Multi-agent conflict-free path planning method based on directed space-time network and forward direction decision

Abstract

The invention discloses a multi-agent conflict-free path planning method based on a directed space-time network and forward direction decision, which comprises the following steps: 1, acquiring environment information, constructing a grid network according to the environment information, and constructing a directed space-time road network for the grid network by using the space-time network, initializing starting and ending point information of the robot group according to the directed space-time road network; 2, establishing a forward direction decision model of the agent group, and avoiding conflicts by using constraint conditions of the model; and 3, solving a forward direction decision model of the agent group, and planning an overall optimal path. According to the method, the directed space-time road network can be constructed by utilizing the space-time network to accurately describe the environment, the agents and the tasks, the path of the agents is planned by utilizing the advancing direction decision model, and the possibility of multi-agent conflict is avoided, so that the feasibility of the path is ensured, and the efficiency and the safety of multi-agent path planning can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of agent path planning, and specifically relates to a multi-agent conflict-free path planning method based on a directed spatio-temporal network and forward direction decision-making. Background Art

[0002] Multi-agent path planning refers to dynamically planning the movement paths of multiple agents in a system containing multiple agents, so that they can efficiently complete specific tasks without collision or conflict. At present, multi-agent path planning technology has a wide range of applications in various fields. In the logistics field, with the rapid development of e-commerce and the express delivery industry, multi-agent path planning can help achieve automated goods handling in warehouses and intelligent logistics distribution, improving logistics efficiency and reducing costs; in the manufacturing industry, multi-agent path planning can be applied to automated production lines to coordinate the movement of multiple industrial agents and complete tasks such as assembly and processing, improving production efficiency and flexibility.

[0003] However, there are still some deficiencies in existing path planning methods in practical applications:

[0004] 1. In complex and dynamic environments, problems such as high computational complexity and long planning time are often faced. Especially in an environment with a large number of obstacles or dynamic changes, it is difficult to find an efficient and feasible path planning method;

[0005] 2. Existing path planning methods often can only ensure the path safety of a single agent, but when multiple agents cooperate, collisions or conflicts between agents are likely to occur. Summary of the Invention

[0006] The present invention is to solve the above-mentioned deficiencies of the existing technology, and proposes a multi-agent conflict-free path planning method based on a directed spatio-temporal network, in order to use the spatio-temporal network to construct a directed spatio-temporal road network to accurately describe the environment, agents and tasks, and use a mathematical model and implicit enumeration method to plan the paths of agents, avoiding the possibility of multi-agent conflicts, so as to ensure the feasibility of the paths and improve the efficiency and safety of multi-agent path planning.

[0007] The present invention adopts the following technical solutions to achieve the above-mentioned invention purpose:

[0008] The characteristics of a multi-agent conflict-free path planning method based on a directed spatio-temporal network and forward direction decision-making of the present invention are as follows: the method is carried out according to the following steps:

[0009] Step 1: Construct a directed spatio-temporal network;

[0010] Step 2: Set basic information:

[0011] Step 2.1: Set the basic information of the agent group:

[0012] Let the agent group be denoted as , representing the total number of agents, representing the k-th agent; let represent the -th agent 's starting point, , representing the starting grid set of the agent group ; let represent the -th agent 's end point; , representing the end grid set of the agent group

[0013] Step 2.2: Calculate the shortest path weight of the agent group :

[0014] Use the Floyd - Warshall algorithm to calculate the shortest path weight from the -th agent in the directed spatio - temporal network at time from the -th grid to the end point of the -th agent , thus obtaining the set of shortest path weights of each agent from its current grid to its respective end point under the time set in the directed spatio - temporal network ;

[0015] Step 3: Based on the directed spatio - temporal network and C, construct the forward direction decision model of the agent group at time :

[0016] Step 4: According to the directed spatio - temporal road network and the forward direction decision model of the agent group , solve the path planning scheme of multiple agents.

