WO2023207803A1 - Multi-agent navigation control method and device based on gene regulatory network, and medium - Google Patents

Abstract

A multi-agent navigation control method and device based on a gene regulatory network, and a medium. The method comprises: acquiring a planar map where a plurality of agents are located (S100); converting the planar map into a grid map, and determining concentration information corresponding to each grid in the grid map (S200); determining an optimal path for an agent from a current grid to a target grid according to the concentration information corresponding to each grid in the grid map (S300); in the process that an agent to be navigated moves along the optimal path, screening for a vacant grid from eight neighborhood grids adjacent to the current grid, and determining a concentration evaluation value, a first distance evaluation value and a second distance evaluation value of each vacant grid (S400); determining a cost evaluation value of each vacant grid according to the concentration evaluation value, the first distance evaluation value and the second distance evaluation value, and taking a vacant grid having a minimum cost evaluation value as a grid to be entered (S500); and determining a distance between the grid to be entered and a grid where the obstacle is located, determining a running state of the agent on the basis of a relationship between an obstacle distance and a preset safety distance, and controlling, according to the running state of the agent, the agent to move towards a target position (S600). An optimal navigation path can be adaptively found out in real time.

Description

Multi-agent navigation control method, equipment and medium based on gene regulatory network Technical field

The present invention relates to but is not limited to the field of navigation technology, and in particular, to a multi-agent navigation control method, equipment and medium based on a gene regulatory network.

Background technique

At present, the A* algorithm is widely used in path planning tasks involved in intelligent navigation systems. The operation process of the A* algorithm requires the cost calculation of all grids on the map in sequence, that is, the movement costs of all possible routes need to be calculated. The calculation cost is high. In addition, when the position of obstacles or target points in the map changes, the obstacles or target points may fall on points that have been traversed, and the A* algorithm will no longer be applicable.

Therefore, it is necessary to improve the existing intelligent navigation system to find the optimal navigation path in real time and adaptively.

Contents of the invention

The following is an overview of the topics described in detail in this article. This summary is not intended to limit the scope of the claims.

Embodiments of the present invention provide a multi-agent navigation control method, equipment and medium based on a gene regulatory network, which can adaptively find the optimal navigation path in real time.

In a first aspect, embodiments of the present invention provide a multi-agent navigation control method based on a gene regulatory network, including:

Step S100, obtain the plane map where multiple agents are located;

Step S200: Convert the flat map into a grid map, and determine the concentration information corresponding to each grid in the grid map; wherein the concentration information includes the concentration information field of the target position and the concentration information of the obstacle position. field;

Step S300: Determine the current grid and target grid of the agent to be navigated, and determine the optimal path for the agent from the current grid to the target grid based on the concentration information corresponding to each grid in the grid map. ; Wherein, the current grid is the grid where the agent is currently located, and the target grid is the grid where the target position is located in the grid map;

Step S400: While the agent is moving along the optimal path, vacant grids are screened out from the eight neighborhood grids adjacent to the current grid, and the concentration evaluation value of each vacant grid is determined. , the first distance evaluation value and the second distance evaluation value; wherein, the vacant grid is a grid passable by the agent, and the concentration evaluation value is the grid where the agent is currently located and the vacant grid. The difference between the concentration information of the grids, the first distance evaluation value is the distance difference between the agent's previous grid and the vacant grid, the second distance evaluation value is the vacant grid The distance difference between the grid and the target grid;

Step S500: Determine the cost evaluation value of each vacant grid according to the concentration evaluation value, the first distance evaluation value and the second distance evaluation value, and use the vacant grid with the smallest cost evaluation value as the grid to be entered;

Step S600: Determine the obstacle distance between the grid to be entered and the grid where the obstacle is located, determine the operating state of the intelligent agent based on the obstacle distance, and control the intelligent agent to move toward the target according to the operating state of the intelligent agent. Position forward.

In some embodiments, converting the flat map into a grid map and determining the concentration information corresponding to each grid in the grid map includes:

Determine the grid where the target position is located in the grid map, the grid where the obstacle is located, and the movable range boundary of the agent;

Import the grid map into the gene regulation network model to generate a target concentration map and an obstacle concentration map; wherein each grid in the target concentration map contains the concentration information field of the target location, and the target concentration map contains Each grid contains the concentration information field of the obstacle location;

The target concentration map and the obstacle concentration map are coupled according to corresponding grids to obtain concentration information corresponding to each grid in the grid map.

In some embodiments, the calculation formula relied upon to generate the target concentration map is:

Among them, p ₁ is the concentration information field generated by the target, b ₁ is the adjustable parameter that affects the concentration value of each grid in the target concentration map, and v ₁ is the concentration diffusion factor, which is used to adjust the mapping relationship between distance parameters and concentration parameters. , r ₁ is the relative distance between the center of each grid in the target concentration map and the location of the target;

The calculation formula relied on to generate the obstacle concentration map is:

Among them, p ₂ is the concentration information field generated by the obstacle, b ₂ is the adjustable parameter that affects the concentration value of each grid in the obstacle concentration map, v ₂ is the concentration diffusion factor, and r ₂ is each grid in the target concentration map. The relative distance between the grid center and the location of the obstacle;

The calculation formula for the concentration information corresponding to each grid in the grid map is:

Among them, g ₁ is the morphological gradient space presented by the target concentration map, g ₂ is the morphological gradient space presented by the obstacle concentration map, g ₃ is the morphology presented by coupling the concentration information field of the target position and the concentration information field of the obstacle position. Gradient space, θ ₁ is an adjustable parameter that affects the concentration value range of the target concentration map, k ₁ is an adjustable parameter that affects the concentration difference between adjacent grids of the target concentration map, θ ₂ is the concentration that affects the obstacle concentration map The adjustable parameter of the value range, k ₂ is the adjustable parameter that affects the concentration difference between adjacent grids in the obstacle concentration map, θ ₃ is the adjustable parameter that affects the concentration value range of the grid in the grid map, k ₃ It is an adjustable parameter that affects the concentration difference between adjacent grids in the raster map.

