Multi-robot distributed surrounding layering method, device, equipment, medium and product

Abstract

The invention discloses a multi-robot distributed surrounding layering method, device and equipment, a medium and a product, and relates to the field of multi-robot distributed motion planning. The method comprises the following steps: acquiring information data; constructing a pursuit task sample; according to the pursuit task sample and the information data, task allocation is carried out on the pursuit, whether secondary allocation is carried out or not is judged, and an allocation result is obtained; determining a final course angle of the pursuit according to a distribution result by adopting a distributed motion planning algorithm based on a following optimal gap method; based on the final course angle, an ORCA algorithm is adopted to carry out speed regulation processing, and the regulation speed is obtained; updating the position information of the escaper and the pursue; and according to the position information, determining whether the device is in a hunting stage based on hunting success conditions. According to the invention, the efficiency of multiple pursuers for hunting escapers can be improved, and hunting time can be shortened.

Description

Technical Field

[0001] The present application relates to the field of multi-robot distributed motion planning, and in particular to a multi-robot distributed encirclement layering method, device, equipment, medium and product. Background Art

[0002] With the widespread application of multi-robot pursuit-evasion game (MPE) in typical mission scenarios such as regional surveillance, target tracking and wildlife monitoring. In this game, a group of robots includes multiple pursuers and one escaper. The pursuers need to adopt a collaborative strategy to capture the adversarial escaper, while the escaper will try to avoid the capture of multiple pursuers. As a typical application scenario of MPE, the pursuers need to collaboratively capture a escaper in a dense environment with multiple obstacles. The main challenge is the cooperation between the pursuers, which requires dynamic reallocation of the capture target points, real-time motion planning to complete collision and obstacle avoidance, rapid approach to limit the motion of the adversarial escaper, and safe capture of the escaper. However, in a complex environment with dense obstacles, multiple pursuers are prone to deadlock when planning the capture path, and the computational complexity of the existing global planner does not meet the real-time requirements, making it difficult to complete the efficient capture of multiple pursuers. At the same time, the existing control methods for capturing adversarial escapees have strict requirements on the initial position and speed of the pursuer. Once the position and speed constraints of the pursuer are weakened, it will be difficult to successfully capture the escapee. It is difficult for the existing methods to encircle randomly appearing escapees, and the pursuer cannot efficiently capture the escapee. Therefore, there is an urgent need for a distributed capture algorithm that takes into account the pursuer's lack of initial position constraints, while connecting the task allocation and motion planning layers, realizing real-time dynamic task allocation and planning the optimal capture path, and efficiently completing the capture task.

[0003] In the prior art, when a control method is used to realize the pursuit and escape game of multiple robots, when the escapee is faster and smarter, its capture success rate depends on the strict initial position and speed conditions of all pursuers, that is, the current position and speed of the pursuers when they start to hunt the escapee. However, in reality, the initial position distribution of the pursuers does not conform to the uniform dispersion around the escapee, and is relatively random. Therefore, there is a lack of research on the effect of the initial position and real-time speed of the pursuers on the hunting effect.

[0004] Moreover, in an environment with dense obstacles, it is easier for multiple pursuers to fail to surround the escapee, and the deadlock problem between pursuers or between pursuers and obstacles is still a difficult problem. How to safely and efficiently plan the movement trajectories of multiple pursuers to achieve rapid capture remains to be solved. Summary of the invention

[0005] The purpose of this application is to provide a multi-robot distributed encirclement layering method, device, equipment, medium and product, which can improve the efficiency of multiple pursuers encircling and capturing fugitives and shorten the encirclement time.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides a multi-robot distributed encirclement and layering method, the multi-robot distributed encirclement and layering method comprising:

[0008] Acquire information data; the information data includes: motion data of the pursuer, motion data of the escaper and motion data of the obstacle; the motion data includes: position and speed; the initial position in the motion data is randomly generated based on the motion model; the motion model is a mathematical model constructed based on the motion data; the speed of the obstacle is always zero;

