Mobile robot path planning method based on improved ant colony algorithm

By improving the ant colony algorithm, combining K-medoids clustering and adaptive heuristic functions, the problems of poor path feasibility and low efficiency of traditional ant colony algorithm in path planning are solved, and faster and higher-quality path planning is achieved.

CN114964261BActive Publication Date: 2024-11-01JIANGNAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210563301.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-20
Publication Date
2024-11-01
Estimated Expiration
2042-05-20

AI Technical Summary

Technical Problem

Traditional ant colony algorithms have problems such as poor path feasibility, easy to fall into deadlocks, and low search efficiency in mobile robot path planning, which cannot meet the real-time navigation needs.

Method used

The path planning method based on the improved ant colony algorithm is adopted, and the map obstacles are clustered through the K-medoids algorithm, the initial pheromone concentration is differentiated, and an adaptive heuristic function and a two-way search mechanism are introduced, combining the marker fallback strategy and pheromone. Reward and punishment strategy, optimize node transfer probability and pheromone update formula to ensure path feasibility and efficiency.

Benefits of technology

It significantly improves the efficiency and accuracy of path planning, shortens the shortest path length, reduces the number of turns and convergence algebra, and improves the quality and feasibility of paths.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114964261B_ABST
    Figure CN114964261B_ABST
Patent Text Reader

Abstract

The present invention discloses a mobile robot path planning method based on an improved ant colony algorithm, and belongs to the field of mobile robot control. The present invention first performs cluster analysis on map obstacles through a K‑medoids algorithm, differentiates the initial pheromone concentration, and improves the early convergence speed; then improves the ant colony algorithm by introducing an adaptive heuristic function, appropriately amplifying the heuristic factor according to the different positions of the nodes to be visited, so as to speed up the search for the target point; a two-way search mechanism is set to convert the problem of finding the target point into a problem of two ants meeting, thereby improving the efficiency of the algorithm and thus improving the speed of path planning. Simulation data shows that compared with the existing path planning scheme, the shortest path length of the robot is shortened by 5%, the number of turns in the process is reduced by 30%, and the convergence algebra is reduced by 46.2%. Therefore, the present invention can find the shortest path of the robot more quickly, and the path quality is better.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a mobile robot path planning method based on an improved ant colony algorithm, and belongs to the field of mobile robot control. Background Art

[0002] In recent years, with the rapid increase in online shopping transactions, the pressure on the express logistics sector has gradually reached saturation. The concept of "smart logistics" aims to improve the accuracy and efficiency of logistics sorting, transportation, and distribution through technologies such as the Internet of Things, robots, and 5G communications. In this context, mobile robots for smart warehousing are a key element in the industry's move towards automation.

[0003] As one of the core technologies for mobile robots to achieve autonomous navigation, path planning is an important research topic in the field of robotics. It refers to the mobile robot finding the shortest feasible path from the starting point to the end point in the working scene according to the established constraints, and the process is safe and collision-free. At present, the algorithms for solving the global path planning problem are mainly divided into two categories, including traditional algorithms such as depth-first search, A* algorithm, Dijkstra algorithm, and swarm intelligence optimization algorithms such as ant colony algorithm, particle swarm algorithm, and simulated annealing algorithm.

[0004] The ant colony algorithm is a heuristic algorithm that seeks the optimal solution to a problem through random search. It is suitable for solving the path planning problem of mobile robots due to its positive feedback mechanism, parallel search, strong robustness, and easy scalability. However, the traditional ant colony algorithm has shortcomings such as slow convergence speed in the early stage, easy to fall into local extreme values, many inflection points, and poor accuracy. When solving the path planning problem of mobile robots, the efficiency and stability of the algorithm cannot meet the requirements.

