Path planning method in complex urban environment

Abstract

The invention discloses a path planning method in a complex urban environment. The path planning method comprises the following steps: S1, generating a feasible initial path in the complex urban environment by adopting a probabilistic road map algorithm based on an A * search algorithm; s2, optimizing the initial path through a particle swarm optimization algorithm improved by an artificial potential field algorithm to obtain an optimized path; and S3, performing route smoothing processing on the optimized path by using a cubic B-spline interpolation method to obtain a final path. According to the method, a path can be found more easily in a complex urban environment by using a probabilistic road map algorithm based on an A * search algorithm. Through the method of combining the artificial potential field algorithm and the particle swarm optimization algorithm, the high robustness of the particle swarm optimization algorithm can be kept while the efficiency is guaranteed. In the process of realizing the artificial potential field algorithm, modeling can be carried out on different urban environments, so that more targeted rapid convergence can be realized when a guiding direction is provided for the particle swarm optimization algorithm.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of path planning, and particularly relates to a path planning method in a complex urban environment. BACKGROUND

[0002] The methods for solving the path planning problem at present mainly fall into two categories: traditional path planning algorithms and intelligent bionic algorithms. Among them, the algorithms used in traditional path planning mainly include Dijkstra algorithm, A* algorithm, PRM and PRT algorithm. Compared with intelligent bionic algorithms, traditional path planning algorithms are more commonly used algorithms at present. For example, in 2020, Elaf Dhulkefl et al. used Dijkstra algorithm to find effective path solutions in two-dimensional and three-dimensional models. Unlike traditional path planning algorithms, intelligent bionic algorithms mainly use intelligent optimization algorithms, such as genetic algorithm and particle swarm optimization algorithm, to find the optimal path. Although this method has better robustness compared with traditional path planning algorithms, it is usually low in efficiency and needs large computational overhead, and cannot be well applied to real-time path planning.

[0003] So far, many researchers have had a deep study on traditional path planning algorithms. A mainstream method in the solution of traditional path planning problems is a graph-based path planning algorithm. This algorithm first needs to abstract the space into a directed or undirected graph, and then uses the method of graph theory to solve the path. The path obtained by this method is usually more accurate. However, the main problem of this method is that as the running range increases, or when a high-dimensional problem is encountered, the curse of dimensionality and other problems will be encountered. Taking Dijkstra algorithm proposed in 1959 as an example, the complexity of this algorithm is wherein is the number of nodes in the graph. As the space increases, the number of nodes will increase exponentially, thereby the efficiency and overhead of the algorithm increase sharply. In view of this problem, Bao Peiming proposed that the shortest path in geographic space can be considered to use only the nodes near the end nodes of the shortest path to find the shortest path, thereby reducing the number of nodes involved in the algorithm and improving the efficiency. However, if traditional path planning needs to be used, the real space needs to be abstracted into a graph, and the path obtained is a polyline, which is not consistent with the real situation and increases the running cost overhead. Li Wen-hui et al. smoothed the path obtained by B-spline interpolation, effectively reducing the path length.

[0004] While intelligent optimization algorithms demonstrate better performance than traditional path planning algorithms, their high overhead and low efficiency prevent them from being used for real-time path planning problems. To address this issue, Huang et al. used the A* algorithm to optimize the initial point of the ant colony search algorithm, thus solving the efficiency problem caused by the randomness of the initial point. This method ensures both efficiency and the quality of the results. Particle swarm optimization (PSO), a commonly used intelligent optimization algorithm, not only suffers from the above drawbacks but is also prone to getting trapped in local optima. To address this, Baoxian Liang et al. proposed a search direction adjustment mechanism based on subpopulation partitioning, solving the problem of PSO easily getting trapped in local optima.

[0005] The intelligent optimization algorithm mentioned above can directly find a path in a simple environment, but in the complex urban environment where the low-altitude economy is located, it is difficult for the intelligent optimization algorithm to find an effective solution directly. In this case, it is necessary to perform local optimization on the path found by the traditional path planning algorithm. Summary of the Invention

[0006] To address the aforementioned shortcomings in existing technologies, this invention provides a path planning method for complex urban environments, which solves the problem of difficult local path processing in existing path planning processes.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by this invention is: a path planning method in a complex urban environment, comprising the following steps: S1. A probabilistic road map algorithm based on the A* search algorithm is used to generate feasible initial paths in complex urban environments; S2. The initial path is optimized using a particle swarm optimization algorithm improved by the artificial potential field algorithm to obtain the optimized path; S3. Use cubic B-spline interpolation to smooth the optimized path and obtain the final path.