[0009] Construct a pursuer task sample; the pursuer task sample is determined based on the number of pursuers, the speed thresholds of the pursuers and the escapers, the capture radius of the escapers, and the successful capture condition; the successful capture condition is: the distance between all pursuers and the designated capture target point is less than the set distance, and the capture target point will be evenly distributed in the capture area with the escaper as the center and the capture radius Rc; the speed thresholds of the pursuers are not all greater than the speed threshold of the escaper;

[0010] According to the pursuer task sample and the information data, tasks are assigned to the pursuer, and it is determined whether to perform secondary assignment to obtain an assignment result; wherein the determination condition for performing secondary assignment includes: the existence of a path intersection;

[0011] Using a distributed motion planning algorithm based on a follow-the-best-gap method, a final heading angle of the pursuer is determined according to the allocation result;

[0012] Based on the final heading angle, an ORCA algorithm is used to perform speed regulation processing to obtain an adjusted speed;

[0013] Updating the position information of the escaper and the pursuer; the position information of the pursuer is determined based on the final heading angle and the adjusted speed; the position information of the escaper is determined based on the escaper's breakout strategy;

[0014] According to the position information, determining whether the hunting is in the hunting stage based on the hunting success condition, and obtaining a determination result;

[0015] If the determination result is yes, then return to the step of "assigning tasks to the pursuers according to the pursuer task sample and the information data, and determining whether to perform secondary assignment to obtain an assignment result";

[0016] If the determination result is no, the roundup mission ends.

[0017] Optionally, the motion model specifically includes:

[0018]

[0019] P = [x, y] T ;

[0020] Where P is the position vector; x is the horizontal coordinate corresponding to the position vector; y is the vertical coordinate corresponding to the position vector; T is the transpose; is the first-order derivative of x; is the first derivative of y; v is the linear velocity; w is the angular velocity; θ is the direction angle; is the first-order derivative of θ.

[0021] Optionally, according to the pursuer task sample and the information data, tasks are assigned to the pursuer, and it is determined whether to perform secondary assignment to obtain an assignment result, which specifically includes:

[0022] According to the pursuer task sample and the information data, the pursuer is initially assigned tasks based on the overall marginal reward function to obtain an initial assignment result; the overall marginal reward function is determined based on the distance discount factor and the relative speed factor using a dynamic reallocation review consistency algorithm;

[0023] Determine whether there is an intersection in the initial allocation result; the intersection is a cross path between pursuers or between pursuers and a capture target point;

[0024] If yes, the assigned capture target point is pre-exchanged, and the overall marginal gain after the pre-exchange is compared with the overall marginal gain before the pre-exchange to obtain a comparison result;

[0025] According to the comparison result, determining whether to perform secondary distribution and obtaining the distribution result;

[0026] If the comparison result is that the overall marginal gain increases after the exchange, the exchange process is performed; otherwise, the corresponding allocated capture target point is not exchanged; the overall marginal gain is determined according to the overall marginal reward function.

[0027] Optionally, a distributed motion planning algorithm based on the best gap following method is used to determine the final heading angle of the pursuer according to the allocation result, specifically including:

[0028] Determining a clearance angle between adjacent obstacles within the pursuer's field of view based on a linear expansion coefficient, wherein the linear expansion coefficient is negatively correlated with a distance between the pursuer and the adjacent obstacle;

[0029] Determine adaptive safety factors;

[0030] Determining a target heading angle according to the clearance angle and the adaptive safety factor;

[0031] Determine whether the pursuer is close to the escapee and obtain a determination result; the determination result is based on the criterion that the distance between the pursuer and the escapee at the next moment is not greater than the distance between the pursuer and the escapee at the current moment under the target heading angle;

[0032] If the judgment result is no, then return to “determining the clearance angle between adjacent obstacles within the pursuer’s field of view based on the linear expansion coefficient”;

[0033] If the judgment result is yes, the final heading angle is determined according to the target heading angle; the calculation formula of the final heading angle is:

[0034]

[0035] Among them, φ final is the final heading angle; α is the adaptive safety factor; d min is the minimum distance between the pursuer and the adjacent obstacle; φ gap_c is the clearance angle; β is the target weight coefficient; φ goal is the target heading angle.

