Mobile robot path planning method and device based on improved RRT algorithm and medium

Abstract

The invention provides a mobile robot path planning method and device based on an improved RRT algorithm and a medium, and belongs to the technical field of mobile robot path planning. The method comprises the steps that a starting point, a target point and obstacle distribution information in a space map are set, and an algorithm is initialized; determining a random sampling point source by using a probability target bias strategy; determining a node closest to the sampling point in the random tree and marking the node as a nearest node; introducing a self-adaptive target gravitation method and a component in a random sampling point direction to determine an expansion direction of a new node; judging whether a connecting line between the new node and the nearest node collides with an obstacle or not, and determining a gravity coefficient; determining an original path from a starting point to a target point, and removing redundant nodes; and performing optimization processing on the path from which the redundant nodes are removed, and outputting an optimized path. According to the method, the path planning time can be greatly shortened, the path is smooth, and the obstacle avoidance path planning efficiency of the mobile robot is further improved.

Description

Technical Field

[0001] The invention relates to a mobile robot path planning method, device and medium based on an improved RRT algorithm, and belongs to the technical field of mobile robot path planning. Background Art

[0002] As a key component in the field of robotics, mobile robots can imitate humans to perceive the surrounding environment, receive instructions from control computers, and have the ability to self-organize, operate independently, and plan paths. With the advancement of technology, the application scope of mobile robots continues to expand, and path planning is one of the key technical challenges in the field of mobile robots. The goal of path planning is to efficiently find a safe and collision-free path for the mobile robot from the starting position to the end position in the face of complex and difficult environments. The requirements of path planning are to achieve goals such as short path length, short time, and efficient obstacle avoidance.

[0003] Among the common path planning methods for mobile robots, the artificial potential field method is prone to fall into local minima, which leads to path planning failure; the A algorithm and the Dijkstra algorithm have high time complexity and are less efficient when facing large-scale or high-precision grid maps. The increase in map size will cause the running time to increase sharply; the RRT algorithm, that is, the rapidly expanding random tree algorithm, is a sampling-based path planning algorithm. It does not require modeling of the map space when used, and the algorithm structure is relatively simple, but there are also problems such as strong randomness when searching for paths, long planned paths, and sometimes the inability to find the ideal path; although the RRT algorithm can optimize the path to a certain extent, the time cost required for its calculation is very high, and the algorithm efficiency is low.

[0004] Therefore, studying how to improve the RRT algorithm to enhance the effect of path planning and make it more efficient and reliable is of great significance to meet the needs of mobile robots in practical applications. Summary of the invention

[0005] The purpose of the present invention is to provide a mobile robot path planning method, device and medium based on an improved RRT algorithm, which improves the path planning efficiency of the mobile robot and optimizes the planned path.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions: A mobile robot path planning method based on an improved RRT algorithm, comprising: Set the starting point, target point, and obstacle distribution information in the spatial map, and initialize the algorithm's random tree T, expansion step size, and target bias probability , initial gravitational coefficient.

[0007] Use a probabilistic target bias strategy to determine the source of random sampling points.

[0008] In the random tree T, the node closest to the sampling point is marked as the nearest node; the adaptive target gravity method and the component of the random sampling point direction are introduced to determine the expansion direction of the new node.

[0009] Determine whether the line between the new node and the nearest node collides with an obstacle. If not, add the new node to the random tree. If so, resample and expand the new node by adjusting the gravity coefficient.

[0010] The above process is iterated to determine an original path from the starting point to the target point, and redundant nodes are removed through the redundant point optimization strategy.

[0011] The particle swarm algorithm is used to optimize the path after removing redundant nodes, and the optimized path is finally output.

[0012] Preferably, the use of the probability target bias strategy to determine the source of random sampling points is specifically as follows: According to the target bias probability and random numbers The size relationship, The value range is (0, 1). If Greater than or equal to , the random sampling point is set to the target point, otherwise the random sampling point is set to a random sampling point generated by multiplying the map size in the x and y dimensions by the low-discrepancy Sobol sequence sampled instead of the pseudo-random sequence in the original algorithm.