[0008] Furthermore, S1 includes the following sub-steps: S11. Random sampling on the map n One node; S12. For each node, search for its accessible nodes and establish connections to form an adjacency list; S13. Initialize the open and closed lists: the open list contains only the starting point, and the closed list is an empty set; S14. Take the point with the minimum cost in the open list as the current node and put it into the closed list. Take the adjacency list of the current node and determine whether the adjacency list contains the endpoint. If so, start from the endpoint and search backwards to the starting point based on the parent node information to generate a feasible initial path; otherwise, proceed to S15. S15, updating the parent node information of the node or adding the node to the open list for each node in the adjacency list of the current node, and returning to S14.

[0009] The above further scheme has the beneficial effect that the probabilistic roadmap algorithm based on the A* search algorithm can effectively reduce the space and time cost of the search path and obtain a more flexible path planning scheme in a complex urban environment.

[0010] Further, in S15, the method of updating the parent node information of the node or adding the node to the open list is specifically: determining whether the node belongs to the open list, if yes, updating the parent node of the node to the current node in response to the cost of the node calculated by the heuristic cost function being less than the cost of the node calculated before; if not, recording the current node as the parent node of the node and adding the node to the open list.

[0011] Further, the heuristic cost function is used to calculate the cost of the node. n In the formula, is the current cost of the node, representing the shortest distance from the starting point to the node along the path found, n is the heuristic cost of the node, representing the estimated cost from the node to the end point. n n n

[0012] Further, S2 includes the following steps: S21, establishing an initialized particle swarm according to the initial path and initializing the iteration number; S22, calculating the fitness of the particles in the particle swarm to obtain the historical optimal position and the global optimal position; S23, calculating the predicted optimal position according to the global optimal position by using the artificial potential field algorithm; S24, updating the position of the particle based on the predicted optimal position by using the position updating formula and the improved velocity formula of the particle swarm optimization algorithm; S25, determining whether the iteration number reaches the maximum iteration number, if yes, outputting the historical optimal position of the particle swarm to generate an optimized path; if not, increasing the iteration number by one and returning to S22.

[0013] ​​​​​The beneficial effects of the above-mentioned further solutions are as follows: This invention combines the artificial potential field algorithm with the particle swarm optimization algorithm, which effectively improves the convergence efficiency of the algorithm and retains the advantage of the high robustness of the particle swarm optimization algorithm.

[0014] Furthermore: S21 specifically refers to: The initial path is copied several times and a small perturbation value is added to obtain the initial values ​​of the particles in the initial particle swarm.

[0015] Furthermore: S23 specifically refers to: S231. Obtain the set of nodes to be optimized through the global optimal position, and initialize the number of iterations; S232. Combine the set of nodes to be optimized with the starting point and the ending point to obtain the complete path; S233. Calculate the attractive and repulsive forces of the intermediate nodes of the complete path, and calculate the possible positions of the intermediate nodes after the resultant forces are updated. S234. If the possible position after the intermediate node update is valid, then update the value of the corresponding node in the set of nodes to be optimized according to the possible position after the intermediate node update. S235. Determine if the number of iterations has reached the maximum iteration value. If yes, obtain the predicted optimal position based on the set of nodes to be optimized. If no, increment the number of iterations by one and return to S232. In S234, the method for verifying the validity of the possible position of the intermediate node after the update is as follows: if the possible position of the intermediate node after the update does not belong to the set of obstacles, and the line connecting the updated intermediate node and its adjacent nodes does not collide with any obstacle, then the possible position of the intermediate node after the update is valid.

[0016] Furthermore, in S24, the expression for updating the particle's position is as follows: In the formula, For the first particle i The position of +1 iteration, For the first particle i The position of the next iteration. For the first particle i The speed of each iteration; In the formula, For the first particle i The speed of -1 iteration For the first particle i The position of the next iteration. , , and For range random numbers, This is the best historical position. The optimal position globally. To predict the optimal location, The coefficients for the random disturbance term are set according to the size and complexity of the map. For its own learning factors, As a group learning factor, To predict the learning factor.

[0017] The beneficial effects of the above-mentioned further scheme are as follows: the use of artificial potential field algorithm to improve the velocity update formula of particle swarm optimization algorithm effectively improves the convergence efficiency of path optimization algorithm and solves the problem of low convergence efficiency of particle swarm optimization algorithm.