[0036] Optionally, the expression for adjusting the speed is:

[0037]

[0038] in, To adjust the speed; is the lateral velocity of the pursuer A; v is the linear velocity; is the set of speeds allowed for the pursuer A within time τ; d A|B (v) is from v to The distance of the symbol from any region edge.

[0039] Optionally, the escapee's escape strategy specifically includes:

[0040] If there is no pursuer in the escaper's sight, the escaper can move randomly; if there is a pursuer, the escaper will move away from the pursuer; at this time, the corresponding first escaper's escape strategy is:

[0041]

[0042] If there are multiple pursuers around the evader, the evader will move away from the geometric center of all pursuers; at this time, the corresponding second evader's breakout strategy is:

[0043]

[0044] When multiple pursuers surround the escaper, the escaper breaks out based on the angle bisector; at this time, the corresponding third escaper's breakout strategy is:

[0045]

[0046] Among them, φ goall is the target heading angle of the escapee corresponding to the first escapee’s breakout strategy; P te is the relative position of the escaper and the pursuer corresponding to the first escaper’s breakout strategy; P e for the location of the escapee; is the geometric position center of multiple pursuers; P te ′ is the relative position of the escaper and the pursuer corresponding to the second escaper’s breakout strategy; r e is the radius of the escaper; x pi is the horizontal coordinate of the position of the pursuer i; pi is the ordinate of the position of the pursuer i; dir is the direction vector from the escaper to the center of the pursuer group; N Pe is the number of escapees; i is the pursuer ID index; φ goal2 is the target heading angle of the escapee corresponding to the second escapee’s breakout strategy; φ goal3 is the target heading angle of the escapee corresponding to the third escapee’s breakout strategy; α i is the clearance angle; β i is the overlap angle.

[0047] In a second aspect, the present application provides a multi-robot distributed encirclement and layering device, the multi-robot distributed encirclement and layering device comprising:

[0048] An information data acquisition module is used to acquire information data; the information data includes: the movement data of the pursuer, the movement data of the escaper and the movement data of the obstacle; the movement data includes: position and speed; the initial position in the movement data is randomly generated based on the movement model; the movement model is a mathematical model constructed based on the movement data; the speed of the obstacle is always zero;

[0049] A sample construction module is used to construct a pursuer task sample; the pursuer task sample is determined based on the number of pursuers, the speed thresholds of the pursuers and the escapers, the capture radius of the escapers, and the successful capture condition; the successful capture condition is: the distance between all pursuers and the designated capture target point is less than the set distance, and the capture target point will be evenly distributed in the capture area with the escaper as the center and the capture radius Rc; the speed thresholds of the pursuers are not all greater than the speed threshold of the escaper;

[0050] The allocation module is used to allocate tasks to the pursuer according to the pursuer task sample and the information data, and determine whether to perform secondary allocation to obtain an allocation result; wherein the judgment condition for performing secondary allocation includes: the existence of a path intersection;

[0051] a final heading angle determination module, configured to determine the final heading angle of the pursuer according to the allocation result by adopting a distributed motion planning algorithm based on a follow-up optimal gap method;

[0052] An adjustment module, configured to perform speed adjustment processing using an ORCA algorithm based on the final heading angle to obtain an adjusted speed;

[0053] An updating module, used for updating the position information of the escaper and the pursuer; the position information of the pursuer is determined based on the final heading angle and the adjusted speed; the position information of the escaper is determined based on the escaper's breakout strategy;

[0054] A determination module, configured to determine whether the hunting is in the hunting phase according to the position information and the hunting success condition, and obtain a determination result;

[0055] A return module, used for returning to the "allocation module" when the determination result is yes;

[0056] The end module is used to end the roundup mission when the determination result is no.

[0057] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the multi-robot distributed encirclement layering method described above.

[0058] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multi-robot distributed encirclement and layering method described above.

[0059] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned multi-robot distributed encirclement and layering method.

