Improved hybrid path planning method for mobile robot based on a* algorithm and dynamic window approach

Abstract

The application discloses a hybrid path planning method for mobile robots by improving A* algorithm and dynamic window method, first, in order to effectively reduce the occurrence frequency of redundant points in the global planning path, the search range is expanded to 5*5, and the search direction is 16. In order to avoid unnecessary node search, a forward and reverse alternating search mechanism is introduced, and an exponential decay constraint is added to the cost function, so that the cost function can be adaptively changed. In order to facilitate local path planning, a nonlinear feedback controller is introduced to smooth the global optimal path obtained by the improved A* algorithm. Finally, in order to adapt to the change of the environment, a fuzzy controller is introduced, the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle are input, and the weight coefficients of three trajectory evaluation functions are output, so that the local trajectory planning of the mobile robot is realized under the dynamic window, and the change of the environment is well adapted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot path planning, and particularly relates to a hybrid path planning method for mobile robots by improving A* algorithm and dynamic window method. BACKGROUND

[0002] In order to complete autonomous navigation in a working environment, a mobile robot needs to have the ability of environment perception, simultaneous localization and mapping, motion control and path planning. Path planning refers to determining the best path for a robot from a starting position to a target position in a given environment. Path planning is an indispensable part of mobile robot navigation technology, which reflects the interaction between the robot and the surrounding environment during movement, and is the basis and safety guarantee for the robot to complete the task.

[0003] Path planning technology for mobile robots includes global path planning and local path planning. However, with the development of mobile robot technology, the working environment of mobile robots has become more complex, and the use of global path planning or local path planning alone cannot meet the requirements of autonomous navigation of mobile robots. At present, researchers have proposed the idea of hybrid path planning to solve the problems of high computational complexity and poor real-time performance of traditional algorithms in complex environments, and to improve the quality of path planning for mobile robots.

[0004] Hybrid path planning generally plans an optimal path to the end point based on global path planning algorithm by perceiving all environmental information offline, and then realizes online path planning based on local path planning algorithm using real-time environmental information obtained by sensors. According to the real-time information of obstacles, the planned path is constantly adjusted. Researchers generally apply A* algorithm to global path planning in the past. Based on the obtained global path, dynamic window method is used for local path planning and adjustment, which comprehensively utilizes the global planning ability of A* algorithm and the local adjustment ability of dynamic window method, and improves the efficiency and quality of path planning.

[0005] For global path planning, many scholars have proposed improvements to the A* algorithm, a heuristic search algorithm. Guruji et al. proposed a time-optimal A* algorithm that only calculates the heuristic value when there is a collision hazard at the node, at the cost of a certain path length and smoothness, significantly improving the search efficiency. Shen Xianqing et al. proposed an improved smooth A* algorithm that classifies the neighborhood nodes into levels and prioritizes the search for the child nodes in the high-level group. Meanwhile, the bidirectional smoothing concept is introduced into the Floyd-Warshall algorithm, reducing the number of inflection points in the path but significantly increasing the search time. Chen et al. proposed an improved A* algorithm that narrows the search range by constructing a cross product based on the angle relationship between the current grid and the starting grid, and reasonably narrows the search direction of the current node in the introduced heuristic function, improving the search efficiency of the A* algorithm but reducing the smoothness of the optimal path.

[0006] Previous researchers generally cannot balance path planning efficiency, path length, and path corner, so it is necessary to optimize the search strategy and cost function of the A* algorithm. Meanwhile, considering the physical structure of mobile robots, the safety distance between the robot and the obstacle must be considered, and most researchers usually consider the mobile robot as a point mass for the convenience of path planning algorithm research, so it is necessary to study path smoothing algorithms.

[0007] For local path planning, many scholars have improved the traditional dynamic window algorithm to further improve the planning efficiency and path quality of the traditional dynamic window method in various application scenarios. Chang et al. proposed an improved dynamic window algorithm based on the Q-Learning algorithm, introducing two new evaluation functions to enhance the global navigation performance of the traditional dynamic window method, but the computational complexity is greatly increased. Liu et al. introduced an evaluation function related to direction change into the trajectory evaluation function of the traditional dynamic window method, reducing the unnecessary turning rate of the robot and improving the planning efficiency of the dynamic window method, but when there is an obstacle between the target position and the current position, the evaluation function cannot be dynamically adjusted.