[0018] Furthermore: In S3, cubic B-spline interpolation method The specific expression is: In the formula, To optimize the path in the first l One control point, To optimize the path in the first l +1 control point, To optimize the path in the first l +2 control points, To optimize the path in the first l +3 control points t For taking values The parameter variables, .

[0019] The beneficial effects of this invention are as follows: This invention provides a path planning method for complex urban environments. First, a probabilistic roadmap algorithm based on the A* search algorithm is used to generate feasible initial paths in complex urban environments. Then, a particle swarm optimization algorithm based on an improved artificial potential field algorithm is proposed. This algorithm uses the artificial potential field algorithm to improve the velocity update formula of the particle swarm optimization algorithm, effectively improving the convergence efficiency of the path optimization algorithm and solving the problem of low convergence efficiency in the particle swarm optimization algorithm. Finally, cubic B-spline interpolation is used to smooth the paths, making the results more closely resemble real-world conditions. Compared with existing technologies, it has the following advantages: (1) Using the probabilistic road map algorithm based on the A* search algorithm can more easily find paths in complex urban environments, improve the efficiency of path planning, help alleviate low-altitude traffic pressure, and reduce the probability of safety accidents caused by flight route conflicts.

[0020] (2) By combining the artificial potential field algorithm and the particle swarm optimization algorithm, the strong robustness of the particle swarm optimization algorithm can be preserved while ensuring efficiency. In the process of implementing the artificial potential field algorithm, different urban environments can be modeled, so that it can converge more quickly and in a more targeted manner when providing guidance for the particle swarm optimization algorithm.

[0021] (3) Using cubic B-spline interpolation can make the route planned by the algorithm more closely resemble the real situation. Attached Figure Description

[0022] Figure 1 This is a flowchart of a path planning method for a complex urban environment according to the present invention. Detailed Implementation

[0023] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0024] like Figure 1 As shown, in one embodiment of the present invention, a path planning method in a complex urban environment includes the following steps: S1. A probabilistic road map algorithm based on the A* search algorithm is used to generate feasible initial paths in complex urban environments; S2. The initial path is optimized using a particle swarm optimization algorithm improved by the artificial potential field algorithm to obtain the optimized path; S3. Use cubic B-spline interpolation to smooth the optimized path and obtain the final path.

[0025] In this embodiment, the present invention first uses a probabilistic road map algorithm based on the A* search algorithm to generate an initial path in a complex urban environment, then uses a particle swarm optimization algorithm improved by the artificial potential field algorithm to optimize the path, and finally uses cubic B-spline interpolation to smooth the path.

[0026] The specific steps of the probabilistic road graph algorithm based on the A* search algorithm are as follows: This algorithm estimates the cost of the current path using a heuristic function, given the known optional nodes. In an environment where the cost of reaching the target is relatively low, the A* search algorithm searches for the path with the lowest cost. At each step, it selects the node with the lowest cost, ensuring the optimal path is found under the current conditions, and in most cases, it has high search efficiency. Traditional A* search algorithms, however, only search within a given starting point. ,end and obstacle sets Two-dimensional grid The shortest path is found in the algorithm, and the optimal path found by the A* search algorithm is recorded as follows: ,in , .

[0027] If the map is gridded, the time and space overhead of the A* search algorithm increases dramatically, especially on large maps or maps with complex obstacles. To simplify the algorithm, map gridding can be avoided. In this case, a combination of probabilistic roadmap algorithms and A* search can be used to search for effective paths on non-gridded maps. This embodiment uses a probabilistic roadmap algorithm based on A* search, which not only effectively reduces the space and time overhead of path searching but also provides more flexible path planning solutions in complex urban environments. A set of obstacles for each obstacle. Map The steps of the probabilistic roadmap algorithm based on the A* search algorithm are as follows: S11. Random sampling on the map n Nodes Let it be a set , among which, the i node Record the starting point ,end ; S12. For each node, search for its walkable nodes and establish connections to form an adjacency list; in this embodiment, for the first node... i node q i The resulting adjacency list , , adjacency list any point in, and q i The lines connecting them do not collide with any obstacles; S13. Initialize the open list and closed list: The open list contains only the starting point, and the closed list is an empty set; in this embodiment, the open list... Closed list Remember the starting point. The cost is In the formula, This refers to the Euclidean distance between the starting point and the ending point. The parent node is an empty set; S14. Take the point with the minimum cost in the open list as the current node and put it into the closed list. Take the adjacency list of the current node and determine whether the adjacency list contains the endpoint. If so, start from the endpoint and search backwards to the starting point based on the parent node information to generate a feasible initial path; otherwise, proceed to S15. S15. For each node in the adjacency list of the current node, update the parent node information of the node or add the node to the open list, and return to S14.

