Multi-robot conflict-free path planning method combining bilevel planning and response algorithm

Abstract

The invention discloses a multi-robot conflict-free path planning method combining bilevel planning and a response algorithm, and the method comprises the steps: 1, obtaining the current environment information, constructing a grid network of a certain scale, and constructing a directed space-time road network by considering a time set; 2, initially setting robot group information, and constructing a robot group path planning scheme including starting and ending points of a plurality of robots, path constraints, space-time paths and path costs; 3, obtaining initial shortest paths of all the robots by using a Dijkstra algorithm, detecting path conflicts, and setting a robot group position index and a walking priority sorting scheme so as to perform path re-planning on the corresponding robots; and 4, transforming the walking priority sequence of the robot group in the upper-layer planning by using a genetic algorithm. According to the method, the directed space-time road network can be utilized to more accurately describe the environment, the robots and the conflicts, and the problem of multi-robot conflict resolution is better solved through the path constraint in the road network, so that the feasibility of the multi-robot path can be ensured.

Description

Technical Field

[0001] The present invention belongs to the field of multi-robot path planning, and in particular to a multi-robot conflict-free path planning method combining double-layer planning and response algorithms. Background Art

[0002] The problem of multi-robot path conflict resolution refers to how to effectively plan their paths to avoid collisions and interference when multiple robots move or perform tasks at the same time; or when multiple robot paths conflict, how to take timely and accurate avoidance measures to ensure the normal path planning of each subsequent robot. This problem involves multiple aspects such as path planning, task allocation, and real-time decision-making. At present, the multi-robot path conflict problem has been widely used in many fields, including automated handling in warehousing and logistics, traffic coordination of unmanned vehicles, operation scheduling in precision agriculture, rapid deployment in emergency rescue, task execution of drone swarms, and action coordination in military applications. These applications demonstrate the importance of effectively resolving path conflicts. With the development of technology, the efficiency and reliability requirements of the system will continue to increase.

[0003] The main methods for resolving path conflicts in multi-robot path planning include the A* algorithm and the conflict search algorithm (CBS). In addition to the above methods, multi-robot path planning also involves other technologies and methods, such as centralized path planning and distributed path planning. However, the above methods all have the following shortcomings:

[0004] 1. The computational complexity doubles. For large-scale problems, especially in complex environments or with many obstacles, the A* algorithm may require a large amount of computation, resulting in slow path planning. Although the CBS algorithm can effectively handle conflicts, when the number of robots increases, the search space increases rapidly, which may lead to a decrease in efficiency.

[0005] 2. Poor real-time performance. Centralized path planning requires global information, and the update and optimization process is slow, which is not suitable for real-time applications in dynamic environments; in distributed path planning, individual robots may lack a global vision, resulting in poor coordination and possible unexpected conflicts;

[0006] 3. The correlation between time and space is not considered. Most of the algorithms currently used consider the spatial path and the time and space window separately, or represent them discretely with characters, which will lead to the separation of time and space relationship and fail to intuitively reflect the conflict of multiple robot paths. Summary of the invention

[0007] The present invention aims to address the deficiencies of the above-mentioned prior art and proposes a multi-robot conflict-free path planning method combining double-layer planning and response algorithm, in order to construct a spatiotemporal road network, construct multi-robot priority walking rights and resolve conflicts by transforming the spatiotemporal road network, so as to more accurately describe the environment, robots and tasks, and re-plan the conflicting multi-robot paths to improve the efficiency and accuracy of multi-robot path planning.

