional RRT algorithm, the above two algorithms are simulated respectively based on a 2.60GHz, i7 processor computer, and MATLAB 2021a is used to write programs, and data is obtained and compared and analyzed.
[0062] First, a computer is used to randomly generate an environment map with a size of 200*200*200 (unit: meters), as shown in Figure 7 .
[0063] The environment map is quantized into a three-dimensional environment matrix, and the information of each point on the map is represented by the matrix, where 0 represents space and 1 represents an obstacle, as described in step 1. According to formula (10), the initial step length is given, and through a large number of repeated experiments, the traditional RRT algorithm and the variable step length RRT algorithm proposed in the application are verified respectively, and the performance parameters of the two algorithms are compared and a conclusion is drawn.
[0064] The simulation experiment results of the variable step length RRT algorithm proposed in the application and the traditional RRT algorithm are as shown in Figure 8 , and the following conclusions can be drawn. Figure 8 (a) is a path planning simulation schematic diagram of the traditional RRT algorithm before improvement, Figure 8 (b) is a simulation schematic diagram of the variable step length RRT algorithm proposed in the application. By comparing the two diagrams, it can be seen that the traditional RRT algorithm does not have environmental adaptability, the generated path is relatively tortuous, and the total distance is longer; the variable step length RRT algorithm proposed in the application can adapt to the local environment in path planning, and can change the step length in real time according to the complexity of the local environment, the generated path is relatively smooth, and the total distance is shorter.
[0065] Further, the superiority of the variable step length RRT algorithm is verified numerically, and the error caused by the randomness of the RRT algorithm is reduced. A multi-sample experiment scheme is adopted, the environmental complexity is randomly given, and different complexity environment maps are generated by a computer. A plurality of groups of repeated tests are performed on each map, and the average value is taken as the final experimental result. The experimental simulation data graph of the traditional RRT algorithm and the variable step length RRT algorithm proposed in the application is as shown in Figure 9 .
[0066] The variable step length RRT algorithm proposed in the application has obvious advantages in performance. As shown in Figure 9 , Figure 9(a) is a path comparison chart of traditional RRT algorithm and variable step RRT algorithm, Figure 9 (b) is a smoothness comparison chart of traditional RRT algorithm and variable step RRT algorithm, Figure 9 (c) is a time comparison chart of traditional RRT algorithm and variable step RRT algorithm, Figure 9 (d) is a total cost comparison chart of traditional RRT algorithm and variable step RRT algorithm. Figure 9 The following conclusions can be drawn: compared with the traditional RRT algorithm, the variable step RRT algorithm generates a path with shorter distance, which is shortened by about 5%-20% on average and is closer to the global optimum. The variable step RRT algorithm generates a path that is smoother than the traditional RRT algorithm, and the smoothness is reduced by about 25% on average, which is beneficial to improve the safety and reliability of the unmanned aerial vehicle in flight. The variable step RRT algorithm takes less time than the traditional RRT algorithm, especially in the environment with complex obstacles, which saves more than 50% of time and improves the efficiency of the algorithm. The variable step RRT algorithm generates a path with smaller total cost than the traditional RRT algorithm, which can reduce energy consumption in actual power equipment operation.
[0067] The variable step RRT algorithm has the following advantages over the traditional RRT algorithm: smoother path, no dead angle, shorter path, closer to global optimum, less time, high efficiency, small total cost, etc.
[0068] The above embodiments of the present application are described in detail, and the principles and implementation modes of the present application are described by applying specific examples. The above embodiment description is only used to help understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed; in summary, the content of the specification should not be understood as a limitation of the present application.
Claims
1. A variable step-size path planning method for unmanned aerial vehicles (UAVs) based on environmental complexity, characterized in that, The method includes a variable step size RRT algorithm, which includes: S1. Quantize the environment graph into a three-dimensional matrix, add the path starting point to the random tree path node list, and determine the initial step size; S2. Using the RRT algorithm, starting from the path start point, when the line connecting the random sampling point and the path start point does not pass through an obstacle, generate a new node with the initial step size and add the new node to the path node list. S3. Construct an ellipsoidal sliding window with the new node as the center point, and calculate the local environmental complexity of the region within the window. S4. Using the total environmental complexity as a dividing point, when the local environmental complexity is less than or equal to the total environmental complexity, the step size remains unchanged; when the local environmental complexity is greater than the total environmental complexity, the step size is recalculated and determined based on the local environmental complexity. S5. Mark the new node one as the new starting point. Starting from the new starting point, generate a new node two with the step size determined by not changing the step size or recalculating the step size. Add the new node two to the random tree list. S6. Repeat the above operation until the distance between the newly generated node and the endpoint is less than the step size.
2. The method according to claim 1, characterized in that, Step S1, which involves determining the initial step size, includes: In the formula, p0 is the initial step size, and c0 is the total complexity of the environment.
3. The method according to claim 1, characterized in that, Step S4, recalculating and determining the step size, includes: In the formula, p is the optimal step size, and c is the local environment complexity.
4. The method according to claim 1, characterized in that, The path planning method also includes: 3D environment modeling, description of path planning performance parameters, exploration of the relationship between optimal step size and environmental complexity, and construction of local environmental complexity and ellipsoidal sliding window.
5. The method according to claim 4, characterized in that, The three-dimensional environment modeling includes: rasterizing the three-dimensional environment map and representing obstacle information using a matrix.
6. The method according to claim 4, characterized in that, The description of path planning performance parameters includes: after path planning is completed, introducing distance, time, and smoothness parameters to evaluate the performance of the planned path; optimizing the parameters and establishing the relationship between the optimal step size and environmental complexity under different environments.
7. The method according to claim 4, characterized in that, The exploration of the relationship between the optimal step size and environmental complexity includes: constructing an elementary function: In the formula, c is the environmental complexity, p is the optimal step size, and α, β, and γ are undetermined coefficients.
8. The method according to claim 7, characterized in that, The exploration of the relationship between optimal step size and environmental complexity also includes: the functional relationship between environmental complexity and optimal step size: .
9. The method according to claim 4, characterized in that, The construction of the local environment complexity and the ellipsoidal sliding window includes: using the ellipsoidal sliding window method to calculate the local environment complexity of the current position in the path search, and constructing the surrounding environment region of the current node during the random tree growth process.
10. The method according to claim 9, characterized in that, The position and size of the ellipsoidal sliding window change with the complexity of the local environment in which the UAV is flying.