[0005] Therefore, many scholars have made relevant improvements to the traditional ant colony algorithm based on the above practical problems. Ma Xianghua et al. (Ma Xianghua, Zhang Qian. Research on improved ant colony algorithm for robot path planning [J]. Computer Engineering and Applications, 2021, 57(5):6.) proposed an improved ant colony algorithm for robot path planning. The initial pheromone matrix is ​​established based on the pre-planned path to avoid blind search in the early stage of the algorithm; the improved ant colony algorithm and the A* algorithm are organically integrated to further improve the search directionality and convergence speed of the ant colony algorithm. The improved algorithm has the following limitations:

[0006] 1. To reduce the blindness of the initial search, connect the starting point and the target point, and increase the initial pheromone concentration near the straight line. This approach does not take into account the distribution of obstacles in the map, and may cause a large number of ants to be deadlocked;

[0007] Second, referring to the A* algorithm, the estimated cost from the current node to the target point is introduced into the heuristic function of the traditional ant colony algorithm, which improves the convergence speed to a certain extent, but the efficiency of the algorithm still needs to be improved;

[0008] Third, the oblique direction is retained when turning, which will affect the accuracy when the robot is actually running.

[0009] In summary, although the improved ant colony algorithm can solve the path planning problem of the robot and has certain improvements in convergence speed and search efficiency, it also has problems such as easy deadlock and poor path feasibility, and the algorithm efficiency still needs to be further improved. Summary of the invention

[0010] In order to solve the problems of poor path feasibility, easy deadlock, low search efficiency, etc. in the current path planning method, the present invention provides a mobile robot path planning method based on an improved ant colony algorithm, the method comprising:

[0011] Step 1: Consider the working environment of the mobile robot as a two-dimensional plane coordinate system, and construct the working environment into a static grid map using a grid method;

[0012] Step 2: The static grid map is preprocessed using a clustering algorithm. The map is divided into a number of cluster partitions. The dredging value and the congestion value are defined to quantify the complexity of the regional environment and differentiate the pheromone concentration.

[0013] Step 3: Initialize various parameters, including: population size M, pheromone weight α, heuristic function weight β, pheromone volatility coefficient ρ, initialize the taboo table Tabu, and set the maximum number of iterations K;

[0014] Step 4: Place M ants at the starting point S and the end point E of the working environment of the mobile robot respectively, and start from the m=1th pair of ants to check whether the walking paths of the pair of ants have an intersection. If not, transfer to the next node according to the node transfer probability formula, and add the previous node to the respective taboo tables; if so, jump to step 5;

[0015] Step 5: Get a path solution based on the walking path of the pair of ants and calculate the path length; determine whether the path finding of M pairs of ants has been completed. If m = M, find the optimal solution and the worst solution of the current iteration; otherwise, m = m + 1, go to step 4, and the next pair of ants starts path finding again;

[0016] Step 6: At the end of each iteration, update the node pheromone concentration, the number of iterations k = k + 1; when k = K, all iterations are completed, and the global optimal path solution and the corresponding shortest path length are output as the optimal path of the mobile robot.

[0017] Optionally, the node transfer probability formula in step 4 is:

[0018]

[0019] in, represents the node transition probability; τ ab (t) is the pheromone concentration between nodes a and b; η ab (t) is the heuristic function from node a to node b; α and β represent the pheromone concentration and the weight of the heuristic function respectively; τ as (t) is the pheromone concentration between nodes a and s, η as (t) is the heuristic function between nodes b and s, and allow represents the set of nodes that the ant is allowed to visit next.

[0020] Optionally, the heuristic function η ab (t) is:

[0021]

[0022] Among them, H ab is the Manhattan distance from node a to node b, H bE is the Manhattan distance from node b to the end point E, a x 、a y is the horizontal and vertical coordinates of node a, b x and b y is the horizontal and vertical coordinates of node b.

[0023] Optionally, the node transfer strategy in step 4 further includes a marking rollback strategy, and the marking rollback strategy is:

[0024] When the ant reaches node B, the set of nodes allowed to be visited next is empty, and it falls into a deadlock. At this time, node B is marked as an obstacle and it falls back to node A to prevent subsequent ants from entering the trap again and save path-finding resources.

[0025] Optionally, the method for updating the node pheromone concentration includes: introducing a reward and punishment strategy and introducing an inflection point evaluation function;

[0026] The reward and punishment strategy introduced is: at the end of each iteration, only the path solution that finds the target point is retained, the pheromone of the path node that exceeds the pseudo-global optimal solution is increased, and the pheromone concentration of the worst solution among all current path solutions is reduced. The node pheromone concentration formula is:

[0027]

[0028]

[0029] Among them, Q is the initial value of pheromone, iter_best is the optimal solution in this iteration, iter_worst is the worst solution in this iteration, best is the pseudo global optimal solution; t represents the tth moment, that is, the tth generation of ants, ρ is the pheromone volatility coefficient;

[0030] The inflection point evaluation function defines the corresponding extended path length according to the different turning angles each time, and the formula is as follows:

[0031]

[0032] Where θ is the turning angle and f(θ) represents the extended path length.