[0028] In S15, the specific methods for updating the parent node information of a node or adding a node to the open list are as follows: Determine if a node belongs to the open list. If so, in response to the fact that the cost of the node calculated by the heuristic cost function is less than the cost of the node calculated previously, update the parent node of the node to the current node. In this embodiment, the node is a point belonging to the adjacency list of the current node. If the node belongs to the open list, it means that there is another path that can reach the node without passing through the current node. Therefore, the node has an original cost, which is the cost previously calculated for the node. Another cost can be calculated based on the path of the current node using the heuristic cost function, which is the cost of the node calculated by the heuristic cost function this time. By comparing the two costs, the better path in the current case can be determined.

[0029] If not, then record the current node as its parent node and add the node to the open list.

[0030] Optimized path obtained by probabilistic path graph algorithm based on A* search algorithm ; In the formula, For the first i node q i coordinates T The transpose symbol is used; S13 to S15 of the above algorithm are the main steps of the A* search algorithm, and S11 to S22 are to simulate map gridding by placing random scattered points on the map and then constructing an adjacency list, so as to make the generated paths more flexible and varied.

[0031] Through heuristic cost function Calculate the first n The cost of a node; In the formula, For the first n The current cost of a node represents the cost from the starting point along the found path to the node. n The shortest distance between nodes, For the first n The heuristic cost of a node represents the cost of starting from the first node. n The estimated cost of a node's distance to the destination is typically expressed using Euclidean distance or Manhattan distance. This embodiment uses Euclidean distance to estimate the heuristic cost. That is, if the current node is... And the set of intermediate nodes for the shortest path to the current node found by the algorithm is: Then the first n Current cost of the node Inspirational cost .

[0032] S2 includes the following steps: S21. Establish an initial particle swarm based on the initial path and initialize the number of iterations; S22. Calculate the fitness of particles in the particle swarm to obtain the historical best position and the global best position; S23. Calculate and predict the optimal position based on the global optimal position using the artificial potential field algorithm; S24. Based on the predicted optimal position, update the particle position using the position update formula of the particle swarm optimization algorithm and the improved velocity formula. S25. Determine if the number of iterations has reached the maximum number of iterations. If yes, output the historical best position of the particle swarm and generate an optimized path. If no, increment the number of iterations and return to S22.

[0033] S21 specifically refers to: The initial path is copied several times and a small perturbation value is added to obtain the initial values ​​of the particles in the initial particle swarm.

[0034] In this embodiment, when using the particle swarm optimization algorithm for path optimization, a path excluding the start and end points is typically treated as a single particle. In a two-dimensional environment, path optimization is usually employed. The two-dimensional array shown represents a path, which needs to be rewritten as follows: The one-dimensional form shown: Then copy the initialization path These initial values ​​are used as the initial values ​​for the particle swarm optimization algorithm, and are recorded in [the relevant database]. matrix Inside.

[0035] Since particle swarm optimization (PSO) is an optimization process that is reflected in the positions of particles, and the initial particle values ​​obtained simply by copying the initial solution are prone to getting trapped in local optima, we will further refine this process here. Add a small perturbation value The initial value for generating particles This increases the diversity of the algorithm's search, thereby improving the algorithm's search efficiency.

[0036] In the formula, as one The matrix contains elements of... random numbers, This represents the upper limit of the small perturbation value, and its value varies depending on the map size and environment.

[0037] In S22, the fitness of each particle is calculated based on the cost function, and the historical best position of each particle and the global best position of the particle swarm are obtained after initialization.

[0038] S23 specifically refers to: S231. Obtain the set of nodes to be optimized through the global optimal position, and initialize the number of iterations; S232. Combine the set of nodes to be optimized with the starting point and the ending point to obtain the complete path; S233. Calculate the attractive and repulsive forces of the intermediate nodes of the complete path, and calculate the possible positions of the intermediate nodes after the resultant forces are updated. S234. If the possible position after the intermediate node update is valid, then update the value of the corresponding node in the set of nodes to be optimized according to the possible position after the intermediate node update. S235. Determine if the number of iterations has reached the maximum iteration value. If yes, obtain the predicted optimal position based on the set of nodes to be optimized. If no, increment the number of iterations by one and return to S232. In S234, the method for verifying the validity of the possible position of the intermediate node after the update is as follows: if the possible position of the intermediate node after the update does not belong to the set of obstacles, and the line connecting the updated intermediate node and its adjacent nodes does not collide with any obstacle, then the possible position of the intermediate node after the update is valid.