[0013] Preferably, an adaptive target gravity method and a component of the random sampling point direction are introduced to determine the expansion direction of the new node, and the calculation formula for the new node is as follows: , in, For the new node, represents the node closest to the sampling point, represents the expansion step size, represents the sampling points generated by the algorithm, represents the gravitational coefficient, Indicates the target point.

[0014] Preferably, if there is a collision, the gravity coefficient is adjusted as follows: , in, represents the initial gravitational coefficient, Represents a random number in the range of 0 to 1. Indicates the distance from the new node to the center of the current obstacle, express The sum of the distance from the center of the obstacle that is currently colliding and the radius of the current obstacle. Indicates dynamic adjustment of the threshold. Indicates the radius of the obstacle.

[0015] Preferably, the The distance from the center of the obstacle that currently collides and the current obstacle radius The distance and The calculation is as follows: ,in, represents the Euclidean distance.

[0016] The dynamically adjusted threshold The specific formula is as follows: , in, represents the base of the exponential function, Indicates the radius of the obstacle.

[0017] Preferably, the optimization processing method of using the particle swarm algorithm to optimize the path after removing the redundant nodes is as follows: The starting point of the path after removing redundant nodes is set as the temporary starting point. The second point and the third point are selected from the path after removing redundant points as the middle point and the temporary target point respectively. A group of particles are randomly generated in the surrounding area of ​​the middle point, and each particle symbolizes a potential path point.

[0018] Calculate the path fitness score between the temporary starting point and the temporary target point based on the position of the particle The size of the fitness function is determined by two factors: the length of the path and whether it intersects with obstacles. The formula is as follows: , in, For two consecutive path points and The Euclidean distance between Indicates the number of points on the path.

[0019] Path collision detection is performed during the process of updating the particle position. If it intersects with an obstacle, the fitness value is added with a penalty value. .

[0020] After a predetermined number of iterations, the point on the determined shortest path is set as a new temporary starting point, and the previous temporary end point is converted to a new intermediate point; the above steps are repeated until the temporary target point is converted to the real target point.

[0021] A mobile robot path planning device based on an improved RRT algorithm comprises a processor and a memory storing program instructions, wherein the processor is configured to execute the mobile robot path planning method based on the improved RRT algorithm when running the program instructions.

[0022] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the mobile robot path planning method based on the improved RRT algorithm.

[0023] The advantages of the present invention are as follows: the present invention significantly improves the quality of random sampling points by adopting a probabilistic target bias strategy in combination with a low-discrepancy sequence to generate sampling points; at the same time, an adaptive target gravity method is introduced to make the expansion of new nodes more target-oriented, and the gravity coefficient can be adjusted according to actual conditions, which is conducive to the random tree to better avoid obstacles.

[0024] The present invention performs pruning optimization on the feasible path found, removes redundant nodes on the path, and simplifies the path. Subsequently, the simplified path is further optimized using the PSO algorithm, and the path is smoothed using the cubic B-spline curve method. These measures further improve the fluency and smoothness of the path.

[0025] The present invention has strong goal orientation and can greatly shorten the path planning time. At the same time, the path is smooth and fluent, which further improves the obstacle avoidance path planning efficiency of the mobile robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0027] Figure 1 It is a schematic diagram of the process of the present invention.

[0028] Figure 2 Schematic diagram of the algorithm flow of the present invention.

[0029] Figure 3 This is the improved new node expansion graph of the present invention.

[0030] Figure 4 This is a schematic diagram of removing redundant nodes according to the present invention.

[0031] Figure 5 This is a schematic diagram of the optimization path of the PSO algorithm of the present invention.

[0032] Figure 6 Graph of running simulation for the basic RRT algorithm.

[0033] Figure 7Graph of running simulation for the Goal-RRT algorithm.

[0034] Figure 8 Run simulation graph for the RRT* algorithm.