[0008] Previous researchers usually only study the improvement of the dynamic window method evaluation function, but in complex environments, the robot cannot accurately determine the surrounding environment, and the situation of obstacles between the target position and the current position frequently occurs. Angle evaluation, speed evaluation, and distance evaluation from obstacles cannot be dynamically adjusted, causing the mobile robot to appear unable to bypass obstacles, unable to pass through narrow areas, and easily miss the target point during local path planning. Therefore, it is necessary to improve the dynamic window method.

[0009] There are many places worth improving and optimizing in the mixed path planning algorithm of the mobile robot through the above analysis. Firstly, for the A* algorithm of the global path planning, not only the length of the optimal path, but also the search efficiency of the path planning should be considered. How to improve the search strategy of the A* algorithm and adjust the proportion of the heuristic function and the evaluation function in the cost function is a problem to be solved. In addition, for the safety distance between the mobile robot and the obstacle, a good performance path smoothing algorithm needs to be designed to reduce the turning angle of the path and avoid the collision between the robot and the obstacle. SUMMARY

[0010] In order to effectively reduce the number of redundant points in the global planning path, the expansion search range is 5x5 and the expansion search direction is 16. In order to avoid unnecessary node search, the forward and reverse alternating search mechanism is introduced, and the exponential decay constraint is added to the cost function, so that the cost function can adaptively change. In order to facilitate local path planning, a nonlinear feedback controller is introduced to smooth the global optimal path obtained by the improved A* algorithm. Finally, in order to adapt to the change of the environment, a fuzzy controller is introduced, which inputs the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle, and outputs the weight coefficients of the three trajectory evaluation functions, so as to realize the local trajectory planning of the mobile robot under the dynamic window.

[0011] The technical scheme adopted by the present application is as follows, a mixed path planning method of mobile robot based on improved A* algorithm and dynamic window method, comprising the following steps:

[0012] S1: obtaining the known environment information of the mobile robot workspace, establishing the corresponding environment model by using the grid method, and determining the positions of the starting point and the target point in the environment model;

[0013] S2: using the improved A* algorithm, exponentially decaying the heuristic function in the cost function, using the bidirectional alternating search algorithm, searching in the expanded search neighborhood and search direction, until the forward node and the reverse node meet, and integrating the forward and reverse paths into a global optimal path from the starting point to the target point;

[0014] S3: extracting the turning point according to the global optimal path, taking it as a local target point, combining the fuzzy controller with the dynamic window method, dynamically adjusting the weight proportion of the target azimuth angle evaluation function, the obstacle gap evaluation function and the speed evaluation function in the trajectory evaluation function according to the real-time information of the environment where the local target point is located, and planning a local path around the obstacle;

[0015] S4: smoothing the global optimal path by using a nonlinear feedback controller;

[0016] S5: After bypassing the obstacle, the mobile robot returns to the global optimal path and continues to move along the global optimal path until it reaches the current local target point, and switches to the next local target point;

[0017] S6: Determine whether the current local target point reached by the mobile robot is the global target point, if not, jump to S4, otherwise, execute S7;

[0018] S7: The mobile robot reaches the global target point, completing the path planning task.

[0019] Further, the improved A* algorithm includes improvements in search direction and search range, as follows:

[0020] Within a 5x5 search range, use 16 connected modes to expand the search step and search area; at the same time, divide the 16 search directions into 4 different search areas, and the search area of the node is guided by the target point, and the area corresponding to the direction of the connecting line between the target point and the path point is the forward area; with the change of the position of the path point and the target point in the search process, the search area also changes.

[0021] Further, the improved A* algorithm includes improvements in search strategy, as follows:

[0022] Introduce a bidirectional search strategy, switch the current target point of each through forward search and reverse search to ensure that the bidirectional search meets near the geometric center of the starting point and the target point, thereby alternately replacing the best current node;

[0023] First, start forward search, take S as the starting point and G as the target point to obtain a forward optimal node S; then, perform reverse search, take G as the starting point and the node obtained by forward search as the target node, at this time, obtain a reverse optimal node G, and repeat iteration until meeting and exiting.