[0017] The feature of a multi - agent conflict - free path planning method based on a directed spatio - temporal network and forward direction decision according to the present invention is also that step 1 is carried out as follows:

[0018] Step 1.1: Construct a grid network; ​

[0019] Obtain environmental information and construct a raster network with a scale of , where , represents a raster set, and , represents the th raster in , and the raster is defined to be connected to the directly adjacent raster in the upper, lower, left, and right directions only. That is, if the th raster is the neighbor raster of raster , then ; represents the set of directed road segments between rasters, and , represents the th raster and the th raster ; represents the set of movement weights between connected rasters, and , is the movement weight of the directed road segment between the th raster and the th raster . If there are no obstacles in the directed road segment between the th raster and the th raster , then let ; if there is at least one obstacle in the directed road segment between the th raster and the th raster , then let ;

[0020] Step 1.2: Construct a directed spatio-temporal road network ;

[0021] Determine the estimated search completion time as , and after discretizing with a unit step size, obtain the time set ; based on the time set and the raster network , construct a directed spatio-temporal road network with a scale of , , represents the raster set under the time set , , where represents The raster set at a moment, and , represents the th raster at a moment; it is defined that a raster is only connected to the directly adjacent rasters in adjacent time steps, that is, if raster is a neighbor raster of raster , then ; represents the set of directed road segments between rasters at different moments, and , represents the th raster and the th raster at a moment represents the set of action weights of the directed road segments between rasters at different moments, and , represents the action weight of the directed road segment .

[0022] Furthermore, step 3 is carried out as follows:

[0023] Step 3.1: Use equation (1) to construct the objective function of the forward direction decision model of the agent group at moment : :

[0024] (1)

[0025] In equation (1), represents whether the kth agent passes through the directed road segment at moment . If so, let = 1, otherwise, let = 0; represents the th raster at moment represents the th raster at moment

[0026] Step 3.2: Use equation (2) to construct the constraint conditions of the forward direction decision model:

[0027] (2)

[0028] In equation (2), represents the an agent at whether it passes through a directed road section at a moment , if so, let = 1, otherwise, let = 0; represents the th agent at whether it passes through a directed road section at a moment ; if so, let , otherwise, let , is the corresponding decision variable, indicating whether the kth agent at is at the ith grid at a moment , if so, let = 1; otherwise, let = 0.

[0029] Furthermore, step 4 is carried out as follows:

[0030] Step 4.1: Let = 1;

[0031] Step 4.2: Let = 1;

[0032] Step 4.3: Define and initialize the set of directed road sections of the kth agent on the directed spatio-temporal network as an empty set, that is = ;

[0033] Step 4.4: According to the forward grid that the kth agent can choose at a moment , obtain the forward strategy candidate set of the kth agent = ; where represents the kth agent being at the th grid at a moment;

[0034] Step 4.5: Traverse the forward strategy candidate set of the kth agent , and take the decision variable corresponding to and the decision variable corresponding to After assigning all values to 1, substitute them into the forward direction decision model in Step 3 for solution to obtain the k-th agent corresponding to the minimum objective function value of the optimal strategy set ;

[0035] Step 4.6: If = , it means that the k-th agent will move from grid to at time , and add the directed road segment to the end of the directed road segment set , and let = ;

[0036] Step 4.7: If the arrival grid of the k-th agent at time is the end point , it means that the final directed road segment set of the k-th agent is obtained, and it forms the path planning scheme of the k-th agent together with ), and execute Step 4.8; otherwise, execute Step 4.9;

[0037] Step 4.8: Assign to , and return to Step 4.4 to execute sequentially until ;

[0038] Step 4.9: After assigning k + 1 to k, return to Step 4.3 until k > K, so as to output the path planning scheme R of the agent group

[0039] .

[0039] An electronic device according to the present invention includes a memory and a processor, characterized in that the memory is used to store a program for supporting the processor to execute the multi-agent conflict-free path planning method, and the processor is configured to execute the program stored in the memory.

[0040] A computer-readable storage medium according to the present invention, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is run by a processor, it executes the steps of the multi-agent conflict-free path planning method.