[0008] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme:

[0009] The multi-robot conflict-free path planning method combining double-level planning and response algorithm of the present invention is characterized in that it is carried out according to the following steps:

[0010] Step 1: Construct a directed spatiotemporal road network based on the grid network;

[0011] Step 1.1: Build a grid network;

[0012] Obtain environmental information and construct a grid network G=(V,E,W) of size L×L, where L is the total number of rows or columns, V represents the grid set, and , Indicates Line Grid of columns; represents the set of connectivity relationships between rasters, and , Represents a grid With Line Column Grid The connectivity relationship between them defines that the grid is only connected to the grids directly adjacent to it in the up, down, left, and right directions, that is, ; represents the set of movement weights between connected grids, and , For Grid With Grid The moving weight between grids With Grid There are no obstacles on the two grids, so If the grid With Grid Connected and grid With Grid If there is at least one obstacle, ;

[0013] Step 1.2: Construct a directed space-time road network;

[0014] Determine the estimated search completion time ,Will After discretization with a unit step, we get the time set ;in, Represents the total moment, based on a time collection and raster networks The construction scale is Directed space-time network ;in, Represents a time collection The grid collection under , Indicated in The moment Row, No. Column grid; represents the set of directed road segments between grids at adjacent times; and , express Grid at the moment and Grid at the moment The directed path between Grid at the moment and Grid at the moment Connectivity; represents the set of movement weights of the directed segments between grids at adjacent times, and , Indicates a directed segment The moving weight; if there is a directional segment If there are no obstacles on the ; If there is a directional section If there is at least one obstacle on ;

[0015] Step 2: Initialize robot group information;

[0016] Step 2.1: Let the robot group be , Represents the total number of robots, represents the kth robot, The path planning scheme is recorded as ,and , Indicates The starting point of the robot, and , Represents the robot group The starting point set, Indicates the robot's endpoint; and , Represents the robot group The endpoint set, Indicates The set of path constraints for each robot, Indicates The path of a robot under path constraints on the space-time network, Indicates Robots on the path The sum of the movement weights on is recorded as the path the cost;

[0017] Make all robots Start at the same time, assuming that Grid As a starting point , then A robot in The grid starting at time is recorded as ;

[0018] The time when all robots reach the end point Assume that Grid As an end point , then A robot in The grid at which time is reached is recorded as ;

[0019] Step 2.2: Define the robot's state in the space-time network, including waiting state and moving state. The waiting state means that the grid of a robot remains unchanged at a number of consecutive moments; the moving state means that the two grids of a robot at adjacent moments are different and connected to each other.

[0020] Step 2.3: Define the conflict:

[0021] If the paths of multiple robots contain the same grid at the same time, it means that there is a point conflict between the corresponding multi-robot paths;

[0022] If the starting grid of the directed path of the kth robot at time t is the ending grid of the k'th robot at time t+1, and the starting grid of the k'th robot at time t is the ending grid of the kth robot at time t+1, then there is an edge conflict between the paths of the kth robot and the k'th robot;

[0023] Step 3: Order = ; Use Dijkstra algorithm to get the A robot starts from the starting point in the grid network. To the end The shortest path ; Thus, the time collection Allocate to On, get The shortest path of a robot on a space-time network , and the shortest path After summing up the moving weights of each directed segment in The shortest path of robots on the space-time network The cost ; Thus, the shortest path set on the space-time network composed of the shortest paths of all robots is obtained , and calculate the total shortest path cost of robot group K' ;

[0024] Step 4: Construct a two-level planning model;

[0025] Step 4.1: Construct the priority walking right sorting mode of the upper robot;

[0026] The position index in the robot group K' is recorded as ,in, represents the kth position index in K';

[0027] Step 4.2: Construct the conflict-free path decision of the lower-level multi-robots;

[0028] Step 4.2.1: Use the best response algorithm to perform conflict resolution path planning for the robot group K', so as to obtain the shortest path cost of all robots and the optimal path planning solution corresponding to their corresponding optimal shortest path. , and calculate the total shortest path optimal cost of the robot group K' ;

[0029] Step 4.2.2: According to Corresponding , change the priority walking right ranking to obtain the optimal priority walking right ranking scheme;

[0030] Step 4.2.3: Process the optimal priority walking right sorting scheme according to step 4.2.1 to obtain the optimal shortest path cost of all robots and the optimal path planning scheme corresponding to their corresponding optimal shortest paths.