[0024] Further, the improved A* algorithm includes improvements in the heuristic function, as follows:

[0025] The heuristic function in the cost function is weighted in an exponential decay manner, as follows:

[0026] f(n)=g(n)+e h(n) ·h(n)

[0027] Wherein, f(n) represents the cost function from the starting node to node n, g(n) represents the dissipation function, i.e. the actual cost from the starting node to node n, and h(n) represents the heuristic function, i.e. the estimated cost from the starting node to node n.

[0028] Further, the improved A* algorithm comprises: smoothing the global optimal path by using a nonlinear feedback controller, specifically as follows:

[0029] Let ρ be the cluster between the current path point and the target point, is the angle between the Y-axis of the current path point vector and the target point vector, and φ is the angle between the X-axis of the current path point vector and the target point, wherein the Y-axis of the current path point vector is consistent with the direction of the current path point;

[0030] By inputting the above angle variables, the expression of the speed command (v, ω) output by the nonlinear feedback controller is:

[0031]

[0032] wherein K ρ , and K φ are the gains of the corresponding variables, and K v is the speed gain.

[0033] In order to ensure the smoothness of the path, the nonlinear feedback controller should satisfy the following conditions:

[0034]

[0035] After obtaining a series of speed commands (v, ω), a series of discrete points constituting the smooth path are obtained within the iteration time Δt according to the commands through iterative calculation, and the iterative formula of the node pose (x, y, θ) is:

[0036]

[0037] wherein x and y are the horizontal and vertical coordinates of the current node, and θ is the direction angle of the current node.

[0038] Further, in the local path planning process, a two-input three-output fuzzy controller with a membership function of trapezoidal function and triangular function is introduced;

[0039] The input quantity is obtained according to the real-time collected environmental information, i.e. the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle; wherein the domain of the distance from the mobile robot to the target point is [0, 4], and the fuzzy set is {PS, PM, PB}, which respectively corresponds to the distance from the mobile robot to the target point as near, medium and far; the domain of the distance from the mobile robot to the nearest obstacle is [0, 2], and the fuzzy set is {PS, PM, PB}, which respectively corresponds to the distance from the mobile robot to the nearest obstacle as near, medium and far.

[0040] The output quantity is obtained by fuzzy logic, i.e. the weights of the target azimuth angle evaluation function, the obstacle gap evaluation function and the speed evaluation function, the domain is [0, 1], the corresponding fuzzy set is {PS, PM, PB}, and the weights are small, medium and large respectively.

[0041] Further, the fuzzy vector is obtained by reasoning the fuzzy logic by using the Mamdain type fuzzy reasoning method, the output control quantity is obtained by using the barycentric method to solve the fuzzification, the weight parameters of the trajectory evaluation function are dynamically adjusted, on the basis of which, the linear speed and angular speed corresponding to the optimal trajectory are selected from the speed combination by using the improved traditional dynamic window method, as the speed instruction in the current motion period, so that the optimal path of local planning is obtained.

[0042] In addition, in order to achieve the above object, the application further provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor implements the steps of the mobile robot hybrid path planning method when executing the program.

[0043] The technical scheme provided by the application has the following beneficial effects:

[0044] The application effectively reduces the occurrence frequency of redundant points in the global planning path by expanding the search range to 5*5 and the search direction to 16. By introducing the forward and reverse alternating search mechanism and adding the exponential decay constraint to the cost function, the cost function is adaptively changed, and the unnecessary node return search is avoided. The nonlinear feedback controller is introduced to smooth the global optimal path obtained by the improved A* algorithm, which facilitates the local path planning. Finally, the fuzzy controller is introduced, the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle are input, and the weight coefficients of the three trajectory evaluation functions are output, so that the local trajectory planning of the mobile robot is realized under the dynamic window, and the change of the environment is well adapted. BRIEF DESCRIPTION OF DRAWINGS

[0045] The application will be further described below in combination with the drawings and examples, and the drawings show:

[0046] Figure 1 The technical roadmap of the improved A* algorithm and dynamic window method for the mobile robot hybrid path planning method of the application;

[0047] Figure 2 The overall flowchart of the improved A* algorithm and dynamic window method for the mobile robot hybrid path planning method of the application;

[0048] Figure 3 The schematic diagram of the search range and search direction of the application;

[0049] Figure 4 Flow chart of bidirectional alternate search A* algorithm of the present application;

[0050] Figure 5 Schematic diagram of mixed path planning result of mobile robot of the present application;

[0051] Figure 6 Schematic diagram of structure of electronic device of the present application. DETAILED DESCRIPTION

[0052] In order to have a clearer understanding of the technical features, objectives and effects of the present application, the specific embodiments of the present application will be described in detail with reference to the drawings.