[0041] 1. The present invention uses a spatio-temporal network to plan the paths of multiple robots. By effectively integrating and utilizing spatio-temporal information, a directed spatio-temporal road network is established, which can significantly improve the efficiency of path planning. Compared with the prior art, the spatio-temporal network structure adopted by the present invention can more accurately describe the motion state of the robots and environmental constraints, thereby reducing the complexity and time cost of path search.

[0042] 2. The present invention introduces collaborative planning of agent groups in the spatio-temporal network and establishes a decision-making model for the forward direction of agent groups. Compared with the prior art, the method adopted by the present invention can effectively avoid collisions and crossing phenomena between agents, and can pre-emptively avoid possible collision situations during the path planning stage, improving the safety and stability of multi-agent collaborative operations.

[0043] 3. The present invention has good dynamic adaptability and can quickly respond to changes in the environment and task requirements. Compared with the prior art, by real-time updating and adjusting the information in the spatio-temporal network, the path planning method used by the present invention can effectively handle the problem of multi-agent collaborative operations in a dynamic environment, improving the flexibility of the multi-agent system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is the grid network diagram of the present invention;

[0045] Figure 2 is the partial grid network diagram of the present invention;

[0046] Figure 3 is the directed spatio-temporal road network diagram of the present invention;

[0047] Figure 4 is the overall flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] In this embodiment, a multi-agent conflict-free path planning method based on a directed spatio-temporal network, as Figure 4 shown, is carried out according to the following steps:

[0049] Step 1: Construct a directed spatio-temporal network and initialize the information of the agent group;

[0050] Step 1.1: Construct a grid network;

[0051] Obtain environmental information and construct a grid network with a scale of where represents the grid set, and and represents the th grid in ​is the total number of rows or columns; it is defined that the grid is only connected to the grids in the upper, lower, left, and right directions that are directly adjacent to itself, that is, if the th grid is the neighbor grid of the grid , then ; represents the set of directed road segments between grids, and , represents the th grid and the th grid between the directed road segments; represents the set of movement weights between connected grids, and , is the th grid and the th grid between the directed road segment movement weight, if the th grid and the th grid between the directed road segments are all obstacle-free, then let ; if the th grid and the th grid between the directed road segments have at least one obstacle, then let ; In specific implementation, the local grid map constructed according to the above method is as Figure 1 shown, the map size is 20×20, where the white grid represents the feasible area and the black grid represents the obstacle.

[0052] Step 1.2: Construct a directed spatio-temporal road network ;

[0053] Determine the estimated search completion time as , and is discretized with a unit step size to obtain the time set ; Based on the time set and the grid network construct a directed spatio-temporal road network with a scale of , represents the grid set under the time set , , where represents the grid set at the moment, and , represents the ​A grid; it is defined that the grid is only connected to the directly adjacent grids in adjacent time steps, that is, if the grid is the neighbor grid of the grid , then ; represents the set of directed road segments between grids at different times, and , represents the th grid and the th grid at time represents the set of action weights of the directed road segments between grids at different times, and , represents the action weight of the directed road segment ; In specific implementation, the directed graph constructed according to the above method is as shown in Figure 2 and Figure 3 , Figure 2 is the local grid network, Figure 3 is the directed spatio-temporal road network constructed based on Figure 2 .

[0054] Step 2: Set basic information:

[0055] Step 2.1: Set the basic information of the agent group:

[0056] Let the agent group be denoted as , represents the total number of agents, represents the kth agent; Let represent the th agent 's starting point, , represents the set of starting point grids of the agent group ; Let represent the th agent 's end point; , represents the set of end point grids of the agent group

[0057] Step 2.2: Calculate the shortest path weight of the agent group :

[0058] Use the Floyd-Warshall algorithm to calculate the in the directed spatio-temporal network th agent in​ The shortest path weight from the th grid to the end point is obtained, so as to obtain a directed spatio-temporal network and the set of shortest path weights from the grid where each agent is located to its respective end point at each moment in the moment set in . .