[0031] The multi-robot conflict-free path planning method combining double-level planning and response algorithm described in the present invention is also characterized in that step 4.2.1 is performed as follows:

[0032] Step 4.2.1.1: Detect conflicts:

[0033] If there are point conflicts or edge conflicts in the paths of all robots on the space-time network, execute step 4.2.1.2; otherwise, the shortest path set on the space-time network consisting of the shortest paths of all robots is As a conflict-free path for multiple robots;

[0034] Step 4.2.1.2: Define and initialize variables =0;

[0035] Step 4.2.1.3: Current iteration number e=0;

[0036] Get the dth position index [d] Corresponding robot As the current robot, the remaining robot set is expressed as - };

[0037] Step 4.2.1.4: Transform the directed space-time network :

[0038] Suppose that on the space-time network graph The sum of the movement weights of each robot on the path itself is , thus obtaining a new directed space-time network graph , and get the current robot In the new directed space-time network graph The path constraint set in ;

[0039] Step 4.2.1.4: Initialize the current robot The shortest path cost ,in, For the current robot The shortest path on the space-time network the cost;

[0040] Step 4.2.1.5: Use Dijkstra's algorithm to find the current robot on the new directed space-time graph. Perform path replanning under the e-th iteration and obtain the current robot The shortest path at iteration e , so that the time is allocated to On, get the current robot The shortest path at iteration e and its cost ,like , then let Assign to ;otherwise, remains unchanged; thus The starting point, end point, and path constraint set of , constitute Path planning solution at the e-th iteration ;

[0041] Step 4.2.1.6: Assign e+1 to e. If e Emax, then return to step 4.2.1.5 to execute sequentially; otherwise, it means that the current robot The shortest path cost and its corresponding optimal shortest path The corresponding optimal path planning solution ; Where Emax represents the maximum number of iterations;

[0042] Step 4.2.1.7: Assign d+1 to d, if , then return to step 4.2.1.3 to execute sequentially; otherwise, it means that the shortest path cost of all robots and the optimal path planning solution corresponding to their corresponding optimal shortest path are obtained. , and calculate the total shortest path optimal cost of the robot group K' .

[0043] Further, the step 4.2.2 is performed as follows:

[0044] Step 4.2.2.1: Define the current algebra as inter and initialize inter=1;

[0045] Let position index The corresponding robot group is regarded as the current individual, and each robot in the robot group is regarded as a gene in the current individual;

[0046] Move the robot corresponding to the first position index in the current individual to the end, and move the remaining robots forward one position in turn to obtain a new individual. Continue to perform the same moving operation on the new individual until K different individuals are obtained and form the inter-generation population;

[0047] Define and initialize the optimal population fitness function value ;

[0048] Step 4.2.2.2: Process each individual in the inter-th generation population according to step 4.2.1 to obtain the optimal total shortest path cost of each individual in the inter-th generation population, and use the reciprocal of the optimal total shortest path cost of each individual in the inter-th generation population as the fitness function value of each individual in the inter-th generation population, so as to take the sum of the fitness function values ​​of all individuals in the inter-th generation population as the fitness function value of the inter-th generation population. ;

[0049] Step 4.2.2.3: If the fitness function value of the inter generation population Greater than the optimal population fitness function value , then Assign to ,otherwise, remain unchanged;

[0050] Step 4.2.2.4: For each individual in the inter generation population, randomly select two position indices and exchange the robots at their corresponding positions, thereby obtaining the inter+1 generation population containing K new individuals, and using it as the new priority walking right sorting scheme;

[0051] Step 4.2.2.5: After assigning inter+1 to inter, return to step 4.2.2.2 and execute sequentially until inter= so far, thus The corresponding population is taken as the optimal population; among them, represents the maximum number of iterations of the population,

[0052] Step 4.2.2.6: Select the individual with the largest fitness function value in the optimal population as the optimal priority walking right sorting scheme.