[0039] In this embodiment, the construction of the artificial potential field needs to refer to the specific map scene. The idea can be similar to the method used when applying the artificial potential field algorithm to dynamic obstacle avoidance, that is, to let obstacles or no-fly zones and map boundaries provide a repulsive force for the point to be optimized, and the neighboring nodes of the node to be optimized provide an attractive force. Finally, the optimization of the target node can be completed through multiple iterations. The pseudocode of the artificial potential field algorithm optimization path is shown in Table 1.

[0040] Table 1. Pseudocode of the path optimization algorithm based on artificial potential field The artificial potential field algorithm described in the pseudocode above requires the starting point to be input. ,end Node set to be optimized A collection of obstacles in the map And update the step size; finally, the output is the optimized node to be optimized. .

[0041] In S24, the derivation steps for updating the particle position using the position update formula and the improved velocity formula of the particle swarm optimization algorithm are as follows: The velocity formula for traditional particle swarm optimization algorithms consists of three velocities: memory velocity, individual cognition velocity, and swarm cognition velocity. Memory velocity refers to the velocity of the particle in the previous iteration, and it is typically assigned a maximum value. ,like ,but The speed of individual cognition and the speed of collective cognition are determined by the particle's historical best position. and global optimal position The decision is made. Therefore, the particle velocity update formula is: Since the initial positions of the particles are generated using a local optimum obtained by an algorithm, the particle swarm optimization algorithm is prone to getting trapped in local optima when used in complex urban environments. Therefore, a random perturbation term is added to the velocity update formula to help the algorithm escape local optima. The velocity update formula for particles with the added random perturbation term is as follows: Predict the optimal location When added to the velocity update formula of the particle swarm optimization algorithm, the new velocity update formula is obtained as follows: In the formula, For the first particle i The speed of -1 iteration For the first particle i The position of the next iteration. , , and For range random numbers, This is the best historical position. The optimal position globally. To predict the optimal location, The coefficients for the random disturbance term are set according to the size and complexity of the map. For its own learning factors, As a group learning factor, To predict the learning factor.

[0042] The particle position can be updated using the position update formula of the particle swarm optimization algorithm: In the formula, For the first particle i The position of +1 iteration, For the first particle i The position of the next iteration. For the first particle i The speed of each iteration; In summary, by following the above algorithmic process, the artificial potential field algorithm and the particle swarm optimization algorithm can be combined, which can effectively improve the convergence efficiency of the algorithm and retain the high robustness of the particle swarm optimization algorithm.

[0043] In S3, cubic B-spline interpolation can be used to smooth the route, making it more closely resemble the actual path. The specific steps are as follows: The combination of optimized path and starting point is obtained from Control points The controlled path, and containing List of node vectors Among them, the cubic B-spline curve segment can be obtained. The specific expression is: In the formula, p l For the first l One control point, The path is The B-spline basis functions, also known as harmonic functions, are commonly obtained using the Cox-deBoor recursive formula. In the formula, The smoothness of the interpolated curve is determined by the cubic B-spline interpolation method. And for List of node vectors ,because Therefore, the first and last nodes are repeated. Next, that is, node The following conditions must be met: for The value is usually obtained using the equal interval method, that is: Will Substitute the value This allows us to obtain the cubic B-spline basis functions: Therefore, when At that time, the segment of the cubic B-spline curve The expression is rewritten as: The matrix form expression is as follows: In the formula, To optimize the path in the first l One control point, To optimize the path in the first l +1 control point, To optimize the path in the first l +2 control points, To optimize the path in the first l +3 control points t For taking values The parameter variables, .

[0044] In the description of this invention, it should be understood that the terms "center," "thickness," "upper," "lower," "horizontal," "top," "bottom," "inner," "outer," and "radial," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying the relative importance or the number of technical features implicitly specified. Therefore, a feature defined by "first," "second," and "third" may explicitly or implicitly include one or more of that feature.

Claims

1. A path planning method in a complex urban environment, characterized in that, The method comprises the following steps: S1, generating an initial feasible path in a complex urban environment by using a probabilistic roadmap algorithm based on an A* search algorithm; S2, optimizing the initial path by using a particle swarm optimization algorithm improved by an artificial potential field algorithm to obtain an optimized path; S3, performing route smoothing processing on the optimized path by using a cubic B-spline interpolation method to obtain a final path.