[0060] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0061] The present application provides a multi-robot distributed encirclement layering method, device, equipment, medium and product. By assigning tasks to pursuers according to pursuer task samples and information data, and judging whether to perform secondary assignment, the assignment result is obtained to solve the conflict problem of path point intersection when assigning tasks to multiple pursuers. A distributed motion planning algorithm based on the best gap following method is adopted to determine the final heading angle of the pursuer according to the assignment result, and then the ORCA algorithm is used for speed regulation to obtain the adjusted speed; the position information of the escaper and the pursuer is updated, and then the determination of whether it is in the encirclement stage is made based on the successful encirclement conditions. The present application introduces a linear expansion coefficient and an adaptive safety factor to determine a safe and fast optimal encirclement gap and smooth the motion trajectory of the pursuer; FOGM is combined with ORCA to solve the problem of missing FOGM speed domain and the deadlock problem of ORCA in complex environments; real-time dynamic task assignment and distributed motion planning are combined to construct a layered method to quickly solve the problem of multiple pursuers encircling the escaper. As a result, the present application can improve the efficiency of multiple pursuers encircling the escaper and shorten the encirclement time. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 A flowchart of the multi-robot distributed encirclement layering method;

[0064] Figure 2 A schematic diagram of the overall process for implementing the method mentioned in this application;

[0065] Figure 3 A schematic diagram of a cross path between pursuers or between pursuers and a target point for capture;

[0066] Figure 4 This is a schematic diagram of the distribution results of the secondary distribution;

[0067] Figure 5 Schematic diagram of the positions of the pursuer, the expanded obstacle, and the escaper at the optimal gap;

[0068] Figure 6 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0070] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0071] This application integrates multi-pursueer task allocation and motion planning levels in real time to solve the cluster roundup problem. A dynamic reallocate review consensus algorithm (DRRC) is proposed at the task allocation layer. By fully sampling the task set, a speed discount factor is introduced to optimize the objective function, and the distance and speed factors are balanced to optimize the allocation of roundup points, and obtain the task allocation result with the maximum return for each pursuer. At the same time, in order to solve the problem that there may be overlapping task points between adjacent pursuers when allocating tasks, the roundup target points of the crossed pursuers are replaced under the premise that the overall return has not been reduced, and the allocation results of the previous moment are effectively used to optimize the allocation of task points at the next moment, thereby improving the real-time allocation performance. In the motion planning layer, FOGM_ORCA (Follow the Optimal Gap Method-Optimal Reciprocal Collision Avoidance) is proposed. The follow gap method is improved through geometric analysis, and the linear expansion coefficient and adaptive safety factor are introduced. Combined with the optimal reciprocal collision avoidance (ORCA), the feasible speed of local planning is provided, so that multiple pursuers have the optimal angle domain and speed domain at the same time, effectively shortening the roundup time and improving the roundup efficiency. It can better connect the task allocation and motion planning levels, reduce the requirements for the initial position of the pursuer in the early stage of the roundup, balance the impact of the initial position on the roundup performance, and complete the efficient and real-time roundup of a fugitive by multiple pursuers.

[0072] Through geometric analysis, the linear expansion coefficient and adaptive safety factor are introduced to control the obstacle avoidance curvature of the pursuer to always remain in the safe gap area, so as to ensure that the planned path is the safest and take into account the influence of the target heading. At the same time, the ORCA algorithm is combined to achieve collision avoidance between pursuers in the team, improve the problem of missing speed domain in FOGM, and use the optimal heading obtained by FOGM to optimize the high-dimensional solution problem in ORCA. The maximum speed can still be used to avoid static obstacles on the opposite side, improving planning efficiency. By combining the dynamic task redistribution level and the distributed motion planning level, the problem of initial position constraints and speed constraints of the pursuer is solved, the efficiency of multiple pursuers in encircling and capturing escapees is improved, and the encirclement time is shortened.

[0073] like Figure 1 As shown, an embodiment of the present application provides a multi-robot distributed encirclement and layering method, the method comprising:

[0074] Step 100: Obtain information data. The information data includes: the motion data of the pursuer, the motion data of the escaper and the motion data of the obstacle; the motion data includes: position and speed; the initial position in the motion data is randomly generated based on the motion model; the motion model is a mathematical model constructed based on the motion data; the speed of the obstacle is always zero.