[0053] The electronic device of the present invention includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the multi-robot conflict-free path planning method, and the processor is configured to execute the program stored in the memory.

[0054] The present invention provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and the computer program executes the steps of the multi-robot conflict-free path planning method when the computer program is executed by a processor.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The present invention uses a spatiotemporal road network to handle multi-robot path planning problems. Based on a grid network graph, the time factor is considered, the estimated search completion time is discretized with a unit step length, a directed spatiotemporal road network is constructed, and the connectivity relationship and movement weight of the grid in the road network are defined. Compared with the prior art, the method used in the present invention effectively combines spatial and temporal information, helps identify conflicts, replan paths, and transforms complex problems into intuitive graph search problems, which can provide a unified basis for various path planning methods, thereby reducing the complexity of path calculation.

[0057] 2. The present invention uses a spatiotemporal network to describe the motion state and conflict of multiple robots. It can detect conflicts intuitively and quickly on the spatiotemporal network, and take timely measures to resolve conflicts. Compared with the prior art, the method used in the present invention can clearly identify potential conflicts between different robots by taking the time dimension into consideration, helping the planning algorithm to avoid path overlap and collision, and can avoid possible collisions in advance at the path planning stage, thereby improving the stability of multi-robot collaborative operations.

[0058] 3. The present invention uses a combination of two-layer planning and genetic algorithm to solve the problem of multi-robot path conflict. The upper layer plans the priority walking right ranking of multiple robots, and the lower layer replans the paths of multiple robots based on this ranking by transforming the spatiotemporal road network, and feeds the results back to the fitness function to form a response optimization mechanism, thereby optimizing the priority walking right ranking of the upper multi-robots. Compared with the prior art, the method adopted by the present invention can solve the conflict problem as a whole, greatly reduce the complexity of local conflict resolution, and improve the flexibility and adaptability of decision-making. At the same time, the lower layer path replanning is used to effectively reduce conflicts and ensure the smooth execution of tasks. The response feedback mechanism enables the system to be continuously optimized and enhances the adaptive ability, thereby improving the overall collaboration efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is the overall grid network diagram of the present invention;

[0060] Figure 2 It is a local grid network node diagram of the present invention;

[0061] Figure 3 It is the directed space-time road network diagram of the present invention;

[0062] Figure 4 This is a waiting state diagram of the robot of the present invention;

[0063] Figure 5 This is the mobile state diagram of the robot of the present invention. DETAILED DESCRIPTION

[0064] In this embodiment, a multi-robot conflict-free path planning method combining a two-level planning and a response algorithm is performed in the following steps:

[0065] Step 1: Construct a directed spatiotemporal road network based on the grid network;

[0066] Step 1.1: Figure 1 As shown, a grid network is constructed;

[0067] Obtain environmental information and construct a grid network G=(V,E,W) of size L×L, where L is the total number of rows or columns, V represents the grid set, and , Indicates Line Grid of columns; represents the set of connectivity relationships between rasters, and , Represents a grid With Line Column Grid The connectivity relationship between them defines that the grid is only connected to the grids directly adjacent to it in the up, down, left, and right directions, that is, ; represents the set of movement weights between connected grids, and , For Grid With Grid The moving weight between grids With Grid There are no obstacles on the two grids, so If the grid With Grid Connected and grid With Grid If there is at least one obstacle, ;

[0068] In the specific implementation, the overall grid network diagram constructed according to the above method is as follows: Figure 1 As shown in Figure 1, the map size is 20×20, where white grids represent feasible areas and black grids represent fixed obstacles. Figure 2 As shown, the correspondence between the grid and the label is shown.