[0033] Optionally, the step 2 uses a K-medoids algorithm to cluster and analyze map obstacles, delineates map partitions based on the center points of each cluster, and uses unblocked value and congestion value indicators to represent the complexity of the regional environment;

[0034] The formula for determining the sum of dissimilarity of cluster centers is as follows:

[0035]

[0036] Among them, TC ih Represents the center point O i By non-central point O h The total cost after replacement, C jih Represents the center point O i By non-central point O h After substitution, the non-central point O j the cost;

[0037] Define the dredge value P i0 Describe the number of grids that can be passed through in the partition and define the congestion value P i Describe the density of obstacles in the partition. The formula is as follows:

[0038] P i0 =N i0 / S i

[0039] P i =N i / S i

[0040] Where N i0 Represents partition Y i The number of internal grids that can be moved straight, N i Represents partition Y i The number of internal obstacle grids, S i Indicates the total number of grids in the partition;

[0041] The dredging value P of each cluster partitioni0 and congestion value P i , define the environment complexity function P(u), the formula is as follows:

[0042] P(u)=exp(P i0 -P i ).

[0043] Optionally, the process of obtaining a path solution according to the walking path of a pair of ants in step 4 and step 5 includes:

[0044] In each iteration, a group of ants is set at the starting point and the end point of the mobile robot's working environment. Each ant at the starting point and the end point is denoted as S k (k=1,2,…m) and E k (k=1,2,…m), select one ant from each of the starting point and the end point to search for the path, and each pair of ants is recorded as (S k ,E k ), where the starting ant searches forward from the starting point, and the ending ant searches backward from the ending point;

[0045] Let Path(S k ) is ant S k The set of path nodes passed, Path(E k ) is ant E k The set of path nodes passed, J(S k ,E k ) is the intersection of the nodes that the pair of ants pass through;

[0046] When J(S k ,E k ) When the set is not empty, it means that the two ants meet on the way, that is, a feasible path is found, and the search stops; the actual path solution Path is the ant S k The node order of ant E k The nodes are connected in reverse order, and the formula is as follows:

[0047] J(S k ,E k )=Path(S k )∩Path(E k )

[0048] Path=Path(S k )+fliplr(Path(E k ))

[0049] Among them, fliplr(*) means left-right flip matrix.

[0050] Optionally, in step 1, constructing the working environment into a static grid map using a grid method includes:

[0051] All grids are numbered in order from left to right and from top to bottom. The expressions of grid numbers and actual coordinates are as follows:

[0052]

[0053] Among them, i is the grid number, i x is the horizontal coordinate of the grid, i y is the grid ordinate, mapLen is the number of rows and columns of the grid map, mod is the remainder operator, and ceil is the rounding-up operator.

[0054] The present invention also provides a mobile robot path planning device, comprising: a memory and a processor coupled to the memory, wherein the processor is configured to: execute the above-mentioned mobile robot path planning method based on the improved ant colony algorithm based on instructions stored in the memory.

[0055] The present invention also provides a computer-readable storage medium, on which a program is stored. When the program is executed by a processor, the above-mentioned mobile robot path planning method based on the improved ant colony algorithm is implemented.

[0056] The beneficial effects of the present invention are:

[0057] The mobile robot path planning method of the present invention first uses the K-medoids algorithm to cluster and analyze map obstacles, differentiates the initial pheromone concentration, and effectively solves the problem of slow convergence speed in the early stage of path planning; then improves the ant colony algorithm by introducing an adaptive heuristic function, appropriately amplifying the heuristic factor according to the different positions of the nodes to be visited, so as to speed up the search for the target point; sets a bidirectional search mechanism, transforms the problem of finding the target point into a problem of two ants meeting, and improves the efficiency of the algorithm; stipulates that ants can only move horizontally and vertically, so that the accuracy is guaranteed and the path feasibility is good.