In some embodiments, the calculation formula of the cost evaluation value is:

Among them, P _e is the cost evaluation value for estimating that the agent to be navigated can reach the grid at the next moment, C _e is the concentration evaluation value for estimating that the agent to be navigated can reach the grid at the next moment, D _p In order to evaluate the first distance evaluation value that the agent to be navigated can reach to the grid at the next moment, D _t is to evaluate the second distance evaluation value that the agent to be navigated can reach the grid at the next moment, a, b and c are both weight parameters, C _N is the concentration of the grid that the agent to be navigated can reach at the next moment. Degree value, C _M is the concentration value of the grid where the agent to be navigated is currently located, C _min is the preset minimum optional grid concentration, C _max is the preset maximum optional grid concentration, N _x is the abscissa coordinate of the grid that the agent to be navigated can reach at the next moment, N _y is the ordinate coordinate of the grid that the agent to be navigated can reach at the next moment, and P _x is the coordinate of the grid that the agent to be navigated can reach at the next moment. The abscissa coordinate of the grid where the navigation agent was located at the last moment, P _y is the ordinate coordinate of the grid where the agent to be navigated was located at the last moment, T _x is the abscissa coordinate of the target location, and T _y is the target location. ordinate of .

In some embodiments, determining the obstacle distance between the grid to be entered and the grid where the obstacle is located, and determining the operating status of the agent based on the relationship between the distance and a preset safety distance includes:

Step S610: Determine whether the obstacle distance is greater than the safety distance. If the obstacle distance is greater than the safety distance, execute step S620; otherwise, execute step S660;

Step S620: Control the intelligent agent to move forward toward the target position in a first state; wherein the first state is to move forward while avoiding obstacles;

Step S630: During the process of the intelligent agent moving forward in the first state, if the trigger points marked by other intelligent agents are detected, the operating state of the intelligent agent is changed from the first state to the fourth state, and the intelligent agent is controlled. The agent moves toward the target position in the fourth state; wherein the fourth state is to move along the shortcut;

Step S640: During the process of the agent moving from the first state to the fourth state, if the end points marked by other agents are detected, the running state of the agent is changed from the fourth state to the fourth state. In one state, the intelligent agent is controlled to move toward the target position in the first state; if the obstacle is detected to be within a safe distance, the operating state of the intelligent agent is transferred from the fourth state to the second state. , controlling the intelligent agent to move forward toward the target position in the second state; wherein the second state is to move along the edge of the obstacle;

Step S650: During the process of the intelligent agent moving from the fourth state to the second state, if it is detected that the obstacle distance is greater than the safe distance, the operating state of the intelligent agent is converted from the second state to the second state. In the fourth state, the intelligent agent is controlled to move toward the target position in the fourth state;

Step S660: Change the running state of the intelligent agent from the first state to the second state, and control the intelligent agent to move toward the target position in the second state;

Step S670: During the process of the intelligent agent changing from the first state to moving forward in the second state, if it is detected that the obstacle distance is greater than the safe distance, the operating state of the intelligent agent is changed from the second state to the second state. In the first state, the intelligent agent is controlled to move forward toward the target position in the first state; if the obstacle distance is detected to be less than the field boundary distance, the location where the obstacle distance is detected to be less than the field boundary distance is marked as a trigger. After clicking, the running state of the intelligent agent is changed from the second state to the third state, and the intelligent agent is controlled to move toward the target position in the third state; wherein the third state is to follow the original path from the trigger point return.

In some embodiments, the method further includes:

During the process of the agent moving forward in the fourth state, if a trigger point is detected, the current operating state and the previous operating state of the agent marking the trigger point are determined;

If the agent marking the trigger point does not move forward according to the third state in the current running state and the previous running state, step S500 is executed;

If the current running state of the agent marking the trigger point is the third state, the agent will move forward in the opposite direction to the current forward path;

If the previous operating state of the agent that marked the trigger point was the third state and is not currently moving forward in the third state, then the agent will follow the agent that marked the trigger point;

Determine whether the agent that marked the trigger point has also marked the end point. If the agent that marked the trigger point has also marked the end point, then the agent follows the agent that marked the trigger point and moves toward the end point.

In a second aspect, embodiments of the present invention also provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following is implemented: The multi-agent navigation control method based on the gene regulatory network described in the first aspect.

In a third aspect, embodiments of the present invention also provide a computer-readable storage medium that stores computer-executable instructions, and the computer-executable instructions are used to execute the multi-agent method based on the gene regulation network as described in the first aspect. Navigation control methods.