[0075] Step 200: Construct a pursuer task sample. The pursuer task sample is determined based on the number of pursuers, the speed thresholds of the pursuers and the escapers, the capture radius of the escapers, and the conditions for successful capture; the conditions for successful capture are: the distance between all pursuers and the designated capture target point is less than the set distance, the capture target point will be evenly distributed in the capture area with the escaper as the center and the capture radius Rc; the speed threshold of the pursuer is not all greater than the speed threshold of the escaper.

[0076] Step 300: According to the pursuer's task sample and information data, the pursuer is assigned tasks, and it is determined whether to perform secondary assignment to obtain an assignment result. The determination condition for performing secondary assignment includes: the existence of a path intersection.

[0077] Step 400: Using a distributed motion planning algorithm based on the best gap following method, the final heading angle of the pursuer is determined according to the allocation result.

[0078] Step 500: Based on the final heading angle, an ORCA algorithm is used to perform speed regulation processing to obtain an adjusted speed.

[0079] Step 600: Update the position information of the escaper and the pursuer. The position information of the pursuer is determined based on the final heading angle and the adjusted speed; the position information of the escaper is determined based on the escaper's breakout strategy.

[0080] Step 700: According to the position information, based on the capture success condition, determine whether it is in the capture stage and obtain a determination result.

[0081] Step 800: If the determination result is yes, return to step 300.

[0082] Step 900: If the determination result is no, the roundup mission ends.

[0083] Motion model, including:

[0084]

[0085] P = [x, y] T ;

[0086] Where P is the position vector; x is the horizontal coordinate corresponding to the position vector; y is the vertical coordinate corresponding to the position vector; T is the transpose; is the first-order derivative of x; is the first derivative of y; v is the linear velocity; w is the angular velocity; θ is the direction angle; is the first-order derivative of θ.

[0087] In one embodiment, tasks are assigned to the pursuers according to the pursuer task samples and information data, and it is determined whether to perform secondary assignment to obtain an assignment result, which specifically includes:

[0088] According to the pursuer's task samples and information data, the pursuer is initially assigned tasks based on the overall marginal reward function to obtain the initial assignment result; the overall marginal reward function is determined based on the distance discount factor and the relative speed factor using a dynamic reallocation review consistency algorithm.

[0089] Determine whether there is an intersection in the initial allocation result; an intersection is when there is an intersecting path between pursuers or between pursuers and the target point of the capture.

[0090] If so, the assigned capture target points are pre-exchanged, and the overall marginal gain after the pre-exchange is compared with the overall marginal gain before the pre-exchange to obtain a comparison result.

[0091] Based on the comparison result, determine whether to perform secondary distribution and obtain the distribution result.

[0092] If the comparison result shows that the overall marginal gain increases after the exchange, the exchange is performed; otherwise, the corresponding assigned capture target point is not exchanged; the overall marginal gain is determined according to the overall marginal reward function.

[0093] In one embodiment, a distributed motion planning algorithm based on the best gap following method is used to determine the final heading angle of the pursuer according to the allocation result, specifically including:

[0094] The clearance angle between adjacent obstacles within the pursuer's field of view is determined based on the linear expansion coefficient; the linear expansion coefficient is negatively correlated with the distance between the pursuer and the adjacent obstacle.

[0095] Determine the adaptive safety factor; determine the target heading angle according to the clearance angle and the adaptive safety factor. Determine whether the pursuer is close to the escapee and obtain a judgment result; the basis for the judgment result to make a close judgment is that at the target heading angle, the distance between the pursuer and the escapee at the next moment is not greater than the distance between the pursuer and the escapee at the current moment.

[0096] If the judgment result is no, return to the step of "determining the clearance angle between adjacent obstacles within the pursuer's field of view based on the linear expansion coefficient".

[0097] If the judgment result is yes, the final heading angle is determined according to the target heading angle; the calculation formula of the final heading angle is:

[0098]

[0099] Among them, φ final is the final heading angle; α is the adaptive safety factor; d min is the minimum distance between the pursuer and the adjacent obstacle; φ gap_c is the clearance angle; β is the target weight coefficient; φ goal is the target heading angle.