[0059] Step 3: Based on the directed spatio-temporal network , construct the forward direction decision model for the agent group at time:

[0060] Step 3.1: Use Equation (1) to construct the objective function of the forward direction decision model for the agent group at time: :

[0061] (1)

[0062] In Equation (1), represents whether the k-th agent passes through the directed road section at time. If so, let = 1; otherwise, let = 0; represents the th grid at time; represents the

[0063] th grid at

[0064] (2)

[0065] In Equation (2), represents whether the th agent passes through the directed road section at time. If so, let = 1; otherwise, let = 0; represents whether the th agent passes through the directed road section at ; If so, then let , otherwise, let , be the corresponding decision variable, indicating whether the k-th agent is at the i-th grid at time . If so, then let = 1; otherwise, let = 0.

[0066] Step 4: According to the directed spatio-temporal road network and the forward direction decision model of the agent group , solve the path planning scheme of multi-agent:

[0067] Step 4.1: Let = 1;

[0068] Step 4.2: Let = 1;

[0069] Step 4.3: Define and initialize the set of directed road segments of the k-th agent on the directed spatio-temporal network as an empty set, i.e., = ;

[0070] Step 4.4: According to the forward grid that the k-th agent can choose at time , obtain the forward strategy candidate set of the k-th agent as = ; where represents the k-th agent at time in the -th grid;

[0071] In specific implementation, according to the forward grid that the k-th agent can choose at time , i.e., the neighbor grid of the directed spatio-temporal road network , obtain the forward strategy candidate set of the k-th agent as , where represents the k-th agent at time in the -th grid; thus obtaining the corresponding k-th agent Candidate set of decision variables 。

[0072] Step 4.5: Traverse the k-th agent from the candidate set of forward strategies ,and substitute the decision variables corresponding to and the decision variables corresponding to both assigned as 1 into the forward direction decision model in Step 3 for solution, and obtain the optimal strategy set of the k-th agent ;

[0073] In specific implementation, traverse the candidate set of forward strategies of the k-th agent ,and assign the decision variables corresponding to and the decision variables corresponding to both as 1; from the candidate set of forward strategies of the k-th agent use the Cartesian product to obtain the candidate set of overall forward strategies of the agent group , ,…, ,… ,from the set of decision variables of the k-th agent use the Cartesian product to obtain the candidate set of overall decision variables of the agent group , ,…, ,… ;

[0074] Substitute the candidate solution into the model in Step 3 to obtain the optimal objective function value corresponding to the feasible solution, denoted as ,and denote the optimal solution set as ,so as to obtain the optimal strategy set of the k-th agent 。

[0075] Step 4.6: If = ,it means that the k-th agent will move from grid to at time ,and add the directed road segment to the set of directed road segments At the end, let = ;

[0076] Step 4.7: If the k-th agent at the arrival grid at the moment is the end point , it means that the final set of road segment sets of the k-th agent is obtained, and it is combined with to form the path planning scheme of the k-th agent ), and step 4.8 is executed; otherwise, step 4.9 is executed;

[0077] Step 4.8: Let be assigned to , and return to step 4.4 to execute sequentially until ;

[0078] Step 4.9: After assigning k + 1 to k, return to step 4.3 until k > K, so as to output the path planning scheme R of the agent group .

[0079] In this embodiment, an electronic device includes a memory and a processor. 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.

[0080] In this embodiment, a computer-readable storage medium stores a computer program on the computer-readable storage medium. When the computer program is run by a processor, it executes the steps of the above method.

Claims

1. A multi-agent conflict-free path planning method based on directed spatiotemporal networks and forward direction decision-making, characterized in that: The steps are as follows: Step 1: Construct a directed spatiotemporal network; Step 2: Set basic information: Step 2.1: Set the basic information of the agent group: Let the agent group be , represents the total number of agents, represents the kth agent; let Indicates Agent The starting point, , Represents an agent group The starting grid set of Indicates Agent The end point; , Represents an agent group The destination grid set of ; Step 2.2: Calculate the agent group The shortest path weight is: Computing directed space-time networks using the Floyd-Warshall algorithm Middle Agent exist Time from the Grid arrive The end point The shortest path weight , thus obtaining a directed space-time network Mid-time collection The shortest path weight set of each agent from the grid to its destination ; Step 3: Based on directed space-time network and C, construct Agent group at the moment The forward direction decision model: Step 4: Based on the directed space-time network and Agent Group The forward direction decision model is used to solve the path planning solution for multiple agents.