[0053] The embodiment of the present application provides a mixed path planning method for mobile robot by improving A* algorithm and dynamic window method, and a technical route diagram is shown in Figure 1 The mixed path planning for mobile robot is realized by combining the two methods of global optimal path and local optimal path. For the global optimal path, the A* algorithm is improved by expanding the search path and search direction, introducing the forward and backward alternate search algorithm and exponentially decaying weighting the cost function, and the global path is roughly planned, and then the global optimal path is obtained by smoothing the global path through the nonlinear feedback controller. For the local optimal path, the distance from the robot to the target point and the distance from the robot to the nearest obstacle are taken as the inputs of the fuzzy controller to obtain the weight coefficients of the trajectory evaluation function, and the weight coefficients are taken as the inputs of the improved dynamic window method to plan the local optimal path. The two path planning methods are combined to obtain the present application, i.e. the mixed path planning method for mobile robot by improving A* algorithm and dynamic window method, and the final result of the mixed path planning is output.

[0054] According to the analysis of the mixed path planning problem of mobile robot, in order to effectively reduce the number of redundant points in the global planning path, the search range is expanded to 5x5 and the search direction is 16. In order to avoid unnecessary nodes in the back and forth search, the forward and backward alternate search mechanism is introduced, and the cost function is added with exponential decay constraint to make the cost function adaptively change. In order to facilitate local path planning, a nonlinear feedback controller is introduced to smooth the global optimal path obtained by the improved A* algorithm. Finally, in order to adapt to the change of the environment, a fuzzy controller is introduced, the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle are input, and the weight coefficients of three trajectory evaluation functions are output, and the local trajectory planning of the mobile robot is realized under the dynamic window.

[0055] Specifically, the overall flow chart of the mixed path planning method for mobile robot by improving A* algorithm and dynamic window method is as shown in Figure 2As shown, first, a global path from the starting point to the target point is planned by the improved A* algorithm, and the surrounding environment information is obtained in real time by the laser radar. When an obstacle appears, the obstacle information is updated. The local path planning is guided by the global path. When there is no random obstacle in the environment, the improved dynamic window method plans a trajectory along the global path. When a random obstacle appears near the path, the robot is blocked and cannot continue to walk along the global path. At this time, the improved dynamic window method adjusts the weight coefficients of each part in the trajectory evaluation function according to the distance from the robot to the target point and the distance from the robot to the nearest obstacle, plans a local path around the obstacle, and then returns to the global path to continue moving forward along the global path until reaching the target point.

[0056] The specific implementation steps are as follows:

[0057] S1: Obtain the known environment information of the mobile robot workspace, establish the corresponding environment model using the grid method, and determine the positions of the starting point and the target point in the environment model;

[0058] S2: Use the improved A* algorithm to exponentially decay the heuristic function in the cost function, and use the bidirectional alternating search algorithm to perform forward and backward searches under the expanded search neighborhood and search direction until the forward nodes and the backward nodes meet. The forward and backward paths are integrated into a global optimal path from the starting point to the target point;

[0059] S3: Extract the turning point according to the global optimal path, use it as a local target point, combine the fuzzy controller with the dynamic window method, and dynamically adjust the weight proportions of the heading, dist, and vel in the trajectory evaluation function according to the real-time information of the environment where the local target point is located. The expressions of the evaluation functions are as follows:

[0060] heading=180°-θ

[0061]

[0062] vel=|v g |

[0063] where θ represents the included angle between the end point of the robot simulation trajectory and the target point, d represents the distance from the point on the mobile robot simulation trajectory to the nearest obstacle, d max represents the threshold distance from the mobile robot to the obstacle, and v g represents the linear velocity corresponding to the mobile robot simulation trajectory.

[0064] A local path around the obstacle is planned according to the above expressions;

[0065] S4: Use a nonlinear feedback controller to smooth the globally optimal path;

[0066] S5: After bypassing the obstacle, the mobile robot returns to the globally optimal path and continues to move along the globally optimal path until it reaches the current local target point, and then switches to the next local target point;

[0067] S6: Determine whether the current local target point reached by the mobile robot is the global target point. If not, jump to S4; otherwise, execute S7.