Embodiments of the present invention include: obtaining a plane map where multiple agents are located; converting the plane map into a grid map, and determining concentration information corresponding to each grid in the grid map; wherein the concentration information includes The concentration information field of the target position and the concentration information field of the obstacle position; determine the current grid and target grid of the agent to be navigated, and determine the agent according to the concentration information corresponding to each grid in the grid map The optimal path from the current grid to the target grid; wherein, the current grid is the grid where the agent is currently located, and the target grid is the grid where the target position is located in the grid map; in While the agent is moving along the optimal path, it selects vacant grids from the eight neighborhood grids adjacent to the current grid, and determines the concentration evaluation value and first distance of each vacant grid. The evaluation value and the second distance evaluation value; wherein, the vacant grid is a grid passable by the agent, and the concentration evaluation value is the concentration of the grid where the agent is currently located and the vacant grid. The difference between the information, the first distance evaluation value is the distance difference between the agent's previous grid and the vacant grid, and the second distance evaluation value is the distance between the vacant grid and the vacant grid. The distance difference between the target grids; determine the cost evaluation value of each vacant grid according to the concentration evaluation value, the first distance evaluation value and the second distance evaluation value, and use the vacant grid with the smallest cost evaluation value as the candidate grid. Enter the grid; determine the obstacle distance between the grid to be entered and the grid where the obstacle is located, determine the operating state of the intelligent agent based on the obstacle distance, and control the direction of the intelligent agent according to the operating state of the intelligent agent. Move forward to the target location.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and obtained by the structure particularly pointed out in the written description, claims and appended drawings.

Description of the drawings

The accompanying drawings are used to provide a further understanding of the technical solutions of the present invention, and constitute a part of the description. Together with the embodiments of the present invention, they are used to explain the technical solutions of the present invention, and do not constitute a limitation of the technical solutions of the present invention.

Figure 1 is a schematic flow chart of a multi-agent navigation control method based on a gene regulatory network in an embodiment of the present invention;

Figure 2 is a specific schematic diagram of the target concentration map applied in the first scenario according to the embodiment of the present invention;

Figure 3 is a specific schematic diagram of the obstacle concentration map applied in the first scenario according to the embodiment of the present invention;

Figure 4 is a specific schematic diagram of the overall concentration map applied in the first scenario according to the embodiment of the present invention;

Figure 5 is a specific schematic diagram of multi-agent navigation applied in the first scenario according to the embodiment of the present invention;

Figure 6 is a specific schematic diagram of the target concentration map applied in the second scenario according to the embodiment of the present invention;

Figure 7 is a specific schematic diagram of the obstacle concentration map applied in the second scenario according to the embodiment of the present invention;

Figure 8 is a specific schematic diagram of the overall concentration map applied in the second scenario according to the embodiment of the present invention;

Figure 9 is a specific schematic diagram of multi-agent navigation applied in the second scenario according to the embodiment of the present invention;

Figure 10 is a specific schematic diagram of the target concentration map applied in the third scenario according to the embodiment of the present invention;

Figure 11 is a specific schematic diagram of the obstacle concentration map applied in the third scenario according to the embodiment of the present invention;

Figure 12 is a specific schematic diagram of the overall concentration map applied in the third scenario according to the embodiment of the present invention;

Figure 13 is a specific schematic diagram of multi-agent navigation applied in the third scenario according to the embodiment of the present invention;

Figure 14 is a specific illustration of the application of the obstacle avoidance control strategy in the fourth scenario of single agent navigation in the embodiment of the present invention. intention;

Figure 15 is a specific schematic diagram of the obstacle avoidance control strategy in the embodiment of the present invention applied to single agent navigation in the fifth scenario;

Figure 16 is a specific schematic diagram of the obstacle avoidance control strategy in the embodiment of the present invention applied to multi-agent navigation in the sixth scenario;

Figure 17 is a specific schematic diagram of the obstacle avoidance control strategy in the embodiment of the present invention applied to multi-agent navigation in the seventh scenario;

Figure 18 is a structural diagram of an electronic device provided by another embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not intended to limit the present invention.

It should be noted that although a logical sequence is shown in the flowchart, in some cases, the steps shown or described may be performed in an order different from that in the flowchart. The terms "first", "second", etc. in the description, claims or the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

The embodiments of the present invention will be further described below with reference to the accompanying drawings.

As shown in Figure 1, Figure 1 is a flow chart of a multi-agent navigation control method based on a gene regulatory network provided by one embodiment of the present invention. The method includes but is not limited to the following steps:

Step S100, obtain the plane map where multiple agents are located;

Step S200: Convert the flat map into a grid map, and determine the concentration information corresponding to each grid in the grid map; wherein the concentration information includes the concentration information field of the target position and the concentration information of the obstacle position. field;

It should be noted that the grid in the grid map is the smallest square that allows a single agent to rotate once;

Step S300: Determine the current grid and target grid of the agent to be navigated, and determine the optimal path for the agent from the current grid to the target grid based on the concentration information corresponding to each grid in the grid map. ; Wherein, the current grid is the grid where the agent is currently located, and the target grid is the grid where the target position is located in the grid map;