[0069] Step 1.2: Construct a directed space-time road network;

[0070] Determine the estimated search completion time ,Will After discretization with a unit step, we get the time set ;in, Represents the total moment, based on a time collection and raster networks The construction scale is Directed space-time network ;in, Represents a time collection The grid collection under , Indicated in The moment Row, No. Column grid; represents the set of directed road segments between grids at adjacent times; and , express Grid at the moment and Grid at the moment The directed path between Grid at the moment and Grid at the moment Connectivity; represents the set of movement weights of the directed segments between grids at adjacent times, and , Indicates a directed segment The moving weight; if there is a directional segment If there are no obstacles on the ; If there is a directional section If there is at least one obstacle on ;

[0071] In the specific implementation, the directed space-time road network diagram constructed according to the above method is as follows: Figure 3 As shown, Figure 3 For Figure 1 and Figure 2 A directed space-time road network graph constructed based on.

[0072] Step 2: Initialize robot group information;

[0073] Step 2.1: Let the robot group be , Represents the total number of robots, represents the kth robot, The path planning scheme is recorded as ,and , Indicates The starting point of the robot, and , Represents the robot group The starting point set, Indicates the robot's endpoint; and , Represents the robot group The endpoint set, Indicates The set of path constraints for each robot, Indicates The path of a robot under path constraints on the space-time network, Indicates Robots on the path The sum of the movement weights on is recorded as the path the cost;

[0074] Make all robots Start at the same time, assuming that Grid As a starting point , then A robot in The grid starting at time is recorded as ;

[0075] The time when all robots reach the end point Assume that Grid As an end point , then A robot in The grid at which time is reached is recorded as .

[0076] Step 2.2: Define the robot's state in the space-time network, including waiting state and moving state. The waiting state means that the grid of a robot remains unchanged at a number of consecutive moments; the moving state means that the two grids of a robot at adjacent moments are different and connected to each other.

[0077] That is, the robot waiting state table is represented as a set of directed sections in the spatiotemporal network , For the current moment, is the waiting time, which is represented by a vertical line segment on the space-time network diagram, such as Figure 4 As shown in the figure, the uniform movement state of the robot is represented in the spatiotemporal network as a set of directed sections between the grids at the current moment and the next moment. , which is represented as a series of oblique line segments on the space-time network diagram, such as Figure 5 shown.

[0078] Step 2.3: Define the conflict:

[0079] If the paths of multiple robots contain the same grid at the same time, that is, if there are multiple Contains the same grid , it means that there are point conflicts between the corresponding multi-robot paths;

[0080] If the starting grid of the directed path of the kth robot at time t is the ending grid of the k'th robot at time t+1, and the starting grid of the k'th robot at time t is the ending grid of the kth robot at time t+1, that is, if there exists Contains directed segments ,path Contains directed segments , it means that there is an edge conflict between the paths of the k-th robot and the k'-th robot.

[0081] Step 3: Order = ; Use Dijkstra algorithm to get the A robot starts from the starting point in the grid network. To the end The shortest path ; Thus, the time collection Allocate to On, get The shortest path of a robot on a space-time network , and the shortest path After summing up the moving weights of each directed segment in The shortest path of robots on the space-time network The cost ; Thus, the shortest path set on the space-time network composed of the shortest paths of all robots is obtained , and calculate the total shortest path cost of robot group K' .

[0082] Step 4: Construct a two-level planning model;

[0083] Step 4.1: Construct the priority walking right sorting mode of the upper robot;

[0084] The position index in the robot group K' is recorded as ,in, represents the kth position index in K';

[0085] Step 4.2: Construct the conflict-free path decision of the lower-level multi-robots;

[0086] Step 4.2.1: Use the best response algorithm to perform conflict resolution path planning for the robot group K';

[0087] Step 4.2.1.1: Detect conflicts:

[0088] If there are point conflicts or edge conflicts in the paths of all robots on the space-time network, execute step 4.2.1.2; otherwise, the shortest path set on the space-time network consisting of the shortest paths of all robots is As a conflict-free path for multiple robots;

[0089] Step 4.2.1.2: Define and initialize variables =0;

[0090] Step 4.2.1.3: Current iteration number e=0;

[0091] Get the dth position index [d] Corresponding robot As the current robot, the remaining robot set is expressed as - }.