[0068] S7: The mobile robot reaches the global target point and completes the path planning task.

[0069] The key points in the above implementation process are as follows:

[0070] 1. A hybrid path planning method combining the improved A* algorithm and the improved dynamic window method is proposed to realize path planning for mobile robots in dynamic obstacle environments.

[0071] 2. To expand the search range and direction of the A* algorithm for mobile robots in global path planning, a strategy of alternating forward and reverse search is proposed, along with an adaptive cost function strategy using an exponentially decaying weighted heuristic function to compensate for the defects of local optima. Simultaneously, based on the kinematic model of the mobile robot, a novel path smoothing method is proposed, introducing a nonlinear feedback controller to smooth the globally optimal path.

[0072] 3. Based on the relationship between the mobile robot's trajectory and the complex environment, a fuzzy rule for dynamically adjusting the weights of the trajectory evaluation function is constructed. A fuzzy controller is introduced to adjust the weights of each evaluation function in the dynamic window in real time, thereby achieving adaptive changes in the environment for local trajectory planning.

[0073] Next, the key technical aspects of an improved hybrid path planning method for mobile robots, combining the A* algorithm and the dynamic window method, will be explained from the following two perspectives.

[0074] I. Improvements to the Global Path Planning Algorithm

[0075] 1. Improvements to search direction and search scope

[0076] The traditional A* algorithm expands the search in the 3*3 neighborhood of the current node, which limits the search range, and searches 4 adjacent nodes each time, which makes the planned path have more turning points and nodes, and reduces the search efficiency. In order to widen the search range and make the search more macro and efficient, the application uses 16 connected modes in the search range of 5*5, expands the search step and search area. At the same time, the 16 search directions are divided into 4 different search areas, and the search area of the node is guided by the target point. The area corresponding to the direction of the connecting line between the target point and the path point is the forward area. With the change of the position of the path point and the target point in the search process, the search area also changes. The schematic diagram of the search range and direction is as shown in Figure 3 The center point represents the current position of the mobile robot, the area occupied by the grid is the search range of the robot, and the arrow points to the search direction of the robot.

[0077] 2. Improvement of search strategy and heuristic function

[0078] In the process of one-way search, the traditional A* algorithm will go back and forth to many unnecessary nodes, which will extend the time-consuming of path planning. In order to improve the search efficiency, the application introduces a bidirectional search strategy, which switches the current target point of the forward search and the reverse search to ensure that the bidirectional search meets near the geometric center of the start point and the target point, so as to alternately replace the best current node. First, start forward search, take S as the start point and G as the target point, and obtain a forward optimal node S; Then, reverse search, take G as the start point and the node obtained by forward search as the target node, and obtain a reverse optimal node G. Repeat iteration until meeting, and the bidirectional alternating search A* algorithm flow chart is as shown in Figure 4As shown, that is: first, the environment model of the mobile robot is constructed and initialized, then the start point and target point are added to the reverse open list and forward open list respectively, then the main loop of the bidirectional alternating search algorithm is entered, in which firstly it is judged whether the forward and reverse open lists are empty, if empty, the search fails, otherwise the minimum cost point in the two open lists is moved to the closed list, and the minimum cost point is set as the current point of each list, then it is judged whether the forward and reverse current nodes meet, if meeting, the forward and reverse paths are integrated and the optimal path is output, otherwise the forward and reverse open lists and the father list are updated according to the 8 adjacent nodes of the current node, and the next round of loop is entered. The proportion of the estimated cost between the forward current node and the reverse current node in the actual path cost in the bidirectional alternating search A* algorithm is related to the position between the forward current node and the reverse current node, and the estimated cost value between the forward current node and the reverse current node is the straight line distance between the two points, which is mostly smaller than the actual path cost. When the forward current node is far away from the reverse current node, the estimated value should be larger, and when the forward current node gradually approaches the reverse current node, the estimated value should be reduced accordingly. For this, the heuristic function in the cost function is weighted in an exponential decay manner, as shown in the following formula:

[0079] f(n) = g(n) + e h(n) ·h(n)

[0080] Wherein, f(n) represents the cost function from the starting node to node n, g(n) represents the dissipation function, that is, the actual cost from the starting node to node n, and h(n) represents the heuristic function, that is, the estimated cost from the starting node to node n.