Step S400: While the agent is moving along the optimal path, vacant grids are screened out from the eight neighborhood grids adjacent to the current grid, and the concentration evaluation value of each vacant grid is determined. , the first distance evaluation value and the second distance evaluation value; wherein, the vacant grid is a grid passable by the agent, and the concentration evaluation value is the grid where the agent is currently located and the vacant grid. The difference between the concentration information of the grids, the first distance evaluation value is the distance difference between the agent's previous grid and the vacant grid, the second distance evaluation value is the vacant grid The distance difference between the grid and the target grid;

It should be noted that in the grid map, with the grid where the agent to be navigated is located as the center, an associated nine-square grid can be established. The grid where the agent is located has 8 adjacent grids (neighborhood grids) ; Confirm again whether these 8 neighborhood grids are occupied, that is, there are no obstacles or other agents on the vacant grids. Use the unoccupied domain grids as vacant grids to get the result that the agent can Passage grid. The previous grid of the agent refers to the grid that the agent has just experienced before the current grid. In one embodiment, based on the mutual communication relationship between multiple agents, the agent to be navigated can obtain the current location information of each other agent in the overall concentration map and the locations of the remaining eight grids. The area ranges are compared one by one, and grids occupied by other agents are obtained and eliminated, thereby preventing the agent to be navigated from colliding with any other agents.

In addition, if the agent to be navigated determines that the remaining 8 grids have been occupied by other agents, the The agent to be navigated regards the current grid elimination process as an abnormal event, and at the same time enters the cyclic position query and occupancy judgment operations to wait for the agent in any one or more grids to start moving until it leaves the association. Nine-square grid, and then re-execute the grid elimination task.

Step S500: Determine the cost evaluation value of each vacant grid according to the concentration evaluation value, the first distance evaluation value and the second distance evaluation value, and use the vacant grid with the smallest cost evaluation value as the grid to be entered;

Step S600: Determine the obstacle distance between the grid to be entered and the grid where the obstacle is located, determine the operating state of the intelligent agent based on the obstacle distance, and control the intelligent agent to move toward the target according to the operating state of the intelligent agent. Position forward.

In the embodiment provided by the present invention, the optimal path for the agent to move to the target is adaptively generated based on the concentration intensity in the grid map. It is not necessary to calculate all feasible paths on the map in order to determine the optimal path for the agent. This can reduce computing costs and make the implementation process simpler. During the movement, the agent uses the cost evaluation value determined by the concentration information and distance information to select and formulate the next moving position, which has strong real-time performance. The operating status of the agent is determined based on the relationship between the distance and the preset safety distance, which can quickly navigate to the target in a complex environment and solve the problem of collision avoidance between group robots and collision avoidance between robots and obstacles. Problems, as well as problems such as adaptively finding the optimal navigation path, are suitable for scenarios where obstacles or target point positions change.

In addition, in one embodiment, in step S200 in the embodiment shown in FIG. 1 , converting the flat map into a grid map and determining the concentration information corresponding to each grid in the grid map includes: :

Step S210: Determine the grid where the target position is located, the grid where the obstacle is located, and the movable range boundary of the agent in the grid map;

Step S220: Import the grid map into the gene regulation network model to generate a target concentration map and an obstacle concentration map; wherein each grid in the target concentration map contains the concentration information field of the target location, and the target concentration Each grid in the map contains a concentration information field for the location of obstacles;

Step S230: Couple the target concentration map and the obstacle concentration map according to corresponding grids to obtain concentration information corresponding to each grid in the grid map.

In some embodiments, first, a gene regulation network is constructed. The gene regulation network used in this embodiment is the gene regulation network disclosed in Publication No. CN112684700A; by obtaining multiple basic elements in the basic element library, according to the multiple The topological structure obtained by combining the basic elements forms a gene regulatory network model; then, the grid map is imported into the gene regulatory network model to generate a target concentration map and an obstacle concentration map; finally, the target concentration map and the obstacle concentration map are The object concentration maps are coupled according to corresponding grids, and the coupled grid map contains the target information and the obstacle information.

Specifically, the grid map is marked with multiple points around the target, multiple points around obstacles, and multiple points on the map boundary. N points around the target (that is, the edge of the area where the target is located) are calculated. The protein concentration of the grid occupied by a point in the grid map is then superimposed to form a target concentration map; similarly, all obstacles surrounding all obstacles (i.e. all obstacles) are obtained in the flat map M points at the edge of the area where each obstacle in the object is located) and K points at the boundary of the movable range, calculate the protein concentration of the grid occupied by each point in the flat map, and then add M+ The protein concentrations of K points are superimposed to form an obstacle concentration map.

Based on the fact that the target information and the obstacle information are changeable in real time, when the embodiment of the present invention is applied to different changing scenarios, multiple sets of corresponding different concentration maps can be generated, as follows:

In the first scene, the target concentration map shown in Figure 2 and the obstacle concentration map shown in Figure 3 are generated, and then coupled to obtain the overall concentration map shown in Figure 4. When the embodiment of the present invention is applied to the first scene When multiple agents are implemented to navigate to the target location, the navigation diagram shown in Figure 5 is obtained;

In the second scene, the target concentration map shown in Figure 6 and the obstacle concentration map shown in Figure 7 are generated, and then coupled The overall concentration map is obtained as shown in Figure 8. When the embodiment of the present invention is applied to the second scenario to realize multiple agents navigating to the target location, a navigation schematic diagram as shown in Figure 9 is obtained;

In the third scene, the target concentration map shown in Figure 10 and the obstacle concentration map shown in Figure 11 are generated, and then coupled to obtain the overall concentration map shown in Figure 12. When the embodiment of the present invention is applied to the third scene When multiple agents are implemented to navigate to the target location, a navigation schematic diagram as shown in Figure 13 is obtained, thereby verifying the feasibility of the embodiment of the present invention.