[0058] The present invention utilizes the improved ant colony algorithm to perform path planning for a mobile robot. Simulation experiments have shown that, compared with the existing path planning scheme utilizing the ant colony algorithm, the path planning method of the present invention shortens the robot's shortest path length by 5%, reduces the number of turns in the process by 30%, and reduces the convergence algebra by 46.2%. Therefore, the present invention can find the shortest path for the robot more quickly, and the path quality is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 This is a schematic diagram of the traditional ant colony algorithm path planning.

[0061] Figure 2 This is a schematic diagram of the existing improved ant colony algorithm path planning.

[0062] Figure 3 It is a cluster partition map after the grid map is preprocessed by the K-medoids clustering algorithm of the second embodiment of the present invention.

[0063] Figure 4 It is a schematic diagram of the improved bidirectional ant colony algorithm path planning according to the second embodiment of the present invention.

[0064] Figure 5 Comparison of iterative convergence curves of the three algorithms. DETAILED DESCRIPTION

[0065] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0066] Embodiment 1:

[0067] This embodiment provides a mobile robot path planning method based on an improved ant colony algorithm, the method comprising:

[0068] Step 1: Consider the working environment of the mobile robot as a two-dimensional plane coordinate system, and construct the working environment into a static grid map using a grid method;

[0069] Step 2: The static grid map is preprocessed using a clustering algorithm. The map is divided into a number of cluster partitions. The dredging value and the congestion value are defined to quantify the complexity of the regional environment and differentiate the pheromone concentration.

[0070] Step 3: Initialize various parameters, including: population size M, pheromone weight α, heuristic function weight β, pheromone volatility coefficient ρ, initialize the taboo table Tabu, and set the maximum number of iterations K;

[0071] Step 4: Place M ants at the starting point S and the end point E of the working environment of the mobile robot respectively, and start from the m=1th pair of ants to check whether the walking paths of the pair of ants have an intersection. If not, transfer to the next node according to the node transfer probability formula, and add the previous node to the respective taboo tables; if so, jump to step 5;

[0072] Step 5: Get a path solution based on the walking path of the pair of ants and calculate the path length; determine whether the path finding of M pairs of ants has been completed. If m = M, find the optimal solution and the worst solution of the current iteration; otherwise, m = m + 1, go to step 4, and the next pair of ants starts path finding again;

[0073] Step 6: At the end of each iteration, update the node pheromone concentration, the number of iterations k = k + 1; when k = K, all iterations are completed, and the global optimal path solution and the corresponding shortest path length are output as the optimal path of the mobile robot.

[0074] Embodiment 2:

[0075] This embodiment provides a mobile robot path planning method based on an improved ant colony algorithm, which is mainly divided into the following parts.

[0076] 1. Environmental modeling and preprocessing

[0077] (1) Environmental Modeling

[0078] The warehouse logistics center where the mobile robot is located can be regarded as a two-dimensional plane coordinate system. The working environment is constructed as a static grid map using the grid method. The original obstacles are irregular in shape, so they are expanded and unified into squares. All grids are numbered in order from left to right and from top to bottom. The expression of grid number and actual coordinates is as follows:

[0079]

[0080] In the formula, i is the grid number, i x is the horizontal coordinate of the grid, i y is the grid ordinate, mapLen is the number of rows and columns of the grid map, mod is the remainder operator, and ceil is the rounding-up operator.

[0081] (2) Map preprocessing

[0082] The K-medoids algorithm is used to cluster the obstacles on the map, and the map partitions are delineated according to the center points of each cluster, and the complexity of the regional environment is represented by the indicators of dredging value and congestion value. The formula for determining the sum of the dissimilarity of the cluster center points is as follows:

[0083]

[0084] Where TC ih Represents the center point O i By non-central point O h The total cost after replacement, C jih Represents the center point O i By non-central point O h After substitution, the non-central point O j The cost.

[0085] Define the dredge value P i0 Describe the number of grids that can be passed through in the partition and define the congestion value P i Describe the density of obstacles in the partition. The formula is as follows:

[0086] P i0 =N i0 / S i

[0087] P i =N i / S i

[0088] Where N i0 Represents partition Y i The number of internal grids that can be moved straight, N i Represents partition Y i The number of internal obstacle grids, S i Indicates the total number of rasters in the partition.

[0089] The environmental complexity function P(u) is defined by the dredging value and congestion value of each cluster partition. The formula is as follows. It is an important indicator for initializing the pheromone of each node in the map. The larger the P(u) value, the fewer obstacles there are in the area where the node u is located, and the less cost the mobile robot spends to pass through the area. Therefore, the initial pheromone concentration can be appropriately increased.