[0092] Step 4.2.1.4: Transform the directed space-time network :

[0093] Suppose that on the space-time network graph The sum of the movement weights of each robot on the path itself is , thus obtaining a new directed space-time network graph , and get the current robot In the new directed space-time network graph The path constraint set in ;

[0094] Step 4.2.1.4: Initialize the current robot The shortest path cost ,in, For the current robot The shortest path on the space-time network the cost;

[0095] Step 4.2.1.5: Use Dijkstra's algorithm to find the current robot on the new directed space-time graph. Perform path replanning under the e-th iteration and obtain the current robot The shortest path at iteration e , so that the time is allocated to On, get the current robot The shortest path at iteration e and its cost ,like , then let Assign to ;otherwise, remains unchanged; thus The starting point, end point, and path constraint set of , constitute Path planning solution at the e-th iteration ;

[0096] Step 4.2.1.6: Assign e+1 to e. If e Emax, then return to step 4.2.1.5 to execute sequentially; otherwise, it means that the current robot The shortest path cost and its corresponding optimal shortest path The corresponding optimal path planning solution ; Where Emax represents the maximum number of iterations;

[0097] Step 4.2.1.7: Assign d+1 to d, if , then return to step 4.2.1.3 to execute sequentially; otherwise, it means that the shortest path cost of all robots and the optimal path planning solution corresponding to their corresponding optimal shortest path are obtained. , and calculate the total shortest path optimal cost of the robot group K' .

[0098] Step 4.2.2: According to Corresponding , change the priority walking right ranking, obtain the final priority walking right ranking scheme and output it;

[0099] Step 4.2.2.1: Define the current algebra as inter and initialize inter=1;

[0100] Let position index The corresponding robot group is regarded as the current individual, and each robot in the robot group is regarded as a gene in the current individual;

[0101] Move the robot corresponding to the first position index in the current individual to the end, and move the remaining robots forward one position in turn to obtain a new individual. Continue to perform the same moving operation on the new individual until K different individuals are obtained and form the inter-generation population;

[0102] Define and initialize the optimal population fitness function value .

[0103] Step 4.2.2.2: Process each individual in the inter-th generation population according to step 4.2.1 to obtain the optimal total shortest path cost of each individual in the inter-th generation population, and use the reciprocal of the optimal total shortest path cost of each individual in the inter-th generation population as the fitness function value of each individual in the inter-th generation population, so as to take the sum of the fitness function values ​​of all individuals in the inter-th generation population as the fitness function value of the inter-th generation population. ;

[0104] Step 4.2.2.3: If the fitness function value of the inter generation population Greater than the optimal population fitness function value , then Assign to ,otherwise, remain unchanged;

[0105] Step 4.2.2.4: For each individual in the inter generation population, randomly select two position indices and exchange the robots at their corresponding positions, thereby obtaining the inter+1 generation population containing K new individuals, and using it as the new priority walking right sorting scheme;

[0106] Step 4.2.2.5: After assigning inter+1 to inter, return to step 4.2.2.2 and execute sequentially until inter= so far, thus The corresponding population is taken as the optimal population; among them, represents the maximum number of iterations of the population,

[0107] Step 4.2.2.6: Select the individual with the largest fitness function value in the optimal population as the optimal priority walking right sorting scheme, and process the optimal priority walking right sorting scheme according to step 4.2.1 to obtain the optimal shortest path cost of all robots and the optimal path planning scheme corresponding to their corresponding optimal shortest path.

[0108] In this embodiment, an electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0109] In this embodiment, a computer-readable storage medium stores a computer program on the computer-readable storage medium, and the computer program executes the steps of the above method when executed by a processor.