In one embodiment, the calculation formula relied upon to generate the target concentration map is:

Among them, p ₁ is the concentration information field generated by the target, b ₁ is the adjustable parameter that affects the concentration value of each grid in the target concentration map, and v ₁ is the concentration diffusion factor, which is used to adjust the mapping relationship between distance parameters and concentration parameters. , r ₁ is the relative distance between the center of each grid in the target concentration map and the location of the target;

The calculation formula relied on to generate the obstacle concentration map is:

Among them, p ₂ is the concentration information field generated by the obstacle, b ₂ is the adjustable parameter that affects the concentration value of each grid in the obstacle concentration map, v ₂ is the concentration diffusion factor, and r ₂ is each grid in the target concentration map. The relative distance between the grid center and the location of the obstacle;

The calculation formula for the concentration information corresponding to each grid in the grid map is:

Among them, g ₁ is the morphological gradient space presented by the target concentration map, g ₂ is the morphological gradient space presented by the obstacle concentration map, g ₃ is the morphology presented by coupling the concentration information field of the target position and the concentration information field of the obstacle position. Gradient space, θ ₁ is an adjustable parameter that affects the concentration value range of the target concentration map, k ₁ is an adjustable parameter that affects the concentration difference between adjacent grids of the target concentration map, θ ₂ is the concentration that affects the obstacle concentration map The adjustable parameter of the value range, k ₂ is the adjustable parameter that affects the concentration difference between adjacent grids in the obstacle concentration map, θ ₃ is the adjustable parameter that affects the concentration value range of the grid in the grid map, k ₃ It is an adjustable parameter that affects the concentration difference between adjacent grids in the raster map. Preferably, in the embodiment of the present invention, parameter b ₁ is set to 1, parameter b ₂ is set to 1.2, parameters v ₁ and v ₂ are both set to 1, and parameters θ ₁ , θ ₂ , and θ ₃ are all set to 1. The value is 0, the value ranges of θ ₁ , θ ₂ and θ ₃ are all [0, 1], the parameters k ₁ , k ₂ and k ₃ are all set to 1, and the values of k ₁ , k ₂ and k ₃ are The value range is [0, 2].

In some embodiments, the calculation formula of the cost evaluation value is:

Among them, P _e is the cost evaluation value for estimating that the agent to be navigated can reach the grid at the next moment, C _e is the concentration evaluation value for estimating that the agent to be navigated can reach the grid at the next moment, D _p In order to evaluate the first distance evaluation value that the agent to be navigated can reach to the grid at the next moment, D _t is to evaluate the second distance evaluation value that the agent to be navigated can reach the grid at the next moment, a, b and c are both weight parameters, C _N is the concentration value of the grid that the agent to be navigated can reach at the next moment, C _M is the concentration value of the grid where the agent to be navigated is at the current moment, C _min is the preset minimum value of the optional grid concentration, C _max is the preset maximum value of the optional grid concentration, N _x is the abscissa coordinate of the grid that the agent to be navigated can reach at the next moment, N _y is the ordinate of the grid where the agent to be navigated can be evaluated at the next moment, P _x is the abscissa of the grid where the agent to be navigated was at the previous moment, and P _y is the intelligence to be navigated. The ordinate of the grid where the body is located at the moment, T _x is the abscissa of the target location, and T _y is the ordinate of the target location. Preferably, in this embodiment of the present invention, parameter a is set to 0.3, parameter b is set to 0.35, and parameter c is set to 0.35.

Among them, the purpose of selecting the concentration evaluation value in the embodiment of the present invention is to enable the agent to be navigated to move along the concentration gradient, and the purpose of selecting the first distance evaluation value is to enable the agent to be navigated to move as effectively as possible. That is to ensure that the grid that the agent to be navigated at the next moment can reach is farther from the grid at the previous moment. The purpose of selecting the second distance evaluation value is to enable the agent to be navigated to move in the direction of the target location. move.

Taking as an example that there are only grid A, grid B and grid C in the associated nine-square grid that are not occupied by any agent, the grid screening process is explained as follows: first, calculate grid A and the intermediate grid according to the above formula The cost evaluation value between grid B and the intermediate grid is P _e,A , the cost assessment value between grid B and the intermediate grid is P _e,B , and the cost assessment value between grid C and the intermediate grid is P _e,C , If it is determined that P _e,A <P _e,B <P _e,C , then the center point of the grid A is designated as the movement position of the agent to be navigated. If it is determined that P _e,A =P _{e, B} <P _e,C , then randomly select a grid from grid A and grid B and designate its center point as the movement position of the agent to be navigated.