[0100] The expression for adjusting the speed is:

[0101]

[0102] in, To adjust the speed; is the lateral velocity of the pursuer A; v is the linear velocity; is the set of speeds allowed for the pursuer A within time τ; d A|B (v) is from v to The distance of the symbol from any region edge.

[0103] As an optional implementation method, the escapee's escape strategy specifically includes:

[0104] If there is no pursuer in the escaper's sight, the escaper can move randomly; if there is a pursuer, the escaper will move away from the pursuer; at this time, the corresponding first escaper's escape strategy is:

[0105]

[0106] If there are multiple pursuers around the evader, the evader will move away from the geometric center of all pursuers; at this time, the corresponding second evader's breakout strategy is:

[0107]

[0108] When multiple pursuers surround the escaper, the escaper breaks out based on the angle bisector; at this time, the corresponding third escaper's breakout strategy is:

[0109]

[0110] Among them, φ goal1 is the target heading angle of the escapee corresponding to the first escapee’s breakout strategy; P te is the relative position of the escaper and the pursuer corresponding to the first escaper’s breakout strategy; P e for the location of the escapee; is the geometric position center of multiple pursuers; P te ′ is the relative position of the escaper and the pursuer corresponding to the second escaper’s breakout strategy; r e is the radius of the escaper; x pi is the horizontal coordinate of the position of the pursuer i; pi is the ordinate of the position of the pursuer i; dir is the direction vector from the escaper to the center of the pursuer group; N Pe is the number of escapees; i is the pursuer ID index; φ goal2 is the target heading angle of the escapee corresponding to the second escapee’s breakout strategy; φ goal3 is the target heading angle of the escapee corresponding to the third escapee’s breakout strategy; α i is the clearance angle; β i is the overlap angle.

[0111] Based on the same inventive concept, the embodiment of the present application also provides a multi-robot distributed encirclement and stratification device for implementing the multi-robot distributed encirclement and stratification method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations of one or more multi-robot distributed encirclement and stratification device embodiments provided below can refer to the limitations of the multi-robot distributed encirclement and stratification method above, and will not be repeated here.

[0112] In one embodiment, a multi-robot distributed encirclement layering device is provided, the device comprising:

[0113] The information data acquisition module is used to acquire information data; the information data includes: the motion data of the pursuer, the motion data of the escaper and the motion data of the obstacle; the motion data includes: position and speed; the initial position in the motion data is randomly generated based on the motion model; the motion model is a mathematical model constructed based on the motion data; the speed of the obstacle is always zero.

[0114] A sample construction module is used to construct a pursuer task sample; the pursuer task sample is determined based on the number of pursuers, the speed thresholds of the pursuers and the escapee, the capture radius of the escapee, and the conditions for successful capture; the conditions for successful capture are: the distance between all pursuers and the designated capture target point is less than the set distance, and the capture target point will be evenly distributed in the capture area with the escapee as the center and the capture radius Rc; the speed thresholds of the pursuers are not all greater than the speed threshold of the escapee.

[0115] The allocation module is used to allocate tasks to the pursuer according to the pursuer's task samples and information data, and to determine whether to perform secondary allocation to obtain the allocation result; wherein the judgment condition for performing secondary allocation includes: the existence of a path intersection.

[0116] The final heading angle determination module is used to determine the final heading angle of the pursuer according to the allocation result by adopting a distributed motion planning algorithm based on the best gap following method.

[0117] The adjustment module is used to perform speed adjustment processing based on the final heading angle using the ORCA algorithm to obtain an adjustment speed.

[0118] The updating module is used to update the position information of the escaper and the pursuer; the position information of the pursuer is determined based on the final heading angle and the adjusted speed; the position information of the escaper is determined based on the escaper's breakout strategy.

[0119] The determination module is used to determine whether the capture phase is in progress according to the position information and the capture success condition, and obtain a determination result.