[0081] 3. Path smoothing strategy

[0082] The path obtained by the traditional A* algorithm after completing path planning has inflection points, although the turning angle of the path planning result of the improved A* algorithm is reduced and the turning number is reduced, but the planned path is still not smooth. Since in actual path planning, the planned path should meet the characteristics of path continuity, path curvature continuity and path easy to be followed by the mobile robot, therefore, the nonlinear feedback controller is adopted to realize path smoothing.

[0083] Let p be the clustering between the current path point and the target point, be the angle between the Y-axis of the current path point vector and the target point vector, and be the angle between the X-axis of the current path point vector and the target point, wherein the Y-axis of the current path point vector is consistent with the direction of the current path point. By inputting the above angle variables, the expression of the speed instruction (v, w) output by the nonlinear feedback controller is:

[0084]

[0085] where K ρ 、 and K φ are the gains of the corresponding variables, and K v is the velocity gain.

[0086] To ensure the smoothness of the path, the nonlinear feedback controller should satisfy the following condition:

[0087]

[0088] After obtaining a series of velocity commands (v, ω), a series of discrete points constituting a smooth path are obtained within an iteration time Δt according to the commands through iterative calculation, and the iterative formula of the node pose (x, y, θ) is as follows:

[0089]

[0090] where x and y are the horizontal and vertical coordinates of the current node, and θ is the direction angle of the current node.

[0091] 2: Improvement of the local path planning algorithm

[0092] Although the traditional dynamic window method takes into account the limitations of the mechanical characteristics of the mobile robot itself and the constraints of the environment on the speed, in the face of complex environments, the constant weight parameter of the trajectory evaluation function leads to poor local path planning results, and there are the following three problems:

[0093] (1) Unable to bypass the obstacles located on the line connecting the mobile robot and the target point;

[0094] (2) Unable to pass through the narrow channel between the obstacles;

[0095] (3) The speed is too fast, causing the mobile robot to miss the target point.

[0096] Therefore, in order to improve the adaptability of the local path planning to the environment, the invention introduces a two-input three-output fuzzy controller with a trapezoidal function and a triangular function as the membership function.

[0097] The input quantity is obtained according to the real-time collected environmental information, that is, the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle. Among them, the universe of discourse of the distance from the mobile robot to the target point is [0, 4], and the fuzzy set is {PS, PM, PB}, which corresponds to the distance from the mobile robot to the target point as near, medium and far, respectively; the universe of discourse of the distance from the mobile robot to the nearest obstacle is [0, 2], and the fuzzy set is {PS, PM, PB}, which corresponds to the distance from the mobile robot to the nearest obstacle as near, medium and far, respectively.

[0098] The output quantity is obtained by using fuzzy logic, i.e. the weights α, β, γ of the target azimuth angle evaluation function, the obstacle gap evaluation function and the speed evaluation function. The domain is [0, 1], and the corresponding fuzzy set is {PS, PM, PB}, which respectively correspond to small, medium and large weights.

[0099] The fuzzy logic is designed according to the following idea:

[0100] (1) When the distance between the mobile robot and the target point and the distance to the nearest obstacle are both greater than 1m, the values of α and γ should be greater than 0.5, and the value of β should be less than 0.5, so that the mobile robot is more inclined to select a trajectory with a large speed and towards the target point, so as to quickly shorten the distance to the target point.

[0101] (2) When the distance between the mobile robot and the target point is greater than 1m, and the distance to the nearest obstacle is less than 1m, the values of α and γ should be less than 0.5, and the value of β should be greater than 0.5, so that the mobile robot is more inclined to select a trajectory with a small speed and away from the obstacle, so as to avoid collision with the obstacle.

[0102] (3) When the distance between the mobile robot and the target point is less than 1m, and the distance to the nearest obstacle is greater than 1m, the values of β and γ should be less than 0.5, and the value of α should be greater than 0.5, so that the mobile robot is more inclined to select a trajectory with a small speed and towards the target point, so as to avoid missing the target point.