In addition, in one embodiment, in step S600 in the embodiment shown in Figure 1, the determination of the obstacle distance between the grid to be entered and the grid where the obstacle is located is based on the distance and the preset safety distance. Relationships determine the operating status of the agent, including:

Step S610: Determine whether the obstacle distance is greater than the safety distance. If the obstacle distance is greater than the safety distance, execute step S620; otherwise, execute step S660;

Step S620: Control the intelligent agent to move forward toward the target position in a first state; wherein the first state is to move forward while avoiding obstacles;

Step S630: During the process of the intelligent agent moving forward in the first state, if the trigger points marked by other intelligent agents are detected, the operating state of the intelligent agent is changed from the first state to the fourth state, and the intelligent agent is controlled. The agent moves toward the target position in the fourth state; wherein the fourth state is to move along the shortcut;

Step S640: During the process of the agent moving from the first state to the fourth state, if the end points marked by other agents are detected, the running state of the agent is changed from the fourth state to the fourth state. In one state, the intelligent agent is controlled to move toward the target position in the first state; if the obstacle is detected to be within a safe distance, the operating state of the intelligent agent is transferred from the fourth state to the second state. , controlling the agent to move forward toward the target position in the second state; wherein the second state is to move along the edge of the obstacle; the end point is the exit of the agent;

Step S650: During the process of the intelligent agent moving from the fourth state to the second state, if it is detected that the obstacle distance is greater than the safe distance, the operating state of the intelligent agent is converted from the second state to the second state. In the fourth state, the intelligent agent is controlled to move toward the target position in the fourth state;

Step S660: Change the running state of the intelligent agent from the first state to the second state, and control the intelligent agent to move toward the target position in the second state;

Step S670: During the process of the agent moving from the first state to the second state, if the obstacle is detected If the distance is greater than the safe distance, then the operating state of the intelligent agent is changed from the second state to the first state, and the intelligent agent is controlled to move toward the target position in the first state; if the distance to the obstacle is detected to be less than the field boundary distance, then mark the location where the obstacle distance is less than the field boundary distance as the trigger point, change the running state of the intelligent agent from the second state to the third state, and control the intelligent agent to press the third state Move forward toward the target position; wherein the third state is to return from the trigger point along the original path.

It should be noted that the obstacle distance is obtained through real-time detection and is a variable; the safety distance and field boundary distance are preset quantifications. The safety distance is greater than the field boundary distance. By combining the obstacle distance, safety distance and field boundary distance, The relationship between the agents is used to determine the operating status of the agent and ensure that the agent reaches the target location as quickly as possible while avoiding obstacles.

In this embodiment, the intelligent agent has four operating states. When encountering different situations, it will switch back and forth between various operating states. In all running states, the agent cannot select the grid where other agents are located when selecting the next position. If there are other agents in the selected grid, the agent abandons the optimal choice and selects a feasible grid near the optimal grid. The agent calculates and selects the vacant grid with the largest cost evaluation value for navigation. A safe distance is set between the agent and obstacles. Set the field boundary distance from the agent to the boundary. The agent sensor has a certain detection range, and the agent can only detect obstacles or boundaries within the detection range. In the first state, the agent realizes basic navigation by evaluating the cost evaluation value of the surrounding grid, and in the second state, it is used to instruct the agent to walk along the edge of the obstacle. When the agent enters this state, it leaves a "trigger point" when faced with a decision, randomly choosing possible directions to move along the edge of the obstacle. If the agent finds the exit, the agent will set the "end point" to remind other agents that they can move directly to the "end point"; if the agent does not find the exit but detects the field boundary, it means that the agent is at the "trigger point" No optimal direction was chosen. At this time, when other agents are near the "trigger point", the agent will directly communicate with other agents and inform other agents of the optimal direction, that is, the opposite direction of the wrong direction selected by the agent at the beginning, and provide other agents with The forward movement of the agent provides a feasible reference and saves the path selection time of other agents. When there is a "trigger point" within the detection range of other agents, but the agent leaving the "trigger point" is not sure of the best direction, the other agents randomly select the direction on their own with a certain probability, for example, when there are two directions , the probability of each of the two directions being selected is 0.5. When the agent reaches the "trigger point", it is found that the agent that left the "trigger point" has entered the third state and has currently left the third state, indicating that the agent has entered the final state obtained through its own experience at this time. In the optimal direction, other agents can directly follow the agent to move. In this embodiment of the present invention, Figures 14 to 17 illustrate the application of the obstacle avoidance control strategy in different scenarios.

In addition, in one embodiment, the method further includes:

During the process of the agent moving forward in the fourth state, if a trigger point is detected, the current operating state and the previous operating state of the agent marking the trigger point are determined;

If the agent marking the trigger point does not move forward according to the third state in the current running state and the previous running state, step S500 is executed;

That is to say, the agent detects the trigger point, but the agent that marks the trigger point does not have a third state in the current operating state and the previous operating state, then step S500 is executed; that is, the agent that marks the trigger point If the agent neither knows the field boundary nor detects the field boundary, the agent (the agent moving forward according to the fourth state) needs to calculate the concentration evaluation value, the first distance evaluation value and the second distance evaluation value by itself. Determine the cost evaluation value of each vacant raster, and use the vacant raster with the smallest cost evaluation value as the raster to be entered.

If the current running state of the agent marking the trigger point is the third state, the agent will move forward in the opposite direction to the current forward path;

That is to say, the agent that marks the trigger point knows the field boundary and determines that its forward direction is blocked. Then the agent that sets the trigger point chooses to move in the opposite direction to the agent that marks the trigger point, thereby avoiding the inability to move forward. line path.