[0120] The return module is used to return to the "allocation module" when the judgment result is yes.

[0121] The end module is used to end the roundup mission when the determination result is no.

[0122] The existing collaborative capture methods do not consider the random initial position setting of the pursuer and the unequal uniform speed of the pursuer and the escapee, and also do not consider the real-time distribution of the pursuer's tasks. By analyzing the real-time status of the escapee, the pursuer is assigned distributed tasks and real-time motion planning, and a safe and capture-oriented direction is selected for movement, so as to design a more scalable pursuer capture strategy that is suitable for a variety of complex environments, such as environments with dense obstacles or escapees in narrow passages. This allows the pursuer to have strong adaptability, thereby improving the success rate of collaborative capture by multiple pursuers and shortening the capture time. The overall process of implementation is as follows: Figure 2 shown.

[0123] The escaper's speed is 0.6m / s, and the pursuer's speed can be set in the range of 0.6m / s-1m / s. The escaper and the pursuer have the same field of view radius R. c = 0.25m, the distance for successful capture is d c = 0.1m, set the safe distance between the pursuer and the obstacle to d 0 =2m, the entire simulation area is set to 50m×50m, the field of view of the pursuer and escaper S=10m, and the initial positions of the pursuer and escaper are randomly generated. The following implementation methods are all completed in MATLAB, with the purpose of realizing the coordinated capture of the escaper by multiple pursuers without initial position constraints.

[0124] like Figure 2 As shown, the specific implementation steps are as follows:

[0125] Step 1: Randomly generate the initial positions of the pursuer, escaper, and obstacle, and determine the initial speeds of the pursuer and escaper.

[0126] Considering that the pursuer needs to have flexible steering and reversing, the motion model of the pursuer and the escaper is simplified. The pursuer and the escaper do not have the constraint of turning radius. The position and velocity vector of the pursuer are: P = [x, y] T and The direction angle between the pursuer and the escaper is θ, the linear velocity and angular velocity of the pursuer are v and w respectively, and the motion model of the pursuer and the escaper is constructed:

[0127]

[0128] Among them, by determining the sampling time t of the pursuer and the escaper s , calculate the position and direction of the pursuer at the next moment of time k, that is, at time k+1:

[0129]

[0130] Determine the number of escapers and pursuers, determine the encirclement formation for the escapers based on the number of pursuers, and specify the maximum speed of the pursuers and escapers. The escapers and pursuers will evade and pursue at the maximum speed. The speed of the pursuers may not necessarily be faster than that of the escapers.

[0131] The pursuer and the escaper have the same field of view, which makes it easy to determine adjacent obstacles and other pursuers, and set the capture radius of the escaper. The condition for successful capture is that the distance between all pursuers and the designated capture target point is less than the set distance d min , and the capture target points will be evenly distributed in a circle with the escapee as the center and a capture radius of R c in the capture area.

[0132] Step 2: Construct the pursuer's task sample, perform the first task assignment to the pursuer, and check whether there are intersections and secondary reallocations. Combine historical data to ensure that the current optimal allocation result always maintains optimal consistency.

[0133] Step 2.1: Perform the first task assignment for all pursuers.

[0134] P i = {P 1 , P 2 , P 3 , …P n} represents the set of pursuers, and the pursuers need to quickly reach their assigned virtual target points, namely the capture points, which are evenly distributed around the escaper as the center. The capture points can be expressed as: T i ={T 1 , T 2 , T 3 ,…,T j ,…,T n Each pursuer can only be assigned one roundup point within the sampling time, and all roundup points have the same priority and importance factor q j =1, T represents the set of task points. In order to minimize the driving distance between the pursuer and the escaper, the influence of their relative speed on the distribution of the pursuer in the dynamic environment is also considered. Therefore, the distance discount factor and the relative speed factor are introduced, which are denoted as λ = 0.95 and χ = 0.95 respectively. The total reward value f of the pursuer i to reach the assigned target i (T i ) can be expressed as:

[0135]

[0136] in, is the path between pursuer i and capture point j, represents the relative speed between the pursuer i and the escaper. In order to obtain the allocation result of the pursuer i, the unallocated target m is added to the set T i Therefore, adding the marginal reward ω of target m i (m) is as shown in formula (4).