Claims

1. A multi-robot conflict-free path planning method combining two-level planning and response algorithm, characterized in that: The steps are as follows: Step 1: Construct a directed spatiotemporal road network based on the grid network; Step 1.1: Build a grid network; Obtain environmental information and construct a grid network G=(V,E,W) of size L×L, where L is the total number of rows or columns, V represents the grid set, and , Indicates Line Grid of columns; represents the set of connectivity relationships between rasters, and , Represents a grid With Line Column Grid The connectivity relationship between them defines that the grid is only connected to the grids directly adjacent to it in the up, down, left, and right directions, that is, ; represents the set of movement weights between connected grids, and , For Grid With Grid The moving weight between grid With Grid There are no obstacles on the two grids, so If the grid With Grid Connected and grid With Grid If there is at least one obstacle, ; Step 1.2: Construct a directed space-time road network; Determine the estimated search completion time ,Will After discretization with a unit step, we get the time set ;in, Represents the total moment, based on a time collection and raster networks The construction scale is Directed space-time network ;in, Represents a time collection The grid collection under , Indicated in The moment Row, No. Column grid; represents the set of directed road segments between grids at adjacent times; and , express Grid at the moment and Grid at the moment The directed path between Grid at the moment and Grid at the moment Connectivity; represents the set of movement weights of the directed segments between grids at adjacent times, and , Indicates a directed segment The moving weight; if there is a directional segment If there are no obstacles on the ; If there is a directional section If there is at least one obstacle on ; Step 2: Initialize robot group information; Step 2.1: Let the robot group be , Represents the total number of robots, represents the kth robot, The path planning scheme is recorded as ,and , Indicates The starting point of the robot, and , Represents the robot group The starting point set, Indicates the robot's endpoint; and , Represents the robot group The endpoint set, Indicates The set of path constraints for each robot, Indicates The path of a robot under path constraints on the space-time network, Indicates Robots on the path The sum of the movement weights on is recorded as the path the cost; Make all robots Start at the same time, assuming that Grid As a starting point , then A robot in The grid starting at time is recorded as ; The time when all robots reach the end point ; Assume that Grid As an end point , then A robot in The grid at which time is reached is recorded as ; Step 2.2: Define the robot's state in the space-time network, including waiting state and moving state. The waiting state means that the grid of a robot remains unchanged at a number of consecutive moments; the moving state means that the two grids of a robot at adjacent moments are different and connected to each other. Step 2.3: Define the conflict: If the paths of multiple robots contain the same grid at the same time, it means that there is a point conflict between the corresponding multi-robot paths; If the starting grid of the directed path of the kth robot at time t is the ending grid of the k'th robot at time t+1, and the starting grid of the k'th robot at time t is the ending grid of the kth robot at time t+1, then there is an edge conflict between the paths of the kth robot and the k'th robot; Step 3: Order = ; Use Dijkstra algorithm to get the A robot starts from the starting point in the grid network. To the end The shortest path ; Thus, the time collection Allocate to On, get The shortest path of a robot on a space-time network , and the shortest path After summing up the moving weights of each directed segment in The shortest path of robots on the space-time network The cost ; Thus, the shortest path set on the space-time network composed of the shortest paths of all robots is obtained , and calculate the total shortest path cost of robot group K' ; Step 4: Construct a two-level planning model; Step 4.1: Construct the priority walking right sorting mode of the upper robot; The position index in the robot group K' is recorded as ,in, represents the kth position index in K'; Step 4.2: Construct the conflict-free path decision of the lower-level multi-robots; Step 4.2.1: Use the best response algorithm to perform conflict resolution path planning for the robot group K', so as to obtain the shortest path cost of all robots and the optimal path planning solution corresponding to their corresponding optimal shortest path. , and calculate the total shortest path optimal cost of the robot group K' ; Step 4.2.2: According to Corresponding , change the priority walking right ranking to obtain the optimal priority walking right ranking scheme; Step 4.2.3: Process the optimal priority walking right sorting scheme according to step 4.2.1 to obtain the optimal shortest path cost of all robots and the optimal path planning scheme corresponding to their corresponding optimal shortest paths.