If the previous operating state of the agent that marked the trigger point was the third state and is not currently moving forward in the third state, then the agent will follow the agent that marked the trigger point;

That is to say, if the agent that marked the trigger point has found the correct path forward, the agent will directly follow the path of the agent that marked the trigger point and move forward.

Determine whether the agent that marked the trigger point has also marked the end point. If the agent that marked the trigger point has also marked the end point, then the agent follows the agent that marked the trigger point and moves toward the end point.

That is to say, if the agent that marked the trigger point has found the correct way forward and also found the exit, then the agent will directly follow the path of the agent that marked the trigger point and move toward the end point.

In addition, referring to Figure 18, one embodiment of the present invention also provides an electronic device 10. The electronic device 10 includes: a memory 11, a processor 12, and a computer program stored on the memory 11 and executable on the processor 12. .

The processor 12 and the memory 11 may be connected through a bus or other means.

The non-transient software programs and instructions required to implement the gene regulation network-based multi-agent navigation control method in the above embodiment are stored in the memory 11. When executed by the processor 12, the gene regulation network-based method in the above embodiment is executed. Multi-agent navigation control method.

In addition, an embodiment of the present invention also provides a computer-readable storage medium that stores computer-executable instructions, and the computer-executable instructions are executed by a processor or controller, for example, by the above-mentioned Execution by a processor in the electronic device embodiment can cause the processor to execute the multi-agent navigation control method based on the gene regulatory network in the above embodiment.

Those of ordinary skill in the art can understand that all or some steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit . Such software may be distributed on computer-readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As is known to those of ordinary skill in the art, the term computer storage media includes volatile and nonvolatile media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. removable, removable and non-removable media. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disk (DVD) or other optical disk storage, magnetic cassettes, tapes, disk storage or other magnetic storage devices, or may Any other medium used to store the desired information and that can be accessed by a computer. Additionally, it is known to those of ordinary skill in the art that communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media .

The above is a detailed description of the preferred implementation of the present invention, but the present invention is not limited to the above-mentioned embodiments. Those skilled in the art can also make various equivalent modifications or substitutions without violating the spirit of the present invention. Equivalent modifications or substitutions are included within the scope of the claims of the present invention.

Claims (8)