[0035] Fig. 9 Figure 2. Simulation diagram of running a simulation to improve the RRT algorithm. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] Example 1 like Figure 1 , Figure 2 As shown, a mobile robot path planning method based on an improved RRT algorithm includes: S1: Set the starting point, target point, and obstacle distribution information in the space map, and initialize the algorithm's random tree, expansion step size, target bias probability, and initial gravity coefficient.

[0038] S2: Use a probabilistic target bias strategy to determine the source of random sampling points.

[0039] S3: Determine in the random tree T that the node closest to the sampling point is marked as the nearest node; introduce the adaptive target gravity method and the component of the random sampling point direction to determine the expansion direction of the new node.

[0040] S4: Determine whether the line between the new node and the nearest node collides with an obstacle. If not, add the new node to the random tree. If so, resample and expand the new node by adjusting the gravity coefficient.

[0041] S5: Iterate the above process to determine an original path from the starting point to the target point, and remove redundant nodes through the redundant point optimization strategy, such as Figure 4 shown.

[0042] S6: The path after removing redundant nodes is optimized using a particle swarm algorithm, and the optimized path is finally output.

[0043] As a refinement of the above embodiment, a probability target bias strategy is used to determine the source of each sampling point of the algorithm, that is, according to the target bias probability and random numbers The size relationship of (value range (0, 1)) determines the random sampling points of the algorithm Set as target point Or generated by low-discrepancy sequence (sobol) The optimized sampling rule expression is as follows: The use of the probability target bias strategy to determine the source of random sampling points is as follows: According to the target bias probability and random numbers The size relationship, The value range is (0, 1). If Greater than or equal to , the random sampling point is set to the target point, otherwise the random sampling point is set to a random sampling point generated by multiplying the map size in the x and y dimensions by the low-discrepancy Sobol sequence sampled instead of the pseudo-random sequence in the original algorithm.

[0044] The formula is as follows: .

[0045] The basic RRT algorithm uses a pseudo-random sequence to generate random sampling points, which may produce some repeated or clustered sampling points, resulting in some invalid sampling situations, and the length cost of the path planning obtained will also increase. The low-discrepancy sequence Sobol is more evenly distributed in a specific dimensional space. Compared with the pseudo-random sequence, its advantage lies in the feature of full coverage. Therefore, this scheme uses the Sobol sequence to generate random sampling points. The Sobol sequence is based on the direction number. A random sequence based on (0< <1), the application coefficient is , the order is The polynomial Number of generated directions , the polynomial is: , in, represents a polynomial variable.

[0046] Number of directions for: , Among them, when You need to specify When hour The recursive expression of is as follows: , In the above formula, 0< < ,and is an odd number, ⊕ represents the XOR operation in binary, where the parameter , , All of them are known and given. Substituting these known parameters into the following formula, we can calculate : Combining the above formulas, the generated Sobol sequence is The number is: In the above formula , , yes The obtained Sobol sequence is evenly distributed in the (0,1) space and is used instead of the pseudo-random sequence used in the original algorithm.

[0047] As a refinement of the above embodiment, in the random tree, the sampling points are determined The node closest to the nearest node is marked as the nearest node The adaptive target gravity method is introduced to jointly determine the expansion direction of the new node with the component of the random sampling point direction obtained by sampling. The calculation formula is as follows: , in, New node, represents the node closest to the sampling point, represents the expansion step size, represents the sampling points generated by the algorithm, represents the gravitational coefficient, Indicates the target point.

[0048] The schematic diagram of the expansion node is as follows Figure 3 shown.

[0049] As a refinement of the above embodiment, after obtaining the new node After that, you need to go through and The conditional judgment of whether the connection line collides with an obstacle. If it does not collide with an obstacle, the node can be used as a new node for expansion and added to the random tree. Otherwise, the initial gravity coefficient Multiply by the attenuation factor to adjust the gravity coefficient and then resample and expand the new node. First calculate The distance from the center of the obstacle that currently collides With the current obstacle radius The distance and : , represents the Euclidean distance; The dynamically adjusted threshold The specific formula is as follows: , in, represents the base of the exponential function, Indicates the radius of the obstacle.