2. A multi-robot conflict-free path planning method combining double-level planning and response algorithm according to claim 1, characterized in that: The step 4.2.1 is performed as follows: Step 4.2.1.1: Detect conflicts: If there are point conflicts or edge conflicts in the paths of all robots on the space-time network, execute step 4.2.1.2; otherwise, the shortest path set on the space-time network consisting of the shortest paths of all robots is As a conflict-free path for multiple robots; Step 4.2.1.2: Define and initialize variables =0; Step 4.2.1.3: Current iteration number e=0; Get the dth position index [d] Corresponding robot As the current robot, the remaining robot set is expressed as - }; Step 4.2.1.4: Transform the directed space-time network : Suppose that on the space-time network graph The sum of the movement weights of each robot on the path itself is , thus obtaining a new directed space-time network graph , and get the current robot In the new directed space-time network graph The path constraint set in ; Step 4.2.1.4: Initialize the current robot The shortest path cost ,in, For the current robot The shortest path on the space-time network the cost; Step 4.2.1.5: Use Dijkstra's algorithm to find the current robot on the new directed space-time graph. Perform path replanning under the e-th iteration and obtain the current robot The shortest path at iteration e , so that the time is allocated to On, get the current robot The shortest path at iteration e and its cost ,like , then let Assign to ;otherwise, remains unchanged; thus The starting point, end point, and path constraint set of , constitute Path planning solution at the e-th iteration ; Step 4.2.1.6: Assign e+1 to e. If e Emax, then return to step 4.2.1.5 to execute sequentially; otherwise, it means that the current robot The shortest path cost and its corresponding optimal shortest path The corresponding optimal path planning solution ; Where Emax represents the maximum number of iterations; Step 4.2.1.7: Assign d+1 to d, if , then return to step 4.2.1.3 to execute sequentially; otherwise, it means that the shortest path cost of all robots and the optimal path planning solution corresponding to their corresponding optimal shortest path are obtained. , and calculate the total shortest path optimal cost of the robot group K' .

3. A multi-robot conflict-free path planning method combining double-level planning and response algorithm according to claim 2, characterized in that: The step 4.2.2 is performed as follows: Step 4.2.2.1: Define the current algebra as inter and initialize inter=1; Let position index The corresponding robot group is regarded as the current individual, and each robot in the robot group is regarded as a gene in the current individual; Move the robot corresponding to the first position index in the current individual to the end, and move the remaining robots forward one position in turn to obtain a new individual. Continue to perform the same moving operation on the new individual until K different individuals are obtained and form the inter-generation population; Define and initialize the optimal population fitness function value ; Step 4.2.2.2: Process each individual in the inter-th generation population according to step 4.2.1 to obtain the optimal total shortest path cost of each individual in the inter-th generation population, and use the reciprocal of the optimal total shortest path cost of each individual in the inter-th generation population as the fitness function value of each individual in the inter-th generation population, so as to take the sum of the fitness function values ​​of all individuals in the inter-th generation population as the fitness function value of the inter-th generation population. ; Step 4.2.2.3: If the fitness function value of the inter generation population Greater than the optimal population fitness function value , then Assign to ,otherwise, remain unchanged; Step 4.2.2.4: For each individual in the inter generation population, randomly select two position indices and exchange the robots at their corresponding positions, thereby obtaining the inter+1 generation population containing K new individuals, and using it as the new priority walking right sorting scheme; Step 4.2.2.5: After assigning inter+1 to inter, return to step 4.2.2.2 and execute sequentially until inter= so far, thus The corresponding population is taken as the optimal population; among them, represents the maximum number of iterations of the population, Step 4.2.2.6: Select the individual with the largest fitness function value in the optimal population as the optimal priority walking right sorting scheme.

4. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the multi-robot conflict-free path planning method described in any one of claims 1 to 3, and the processor is configured to execute the program stored in the memory.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-robot conflict-free path planning method according to any one of claims 1 to 3 are executed.

Images