[0103] (4) When the distance between the mobile robot and the target point and the distance to the nearest obstacle are both less than 1m, the value of α should be greater than 0.5, the value of γ should be less than 0.5, and the value of β should be taken as the moderate value 0.5, so that the mobile robot is more inclined to select a trajectory with a small speed and towards the target point, while avoiding collision with the obstacle.

[0104] The fuzzy vector is obtained by using Mamdain type fuzzy reasoning method to reason the above fuzzy logic, the output control quantity is obtained by using the barycentric method to solve the fuzzification, and the weight parameters of each part of the trajectory evaluation function are dynamically adjusted. On this basis, the linear speed and angular speed corresponding to the optimal trajectory are selected from the speed combination by using the improved traditional dynamic window method, as the speed command in the current motion period, to obtain the optimal path of local planning.

[0105] The optimal path finally obtained by the mobile robot hybrid path planning method based on the application is shown in Figure 5 , which represents the optimal path of the mobile robot in the grid method environment modeling background of the complex environment.

[0106] As shown in Figure 6As shown, an entity structure diagram of an electronic device is exemplified, which can include: a processor 610, a communications interface 620, a memory 630 and a communications bus 640, wherein the processor 610, the communications interface 620 and the memory 630 complete mutual communication through the communications bus 640. The processor 610 can call the logic instruction in the memory 630 to execute the steps of the mobile robot hybrid path planning method described above, specifically including: S1: obtaining the known environment information of the mobile robot workspace, establishing the corresponding environment model using the grid method, and determining the positions of the starting point and the target point in the environment model; S2: using the improved A* algorithm, exponentially decaying weighting the heuristic function in the cost function, to use the bidirectional alternating search algorithm to perform forward and reverse searches under the expanded search neighborhood and search direction until the forward node and the reverse node meet, and integrate the forward and reverse paths into a global optimal path from the starting point to the target point; S3: extracting the turning point according to the global optimal path, taking it as a local target point, combining the fuzzy controller with the dynamic window method, dynamically adjusting the weight proportion of the target azimuth angle evaluation function, the obstacle gap evaluation function and the speed evaluation function in the trajectory evaluation function according to the real-time information of the environment where the local target point is located, and planning a local path that avoids obstacles; S4: smoothing the global optimal path using a nonlinear feedback controller; S5: after avoiding obstacles, the mobile robot returns to the global optimal path and continues to move along the global optimal path until it reaches the current local target point, and switches to the next local target point; S6: determining whether the current local target point reached by the mobile robot is the global target point, if not, jumping to S4, otherwise, executing S7; S7: the mobile robot reaches the global target point, completing the path planning task.

[0107] In addition, the logic instruction in the memory 630 described above can be implemented in the form of a software function unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0108] It should be noted that in the present application, the terms "comprising", "containing" or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such a process, method, article or system. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0109] The above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. In the unit claims in which several devices are listed, several of these devices can be embodied by the same hardware item. The use of the words first, second, and third does not indicate any order, and these words can be interpreted as identifiers.

[0110] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the present application specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A hybrid path planning method for mobile robots that combines the improved A* algorithm and the dynamic window method, characterized in that, Includes the following steps: S1: Obtain known environmental information of the mobile robot's workspace, establish a corresponding environmental model using the grid method, and determine the positions of the starting point and the target point in the environmental model; S2: Using the improved A* algorithm, the heuristic function in the cost function is exponentially decayed and weighted to utilize the bidirectional alternating search algorithm to perform forward and reverse searches in the expanded search neighborhood and search direction until the forward and reverse nodes meet, integrating the forward and reverse paths into a globally optimal path from the starting point to the target point. S3: Extract turning points based on the global optimal path, use them as local target points, combine the fuzzy controller with the dynamic window method, and dynamically adjust the weight ratio of the target azimuth evaluation function, obstacle gap evaluation function and velocity evaluation function in the trajectory evaluation function according to the real-time information of the environment where the local target point is located, and plan a local path to bypass the obstacles. S4: Use a nonlinear feedback controller to smooth the globally optimal path; S5: After bypassing the obstacle, the mobile robot returns to the globally optimal path and continues to move along the globally optimal path until it reaches the current local target point, and then switches to the next local target point; S6: Determine whether the current local target point reached by the mobile robot is the global target point. If not, jump to S4; otherwise, execute S7. S7: The mobile robot reaches the global target point and completes the path planning task.