[0050] The dynamically adjusted threshold The specific formula of the combined attenuation function is introduced as follows: , in, represents the base of the exponential function, Indicates the radius of the obstacle.

[0051] New nodes also need to be calculated Distance to the center of the current obstacle ,if Greater than , then the variation law of the gravitational coefficient is ,in is a random number between 0 and 1. Less than But greater than , then the variation law of the gravitational coefficient is .if Less than , then the variation law of the gravitational coefficient is Specifically, the gravity coefficient adjustment can be expressed as follows: , in, represents the initial gravitational coefficient, Represents a random number in the range of 0 to 1. Indicates the distance from the new node to the center of the current obstacle, express The sum of the distance from the center of the obstacle that is currently colliding and the radius of the current obstacle. Indicates dynamic adjustment of the threshold. Indicates the radius of the obstacle.

[0052] As a refinement of the above embodiment, the schematic diagram of the PSO algorithm optimization path is as follows: Figure 5 As shown, the method of using the particle swarm algorithm to optimize the path after removing redundant nodes is as follows: The starting point is set as the temporary starting point, and the second and third points are selected from the path without redundant points as the middle point and the temporary target point respectively. A group of particles are randomly generated in the surrounding area of ​​the middle point, and each particle represents a potential path point.

[0053] According to the position of the particle, the path fitness score between the temporary starting point and the temporary target point is calculated. The lower the score, the better the path. The size of the fitness function is determined by two factors: the length of the path and whether it intersects with obstacles.

[0054] , in, For two consecutive path points and The Euclidean distance between Indicates the number of points on the path.

[0055] In the process of updating the particle position, path collision detection is performed to ensure that all particles avoid obstacles and stay in a safe area. If they intersect with an obstacle, a large penalty value is added to the fitness value. , so that the score becomes larger.

[0056] After a predetermined number of iterations, the point on the determined shortest path is set as a new temporary starting point, and the previous temporary end point is converted to a new intermediate point. Repeat the above steps until the temporary target point is converted to the real target point, at which point the optimization process is completed. Finally, the optimized path is output to ensure that the path maintains its original continuity while minimizing collisions with obstacles. The path is then optimized using the cubic B-spline curve method.

[0057] In order to evaluate the performance and applicability of the proposed improved algorithm, we conducted experiments on the MATLAB simulation platform. In the experiment, we simulated the basic RRT algorithm, the target probability biased RRT algorithm, the RRT* algorithm, and the improved algorithm proposed in this paper in a two-dimensional map environment. The simulation runs of each algorithm are as follows: Figure 6-Figure 9 As shown, 50 independent simulation experiments were recorded and the average values ​​were listed in Table 1.

[0058] Table 1. Operation data of obstacle avoidance path planning of each algorithm Example 2 The disclosed embodiment also provides a mobile robot path planning device based on an improved RRT algorithm, including a processor and a memory. Optionally, the device may also include a communication interface and a bus. The processor, the communication interface, and the memory may communicate with each other through the bus. The communication interface may be used for information transmission. The processor may call the logic instructions in the memory to execute the mobile robot path planning method based on the improved RRT algorithm of the above embodiment.

[0059] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0060] The memory, as a computer-readable storage medium, can be used to store software programs and computer executable programs, such as program instructions / modules corresponding to the method in the embodiment of the present disclosure. The processor executes the function application and data processing by running the program instructions / modules stored in the memory, that is, the mobile robot path planning method based on the improved RRT algorithm in the above embodiment is implemented.

[0061] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory may include a high-speed random access memory and may also include a non-volatile memory.

[0062] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned mobile robot path planning method based on the improved RRT algorithm.

[0063] The computer-readable storage medium mentioned above may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0064] The technical solution of the embodiment of the present disclosure can be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in the embodiment of the present disclosure. The aforementioned storage medium may be a non-transient storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, and other media that can store program codes, or a transient storage medium.