[0090] P(u)=exp(P i0 -P i )

[0091] 2. Bidirectional search mechanism

[0092] Each iteration sets a group of ants at the starting point and the end point, and each ant is denoted as S k (k=1,2,…m) and E k (k=1,2,…m), select one ant from each group S and E to search for a path, and each pair of ants is recorded as (S k ,E k ). The S group of ants searches forward from the starting point, and the E group of ants searches backward from the target point.

[0093] Let Path(S k ) is ant S kThe set of path nodes passed, Path(E k ) is ant E k The set of path nodes passed, J(S k ,E k ) is the intersection of the nodes that the pair of ants pass through. k ,E k )When the set is not empty, it means that the two ants meet on the way, that is, a feasible path is found, and the search stops. The actual path solution Path is the ant S k The node order of ant E k The nodes are connected in reverse order, and the formula is as follows.

[0094] J(S k ,E k )=Path(S k )∩Path(E k )

[0095] Path=Path(S k )+fliplr(Path(E k ))

[0096] 3. Improved node transfer probability formula

[0097] The node transfer probability is the core formula of the ant colony algorithm. Each ant uses it to select the location of the next node. The specific formula is as follows. represents the node transition probability; τ ab (t) is the pheromone concentration between nodes a and b; η ab (t) is the heuristic function from node a to node b; α and β represent the pheromone concentration and the weight of the heuristic function respectively, τ as (t) is the pheromone concentration between nodes a and s, η as (t) is the heuristic function between nodes b and s, and allow represents the set of nodes that the ant is allowed to visit next.

[0098]

[0099] (1) Introducing adaptive heuristic function

[0100] In order to increase the transition probability gap between the nodes to be selected, an adaptive heuristic function is introduced. The heuristic factor is appropriately amplified according to the different positions of the nodes to be selected to speed up the search for the target point. The formula is as follows:

[0101]

[0102] In the formula, H ab is the Manhattan distance from node a to node b, H bEis the Manhattan distance from node b to the end point E, a x 、a y is the horizontal and vertical coordinates of node a, b x and b y is the horizontal and vertical coordinates of node b.

[0103] (2) Deadlock mark rollback processing

[0104] If there are U-shaped obstacles in the map environment, some ants will fall into deadlock during the pathfinding process, so a marking and retreating strategy is designed. When the ant reaches node B, the set of candidate nodes is empty and it falls into deadlock. At this time, marking node B as an obstacle and retreating to node A can prevent subsequent ants from entering the trap again and save pathfinding resources.

[0105] 4. Improved pheromone update formula

[0106] (1) Pheromone reward and punishment strategy

[0107] The concentration of pheromones has a crucial impact on the performance of the ant colony algorithm. When ants are searching for a path, they release pheromones to mark the better path solution. At the same time, as the pheromones evaporate, after several iterations, the next generation of ants gradually tend to the optimal path with the help of pheromones.

[0108] In order to avoid the algorithm from falling into the local optimum and improve the convergence speed, this paper introduces a reward and punishment strategy for pheromone concentration, the formula is as follows. At the end of each iteration, only the path solution that finds the target point is retained, the pheromone of the path node that exceeds the pseudo-global optimal solution is increased, and the pheromone concentration of the worst solution among all contemporary path solutions is reduced.

[0109]

[0110]

[0111] Where Q is the initial value of pheromone, iter_best is the optimal solution in this iteration, iter_worst is the worst solution in this iteration, and best is the pseudo-global optimal solution.

[0112] (2) Introducing the inflection point evaluation function

[0113] Since the number and angle of turns of the mobile robot in actual working scenarios will affect the cost of the driving process, the present invention introduces an inflection point evaluation function and defines the corresponding extended path length according to the different turning angles each time. The formula is as follows. 45° oblique turning cannot guarantee driving accuracy, so it is not considered.

[0114]

[0115] Where θ is the turning angle and f(θ) represents the extended path length.