A multi-agent navigation control method based on gene regulation network, which is characterized by including: Step S100, obtain the plane map where multiple agents are located; Step S200: Convert the flat map into a grid map, and determine the concentration information corresponding to each grid in the grid map; wherein the concentration information includes the concentration information field of the target position and the concentration information of the obstacle position. field; Step S300: Determine the current grid and target grid of the agent to be navigated, and determine the optimal path for the agent from the current grid to the target grid based on the concentration information corresponding to each grid in the grid map. ; Wherein, the current grid is the grid where the agent is currently located, and the target grid is the grid where the target position is located in the grid map; Step S400: While the agent is moving along the optimal path, vacant grids are screened out from the eight neighborhood grids adjacent to the current grid, and the concentration evaluation value of each vacant grid is determined. , the first distance evaluation value and the second distance evaluation value; wherein, the vacant grid is a grid passable by the agent, and the concentration evaluation value is the grid where the agent is currently located and the vacant grid. The difference between the concentration information of the grids, the first distance evaluation value is the distance difference between the agent's previous grid and the vacant grid, the second distance evaluation value is the vacant grid The distance difference between the grid and the target grid; Step S500: Determine the cost evaluation value of each vacant grid according to the concentration evaluation value, the first distance evaluation value and the second distance evaluation value, and use the vacant grid with the smallest cost evaluation value as the grid to be entered; Step S600: Determine the obstacle distance between the grid to be entered and the grid where the obstacle is located, determine the operating state of the intelligent agent based on the obstacle distance, and control the intelligent agent to move toward the target according to the operating state of the intelligent agent. Position forward. The multi-agent navigation control method based on gene regulation network according to claim 1, characterized in that: converting the flat map into a grid map, determining the concentration corresponding to each grid in the grid map information, including: Step S210: Determine the grid where the target position is located, the grid where the obstacle is located, and the movable range boundary of the agent in the grid map; Step S220: Import the grid map into the gene regulation network model to generate a target concentration map and an obstacle concentration map; wherein each grid in the target concentration map contains the concentration information field of the target location, and the target concentration Each grid in the map contains a concentration information field for the location of obstacles; Step S230: Couple the target concentration map and the obstacle concentration map according to corresponding grids to obtain concentration information corresponding to each grid in the grid map. The multi-agent navigation control method based on gene regulation network according to claim 2, characterized in that the calculation formula relied on to generate the target concentration map is:
Among them, p ₁ is the concentration information field generated by the target, b ₁ is the adjustable parameter that affects the concentration value of each grid in the target concentration map, and v ₁ is the concentration diffusion factor, which is used to adjust the mapping relationship between distance parameters and concentration parameters. , r ₁ is the relative distance between the center of each grid in the target concentration map and the location of the target; The calculation formula relied on to generate the obstacle concentration map is:
Among them, p ₂ is the concentration information field generated by the obstacle, b ₂ is the adjustable parameter that affects the concentration value of each grid in the obstacle concentration map, v ₂ is the concentration diffusion factor, and r ₂ is each grid in the target concentration map. The relative distance between the grid center and the location of the obstacle; The calculation formula for the concentration information corresponding to each grid in the grid map is:
Among them, g ₁ is the morphological gradient space presented by the target concentration map, g ₂ is the morphological gradient space presented by the obstacle concentration map, g ₃ is the morphology presented by coupling the concentration information field of the target position and the concentration information field of the obstacle position. Gradient space, θ ₁ is an adjustable parameter that affects the concentration value range of the target concentration map, k ₁ is an adjustable parameter that affects the concentration difference between adjacent grids of the target concentration map, θ ₂ is the concentration that affects the obstacle concentration map The adjustable parameter of the value range, k ₂ is the adjustable parameter that affects the concentration difference between adjacent grids in the obstacle concentration map, θ ₃ is the adjustable parameter that affects the concentration value range of the grid in the grid map, k ₃ It is an adjustable parameter that affects the concentration difference between adjacent grids in the raster map. The multi-agent navigation control method based on gene regulation network according to claim 1, characterized in that the calculation formula of the cost evaluation value is:
Among them, P _e is the cost evaluation value for estimating that the agent to be navigated can reach the grid at the next moment, C _e is the concentration evaluation value for estimating that the agent to be navigated can reach the grid at the next moment, D _p In order to evaluate the first distance evaluation value that the agent to be navigated can reach to the grid at the next moment, D _t is to evaluate the second distance evaluation value that the agent to be navigated can reach the grid at the next moment, a, b and c are both weight parameters, C _N is the concentration value of the grid that the agent to be navigated can reach at the next moment, C _M is the concentration value of the grid where the agent to be navigated is at the current moment, C _min is the preset minimum value of the optional grid concentration, C _max is the preset maximum value of the optional grid concentration, N _x is the abscissa coordinate of the grid that the agent to be navigated can reach at the next moment, N _y is the ordinate of the grid where the agent to be navigated can be evaluated at the next moment, P _x is the abscissa of the grid where the agent to be navigated was at the previous moment, and P _y is the intelligence to be navigated. The ordinate of the grid where the body is located at the moment, T _x is the abscissa of the target location, and T _y is the ordinate of the target location. The multi-agent navigation control method based on gene regulation network according to claim 1, characterized in that the determination of the obstacle distance from the grid to be entered to the grid where the obstacle position is located is based on the distance and a preset The relationship between safe distance determines the operating status of the agent, including: Step S610: Determine whether the obstacle distance is greater than the safety distance. If the obstacle distance is greater than the safety distance, execute step S620; otherwise, execute step S660; Step S620: Control the intelligent agent to move forward toward the target position in a first state; wherein the first state is to move forward while avoiding obstacles; Step S630: During the process of the intelligent agent moving forward in the first state, if the trigger points marked by other intelligent agents are detected, the operating state of the intelligent agent is changed from the first state to the fourth state, and the intelligent agent is controlled. The agent moves towards the target position according to the fourth state Set to move forward; wherein, the fourth state is to move forward along the shortcut; Step S640: During the process of the agent moving from the first state to the fourth state, if the end points marked by other agents are detected, the running state of the agent is changed from the fourth state to the fourth state. In one state, the intelligent agent is controlled to move toward the target position in the first state; if the obstacle is detected to be within a safe distance, the operating state of the intelligent agent is transferred from the fourth state to the second state. , controlling the intelligent agent to move forward toward the target position in the second state; wherein the second state is to move along the edge of the obstacle; Step S650: During the process of the intelligent agent moving from the fourth state to the second state, if it is detected that the obstacle distance is greater than the safe distance, the operating state of the intelligent agent is converted from the second state to the second state. In the fourth state, the intelligent agent is controlled to move toward the target position in the fourth state; Step S660: Change the running state of the intelligent agent from the first state to the second state, and control the intelligent agent to move toward the target position in the second state; Step S670: During the process of the intelligent agent changing from the first state to moving forward in the second state, if it is detected that the obstacle distance is greater than the safe distance, the operating state of the intelligent agent is changed from the second state to the second state. In the first state, the intelligent agent is controlled to move forward toward the target position in the first state; if the obstacle distance is detected to be less than the field boundary distance, the location where the obstacle distance is detected to be less than the field boundary distance is marked as a trigger. After clicking, the running state of the intelligent agent is changed from the second state to the third state, and the intelligent agent is controlled to move toward the target position in the third state; wherein the third state is to follow the original path from the trigger point return. The multi-agent navigation control method based on gene regulation network according to claim 5, characterized in that the method further includes: During the process of the agent moving forward in the fourth state, if a trigger point is detected, the current operating state and the previous operating state of the agent marking the trigger point are determined; If the agent marking the trigger point does not move forward according to the third state in the current running state and the previous running state, step S500 is executed; If the current running state of the agent marking the trigger point is the third state, the agent will move forward in the opposite direction to the current forward path; If the previous operating state of the agent that marked the trigger point was the third state and is not currently moving forward in the third state, then the agent will follow the agent that marked the trigger point; Determine whether the agent that marked the trigger point has also marked the end point. If the agent that marked the trigger point has also marked the end point, then the agent follows the agent that marked the trigger point and moves toward the end point. An electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that when the processor executes the computer program, any one of claims 1 to 6 is achieved. A multi-agent navigation control method based on a gene regulatory network. A computer-readable storage medium storing computer-executable instructions, characterized in that the computer-executable instructions are used to perform multi-agent navigation based on a gene regulation network as described in any one of claims 1 to 6 Control Method.