[0065] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A mobile robot path planning method based on an improved RRT algorithm, characterized in that: include: Set the starting point, target point, and obstacle distribution information in the spatial map, and initialize the algorithm's random tree T, expansion step size, and target bias probability , initial gravitational coefficient; Use the probability target bias strategy to determine the source of random sampling points; In the random tree T, the node closest to the sampling point is marked as the nearest node; the adaptive target gravity method and the component of the random sampling point direction are introduced to determine the expansion direction of the new node; Determine whether the line between the new node and the nearest node collides with an obstacle. If not, add the new node to the random tree. If so, resample and expand the new node by adjusting the gravity coefficient. Iterate the above process to determine an original path from the starting point to the target point, and remove redundant nodes through the redundant point optimization strategy; The particle swarm algorithm is used to optimize the path after removing redundant nodes, and the optimized path is finally output.

2. The mobile robot path planning method based on the improved RRT algorithm according to claim 1 is characterized in that: The use of the probability target bias strategy to determine the source of random sampling points is as follows: According to the target bias probability and random numbers The size relationship, The value range is (0, 1). If Greater than or equal to , the random sampling point is set to the target point, otherwise the random sampling point is set to a random sampling point generated by multiplying the map size in the x and y dimensions by the low-discrepancy Sobol sequence sampled instead of the pseudo-random sequence in the original algorithm.

3. The mobile robot path planning method based on the improved RRT algorithm according to claim 2 is characterized in that: The adaptive target gravity method and the component of the random sampling point direction are introduced to determine the expansion direction of the new node. The calculation formula of the new node is as follows: ， in, For the new node, represents the node closest to the sampling point, represents the expansion step size, represents the sampling points generated by the algorithm, represents the gravitational coefficient, Indicates the target point.

4. The mobile robot path planning method based on the improved RRT algorithm according to claim 3 is characterized in that: In case of collision, the gravity coefficient is adjusted as follows: ， in, represents the initial gravitational coefficient, Represents a random number in the range of 0 to 1. Indicates the distance from the new node to the center of the current obstacle, express The sum of the distance from the center of the obstacle that is currently colliding and the radius of the current obstacle. Indicates dynamic adjustment of the threshold. Indicates the radius of the obstacle.

5. The mobile robot path planning method based on the improved RRT algorithm according to claim 4 is characterized in that: Said The distance from the center of the obstacle that currently collides and the current obstacle radius The distance and The calculation is as follows: ,in, represents the Euclidean distance; The dynamically adjusted threshold The specific formula is as follows: ， in, represents the base of the exponential function, Indicates the radius of the obstacle.

6. The mobile robot path planning method based on the improved RRT algorithm according to claim 4 is characterized in that: The specific method of using the particle swarm algorithm to optimize the path after removing redundant nodes is as follows: The starting point of the path after removing redundant nodes is set as the temporary starting point. The second and third points are selected from the path after removing redundant points, respectively, as the middle point and the temporary target point. A group of particles are randomly generated in the surrounding area of ​​the middle point, and each particle represents a potential path point. The path fitness score between the temporary starting point and the temporary target point is calculated based on the position of the particle; the size of the fitness function is determined by two factors: the path length and whether it intersects with obstacles. The formula is as follows: ， in, For two consecutive path points and The Euclidean distance between Indicates the number of points on the path. Path collision detection is performed during the process of updating the particle position. If it intersects with an obstacle, the fitness value is added with a penalty value. ; After a predetermined number of iterations, the point on the determined shortest path is set as a new temporary starting point, and the previous temporary end point is converted to a new intermediate point; the above steps are repeated until the temporary target point is converted to the real target point.

7. A mobile robot path planning device based on an improved RRT algorithm, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the mobile robot path planning method based on the improved RRT algorithm as described in any one of claims 1-6 when running the program instructions.

8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the method described in any one of claims 1 to 6 is implemented.

Images