2. The mobile robot hybrid path planning method according to claim 1, characterized in that, The improved A* algorithm includes improvements in search direction and search range, as detailed below: Within a 5×5 search range, a 16-connected approach is used to expand the search step size and search area; at the same time, the 16 search directions are divided into 4 different search regions, with the search region of a node guided by the target point, and the region corresponding to the direction of the line connecting the target point and the path point is the forward region. As the location of the waypoints and target points changes during the search process, the search area also changes accordingly.

3. The mobile robot hybrid path planning method according to claim 1, characterized in that, The improved A* algorithm includes improvements to the search strategy, as detailed below: A bidirectional search strategy is introduced, which switches between the current target points of forward search and reverse search to ensure that the bidirectional search meets near the geometric center of the starting point and the target point, thereby alternately changing the best current node. First, a forward search begins, starting with S and targeting G, to obtain a forward optimal node S. Then, a reverse search begins, starting with G and targeting the node obtained in the forward search, to obtain a reverse optimal node G. This process is repeated until the nodes meet and the search is exited.

4. The hybrid path planning method for mobile robots according to claim 1, characterized in that, The improved A* algorithm includes improvements to the heuristic function, as detailed below: The heuristic functions in the cost function are weighted using an exponentially decaying method, as shown in the following equation: f(n)=g(n)+e h(n) ·h(n) Where f(n) represents the cost function from the starting node to node n, g(n) represents the dissipation function, i.e. the actual cost from the starting node to node n, and h(n) represents the heuristic function, i.e. the estimated cost from the starting node to node n.

5. The mobile robot hybrid path planning method according to claim 1, characterized in that, The improved A* algorithm includes: smoothing the globally optimal path using a nonlinear feedback controller, as detailed below: Let ρ be the clustering between the current path point and the target point. φ is the angle between the current path point vector Y-axis and the target point vector, and φ is the angle between the current path point vector X-axis and the target point, where the current path point vector Y-axis is in the same direction as the current path point; By inputting the aforementioned included angle variables, the expression for the output speed command (v, ω) of the nonlinear feedback controller is: Among them, K ρ , and K φ These are the gains of the corresponding variables, K. v Here, v is the current linear velocity, and ω is the current angular velocity; To ensure the smoothness of the path, the nonlinear feedback controller must satisfy the following condition: After receiving a series of velocity commands (v, ω), a series of discrete points constituting a smooth path are obtained through iterative calculation within the iteration time Δt based on these commands. The iterative formula for the node pose (x, y, θ) is as follows: Where x and y are the horizontal and vertical coordinates of the current node, and θ is the direction angle of the current node.

6. The hybrid path planning method for mobile robots according to claim 1, characterized in that, In the local path planning process, a two-input, three-output fuzzy controller with membership functions of trapezoidal and triangular functions is introduced; The input quantities are obtained based on the environmental information collected in real time, namely the distance from the mobile robot to the target point and the distance from the mobile robot to the nearest obstacle; wherein, the universe of discourse for the distance from the mobile robot to the target point is [0,4], and the fuzzy set is {PS,PM,PB}, which correspond to the distances from the mobile robot to the target point being near, medium, and far, respectively; the universe of discourse for the distance from the mobile robot to the nearest obstacle is [0,2], and the fuzzy set is {PS,PM,PB}, which correspond to the distances from the mobile robot to the nearest obstacle being near, medium, and far, respectively. The output is obtained using fuzzy logic, namely the weights α, β, and γ of the target azimuth evaluation function, obstacle gap evaluation function, and velocity evaluation function; the universe of discourse is [0,1], corresponding to the fuzzy set {PS,PM,PB}, which correspond to small, medium, and large weights, respectively.

7. The mobile robot hybrid path planning method according to claim 6, characterized in that, The Mamdain-type fuzzy inference method is used to infer fuzzy logic to obtain fuzzy vectors. The centroid method is used to defuzzify and obtain the output control quantity. The weight parameters of each part of the trajectory evaluation function are dynamically adjusted. Based on this, the improved traditional dynamic window method is used to select the linear velocity and angular velocity corresponding to the optimal trajectory from the velocity combination as the velocity command in the current motion cycle, thus obtaining the optimal path of local planning.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the mobile robot hybrid path planning method as described in any one of claims 1-7.

Images