2. According to claim 1, a multi-agent conflict-free path planning method based on directed spatiotemporal network and forward direction decision-making is characterized in that: The step 1 is carried out as follows: Step 1.1: Build a grid network; Get environment information and build a Grid network ,in, represents a collection of rasters, and , express The grids, is the total number of rows or columns; the grid is defined to be connected only with the grids directly adjacent to it in the upper, lower, left, and right directions. Grid For Grid The neighbor grid of ; represents a set of directed road segments between grids, and , Indicates Grid With Grid Directed sections between; represents the set of movement weights between connected grids, and , For the Grid With Grid The moving weight of the directed road segment between Grid With Grid There are no obstacles on the directional sections between ; If Grid With Grid There is at least one obstacle in the directed road segment between ; Step 1.2: Construct a directed space-time network ; Determine the estimated search completion time ,Will After discretization with a unit step, we get the time set ; Time-based collection and raster networks The construction scale is Directed space-time network , Represents a time collection The following grid collection, ,in, express The grid set at the time, and , express The moment grids; define the grid to be connected only to the directly adjacent grids of the adjacent time steps, that is, if the grid Is a grid The neighbor grid of ; represents the set of directed road segments between grids at different times, and , express The moment Grid and The moment Grid Directed sections between; represents the action weight set of the directed road segments between grids at different times, and , Indicates a directed segment action weight.

3. The multi-agent conflict-free path planning method based on directed spatiotemporal network and forward direction decision according to claim 2, characterized in that: The step 3 is carried out as follows: Step 3.1: Use formula (1) to construct Agent group at the moment The objective function of the forward direction decision model is : (1) In formula (1), represents the kth agent exist Whether the moment passes through a directed segment If so, then let =1, otherwise, let =0; express The moment grids; express The moment grids; Step 3.2: Use equation (2) to construct the constraint conditions of the forward direction decision model: (2) In formula (2), Indicates Agent exist Whether the moment passes through a directed segment If so, then let =1, otherwise, let =0; Indicates Agent exist Whether the moment passes through a directed segment If so, then , otherwise, let , for The corresponding decision variable represents the kth agent exist Whether the moment is in the i-th grid If so, then let =1; otherwise, let =0.

4. The multi-agent conflict-free path planning method based on directed spatiotemporal network and forward direction decision according to claim 3, characterized in that: The step 4 is carried out as follows: Step 4.1: Order =1; Step 4.2: Order =1; Step 4.3: Define and initialize the kth agent In directed space-time networks The set of directed segments on Set to an empty set, that is = ; Step 4.4: According to the kth agent exist time Selectable forward grid , get the kth agent The forward strategy candidate set = ;in, represents the kth agent The moment is in grids; Step 4.5: Traverse the kth agent The forward strategy candidate set ,Will The corresponding decision variable and The corresponding decision variable After all values ​​are assigned to 1, they are substituted into the forward direction decision model in step 3 to solve and obtain the kth agent corresponding to the minimum objective function value. The optimal strategy set ; Step 4.6: If = , then it means the kth agent exist The moment will be from the grid Move to , and the directed segment Add to the directed segment collection At the end of = ; Step 4.7: If the kth agent exist Arrival grid at time For the end , then it means that the kth agent is obtained The final set of road segments , and with Form the kth agent Path planning scheme ), and execute step 4.8; otherwise, execute step 4.9; Step 4.8: Order Assign to , return to step 4.4 and execute sequentially until until; Step 4.9: After assigning k+1 to k, return to step 4.3 until k>K, thereby outputting the agent group The path planning scheme R .

5. 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-agent conflict-free path planning method described in any one of claims 1-4, and the processor is configured to execute the program stored in the memory.

6. 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-agent conflict-free path planning method described in any one of claims 1 to 4 are executed.

Images