[0116] Based on the above improvement of the ant colony algorithm, the mobile robot path planning method of this embodiment specifically includes the following steps:

[0117] Step 1: Consider the working environment of the mobile robot as a two-dimensional plane coordinate system, and construct the working environment into a static grid map using a grid method;

[0118] Step 2: The static grid map is preprocessed using the K-medoids algorithm. The map is divided into several cluster partitions, and the dredging value and congestion value are defined to quantify the complexity of the regional environment and differentiate the pheromone concentration;

[0119] Step 3: Initialize various parameters, including: population size M, pheromone weight α, heuristic function weight β, pheromone volatility coefficient ρ, initialize the taboo table Tabu, and set the maximum number of iterations K;

[0120] Step 4: Place M ants at the starting point S and the end point E of the working environment of the mobile robot respectively, and start from the m=1th pair of ants to check whether the walking paths of the pair of ants have an intersection. If not, transfer to the next node according to the node transfer probability formula, and add the previous node to the respective taboo tables; if so, jump to step 5;

[0121] Step 5: Get a path solution based on the walking path of the pair of ants and calculate the path length; determine whether the path finding of M pairs of ants has been completed. If m = M, find the optimal solution and the worst solution of the current iteration; otherwise, m = m + 1, go to step 4, and the next pair of ants starts path finding again;

[0122] Step 6: At the end of each iteration, update the node pheromone concentration, the number of iterations k = k + 1; when k = K, all iterations are completed, and the global optimal path solution and the corresponding shortest path length are output as the optimal path of the mobile robot.

[0123] In order to verify the effect of the present invention, a grid map with a specification of 20*20 is used to test the mobile robot path planning method based on the improved bidirectional ant colony algorithm of the present invention, and the experimental results are compared with the path planning method using the traditional ant colony algorithm and the existing improved ant colony algorithm. The comparison results are shown in Table 1 (for the convenience of comparison, only horizontal and vertical movement are retained for the traditional ant colony algorithm and the existing improved ant colony algorithm).

[0124] Table 1 Comparison between the present invention and the existing ant colony algorithm

[0125] algorithm Shortest path length Number of turns Convergent Algebra Traditional ant colony algorithm 44 18 23 Existing improved ant colony algorithm 40 15 13 Improved Bidirectional Ant Colony Algorithm 38 12 7

[0126] It can be seen from Table 1 that, in the path planning method of the present invention, the shortest path length of the mobile robot is 38, the number of turns is 12, and the algorithm convergence algebra is 7. Compared with the two solutions in the prior art, not only the path is optimized, but also the planning speed is significantly improved. Therefore, the mobile robot path planning method of the present invention can find the shortest path faster, and the path solution quality is better and more feasible.