2. The path planning method in a complex urban environment according to claim 1, wherein, S1 comprises the following steps: S11, randomly sampling in the map n a node; S12, searching for passable nodes for each node and establishing a connection to form an adjacency list; S13, initializing an open list and a closed list: the open list only has a starting point, and the closed list is an empty set; S14, taking a node with the minimum cost in the open list as a current node, putting the current node into the closed list, taking an adjacency list of the current node, judging whether the adjacency list contains an end point, if yes, searching for an effective path from the end point to the starting point in a reverse direction according to parent node information to generate an initial feasible path; if no, entering S15; S15, updating parent node information of each node in the adjacency list of the current node or adding the node into the open list, and returning to S14.

3. The path planning method in complex urban environment according to claim 2, characterized in that, In S15, the method of updating the parent node information of the node or adding the node into the open list is specifically as follows: judging whether the node belongs to the open list, if yes, updating the parent node of the node to the current node in response to the fact that the cost of the node calculated by using a heuristic cost function this time is less than the cost of the node calculated before; if no, recording the current node as the parent node of the node, and adding the node into the open list.

4. The path planning method in a complex urban environment according to claim 3, wherein, By heuristics cost function Computing the cost of the n Node's cost; In the formula, For the first n The current cost of a node represents the cost from the starting point along the found path to the node. n The shortest distance between nodes, For the first n The heuristic cost of a node represents the cost of starting from the first node. n Estimated cost of node distance from the destination.

5. The path planning method in complex urban environment according to claim 1, characterized in that, S2 comprises the following steps: S21, establishing an initialized particle swarm according to the initial path, and initializing an iteration number; S22, calculating the fitness of particles in the particle swarm to obtain a historical optimal position and a global optimal position; S23, calculating a predicted optimal position according to the global optimal position by using an artificial potential field algorithm; S24, updating the position of the particle based on the predicted optimal position by using a position updating formula and an improved speed formula of the particle swarm optimization algorithm; S25, judging whether the iteration number reaches a maximum iteration number, if yes, outputting the historical optimal position of the particle swarm to generate an optimized path; if no, increasing the iteration number by one, and returning to S22.

6. The path planning method in a complex urban environment according to claim 5, wherein, S21 is specifically as follows: copying the initial path several times, and adding a small disturbance value to obtain initial values of particles in the initialized particle swarm.

7. The path planning method in complex urban environment according to claim 5, characterized in that, S23 is specifically as follows: S231, obtaining a set of nodes to be optimized by using the global optimal position, and initializing an iteration number; S232, combining the set of nodes to be optimized with a starting point and an end point to obtain a complete path; S233, calculating the gravity and the repulsion of an intermediate node of the complete path to calculate a possible position of the intermediate node after updating; S234, in response to the fact that the possible position of the intermediate node after updating is legal, updating the value of a corresponding node in the set of nodes to be optimized according to the possible position of the intermediate node after updating; S235, judging whether the iteration number reaches a maximum iteration number, if yes, obtaining the predicted optimal position according to the set of nodes to be optimized; if no, increasing the iteration number by one, and returning to S232; In S234, the method for checking whether the updated possible position of the intermediate node is legal is specifically as follows: in response to the updated possible position of the intermediate node not belonging to the obstacle set, and the connecting line between the updated intermediate node and the adjacent node not colliding with the obstacle, the updated possible position of the intermediate node is legal.

8. The path planning method in complex urban environment according to claim 5, characterized in that, In S24, the expression for updating the position of the particle is specifically as follows: wherein is the position of the particle at the first iteration, i is the position of the particle at the first iteration, is the position of the particle at the first iteration, i is the position of the particle at the first iteration, is the position of the particle at the first iteration, i is the position of the particle at the first iteration, wherein, is the position of the particle at the i-th iteration, i is the velocity of the particle at the i-th iteration, is the position of the particle at the i-th iteration, i is the position of the particle at the i-th iteration, , , and is a random number in the range , is the history optimal position, is the global optimal position, is the predicted optimal position, is the coefficient of the random perturbation term, which is set according to the size and complexity of the map, is the self-learning factor, is the group-learning factor, is the prediction-learning factor.

9. The path planning method in complex urban environment according to claim 5, characterized in that, In S3, cubic B-spline interpolation method The specific expression is: In the formula, To optimize the path in the first l One control point, To optimize the path in the first l +1 control point, To optimize the path in the first l +2 control points To optimize the path in the first l +3 control points t For taking values The parameter variables, .

Images