[0127] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0128] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A mobile robot path planning method based on an improved ant colony algorithm, characterized in that: The method comprises: Step 1: Consider the working environment of the mobile robot as a two-dimensional plane coordinate system, and construct the working environment into a static grid map using a grid method; Step 2: The static grid map is preprocessed using a clustering algorithm. The map is divided into a number of cluster partitions. The dredging value and the congestion value are defined to quantify the complexity of the regional environment and differentiate the pheromone concentration. Step 3: Initialize various parameters, including: population size M, pheromone weight α, heuristic function weight β, pheromone volatility coefficient ρ, initialize the taboo table Tabu, and set the maximum number of iterations K; Step 4: Place M ants at the starting point S and the end point E of the working environment of the mobile robot respectively, and start from the m=1th pair of ants to check whether the walking paths of the pair of ants have an intersection. If not, transfer to the next node according to the node transfer probability formula, and add the previous node to the respective taboo tables; if so, jump to step 5; Step 5: Get a path solution based on the walking path of the pair of ants and calculate the path length; determine whether the path finding of M pairs of ants has been completed. If m = M, find the optimal solution and the worst solution of the current iteration; otherwise, m = m + 1, go to step 4, and the next pair of ants starts path finding again; Step 6: At the end of each iteration, the node pheromone concentration is updated, and the number of iterations k=k+1; when k=K, all iterations are completed, and the global optimal path solution and the corresponding shortest path length are output as the optimal path of the mobile robot; The node transfer probability formula in step 4 is: in, represents the node transition probability; τ ab (t) is the pheromone concentration between nodes a and b; η ab (t) is the heuristic function from node a to node b; α and β represent the pheromone concentration and the weight of the heuristic function respectively; τ as (t) is the pheromone concentration between nodes a and s, η as (t) is the heuristic function between nodes b and s, and allow represents the set of nodes that the ant is allowed to visit next; The heuristic function η ab (t) is: Among them, H ab is the Manhattan distance from node a to node b, H bE is the Manhattan distance from node b to the end point E, a x 、a y is the horizontal and vertical coordinates of node a, b x and b y is the horizontal and vertical coordinates of node b; The node transfer strategy in step 4 also includes a mark-back strategy, and the mark-back strategy is: When the ant reaches node B, the set of nodes allowed to be visited next is empty, and the ant falls into a deadlock. At this time, node B is marked as an obstacle and the ant returns to node A to prevent subsequent ants from entering the trap again and save path-finding resources. The method for updating the node pheromone concentration includes: introducing a reward and punishment strategy and introducing an inflection point evaluation function; The reward and punishment strategy introduced is: at the end of each iteration, only the path solution that finds the target point is retained, the pheromone of the path node that exceeds the pseudo-global optimal solution is increased, and the pheromone concentration of the worst solution among all current path solutions is reduced. The node pheromone concentration formula is: Among them, Q is the initial value of pheromone, iter_best is the optimal solution in this iteration, iter_worst is the worst solution in this iteration, best is the pseudo global optimal solution; t represents the tth moment, that is, the tth generation of ants, ρ is the pheromone volatility coefficient; The inflection point evaluation function defines the corresponding extended path length according to the different turning angles each time, and the formula is as follows: Where θ is the turning angle, and f(θ) represents the extended path length; The step 2 uses the K-medoids algorithm to cluster the map obstacles, delineate the map partitions according to the center points of each cluster, and use the dredging value and congestion value indicators to represent the complexity of the regional environment; The formula for determining the sum of dissimilarity of cluster centers is as follows: Among them, TC ih Represents the center point O i By non-central point O h The total cost after replacement, C jih Represents the center point O i By non-central point O h After substitution, the non-central point O j the cost; Define the dredge value P i0 Describe the number of grids that can be passed through in the partition and define the congestion value P i Describe the density of obstacles in the partition. The formula is as follows: P i0 =N i0 / S i P i =N i / S i Where N i0 Represents partition Y i Number of grids that can be moved straight inside, N i Represents partition Y i The number of internal obstacle grids, S i Indicates the total number of grids in the partition; The dredging value P of each cluster partition i0 and congestion value P i , define the environment complexity function P(u), the formula is as follows: P(u)=exp(P i0 -P i ) The process of obtaining a path solution according to the walking path of a pair of ants in step 4 and step 5 includes: In each iteration, a group of ants is set at the starting point and the end point of the mobile robot's working environment. Each ant at the starting point and the end point is denoted as S k (k=1,2,…m) and E k (k=1,2,…m), select one ant from each of the starting point and the end point to search for the path, and each pair of ants is recorded as (S k ,E k ), where the starting ant searches forward from the starting point, and the ending ant searches backward from the ending point; Let Path(S k ) is ant S k The set of path nodes passed, Path(E k ) is ant E k The set of path nodes passed, J(S k ,E k ) is the intersection of the nodes that the pair of ants pass through; When J(S k ,E k ) When the set is not empty, it means that the two ants meet on the way, that is, a feasible path is found, and the search stops; the actual path solution Path is the ant S k The node order of ant E k The nodes are connected in reverse order, and the formula is as follows: J(S k ,E k )=Path(S k )∩Path(E k ) Path=Path(S k )+fliplr(Path(E k )) Among them, fliplr(*) means left-right flip matrix.

2. The method according to claim 1, characterized in that In the step 1, constructing the working environment into a static grid map by using the grid method includes: All grids are numbered in order from left to right and from top to bottom. The expressions of grid numbers and actual coordinates are as follows: Among them, i is the grid number, i x is the horizontal coordinate of the grid, i y is the grid ordinate, mapLen is the number of rows and columns of the grid map, mod is the remainder operator, and ceil is the rounding-up operator.

3. A mobile robot path planning device, characterized in that: include: A memory and a processor coupled to the memory, wherein the processor is configured to: execute a mobile robot path planning method based on an improved ant colony algorithm as described in any one of claims 1-2 based on instructions stored in the memory.

4. A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements a mobile robot path planning method based on an improved ant colony algorithm as described in any one of claims 1-2.