[0137]

[0138] Therefore, the marginal reward ω i (m) can be expressed as shown in formula (5).

[0139]

[0140] Due to the marginal reward ω i (m), and the objective function is a submodular function, the marginal reward is always decreasing. Therefore, the overall marginal reward function f(T) of dynamic task redistribution can be expressed as:

[0141]

[0142] Once pursuer i determines the allocation result, the optimal allocation result will be removed from the current task list until all tasks are fully allocated.

[0143] N a represents the pursuer's task sample set, T i represents the task currently assigned to pursuer i. Each pursuer can dynamically update T in each iteration i , no communication is required. At each sampling time t s , j * is the maximum marginal reward ω that can be provided to pursuer i * The task is, where i is the set of IDs of all pursuers. The Dynamic Redistribution Review Consensus (DRRC) algorithm consists of two different phases: the auction phase and the consensus phase.

[0144] In the auction phase, in order to obtain the global optimal solution as much as possible, the sample sampling frequency of each pursuer is set to P = 1. This means that each pursuer will consider all tasks to improve their respective task sample set N a , calculate the maximum marginal gain of each pursuer until all pursuers have completed the auction process. At this stage, pursuers share their ID, best task j * , marginal gain ω * , current position and speed with neighboring pursuers. At the same time, the pursuer can obtain useful information from its neighbors. Once both parties confirm that all shared information has been received, the pursuer will proceed to the next consensus stage.

[0145] In the consensus phase, once all pursuers have received all the shared information, they negotiate to determine which pursuer will get the current global best task. Select the corresponding maximum marginal gain ω * If a*=i, the ID of the assigned pursuer i and the corresponding task will be deleted. If there are multiple tasks with the same maximum marginal gain, the remaining tasks will be reallocated until the allocation is completed.

[0146] In the redistribution review consistency strategy, a is obtained through the MaxCons function * and Best Mission Directly assigned to pursuer a * , and other pursuers with different best tasks are not considered in this allocation. Otherwise, pursuers with the same best task The pursuers will compare their marginal gains, and the pursuer with the highest marginal gain will be assigned the task.

[0147] Step 2.2: After all the pursuers’ tasks are assigned, check whether there are any intersections between the assignment results of all adjacent pursuers, that is, there are intersection paths between pursuers or between pursuers and the target point, such as Figure 3 As shown, if there is a task allocation conflict, the allocated capture target points are exchanged, and the overall marginal gain after the exchange is compared with the overall marginal gain before the exchange. If the overall marginal gain increases after the exchange, the task exchange is considered to be optimal, otherwise, the corresponding allocated capture target points are not exchanged.

[0148] Step 2.3: Store the current optimal allocation result of the pursuer. The next allocation will be combined with the historical optimal allocation result value. That is, after repeating steps 2.1 and 2.2 to complete the task allocation, consider whether the current marginal gain is the smallest deviation from the value after the last allocation result. At the same time, the total marginal reward sum is still required to be the highest to ensure the consistency of each optimal allocation result. Figure 4 shown.

[0149] The optimal allocation result of the current pursuer is stored. The next allocation will be continued in combination with the optimal result value of the historical allocation. That is, consider whether the current marginal gain is reduced after combining the value of the last allocation result. At the same time, the total marginal reward sum is still required to be the highest to ensure the consistency of the optimal allocation result each time.

[0150] Step 3: Select the appropriate clearance angle for encirclement based on FOGM, always move at the maximum speed after determining the safe area, and integrate ORCA to provide the speed domain.

[0151] Considering the collision and obstacle avoidance problems of the pursuer with static and dynamic obstacles, a distributed motion planning algorithm based on the Follow the Gap Method (FOGM) is proposed. The algorithm combines the adaptive safety factor and the linear expansion coefficient to improve the original Follow the Gap Method (FGM), which belongs to geometric obstacle avoidance. A series of gap matrices are constructed by obtaining adjacent obstacles, and the optimal heading of the pursuer is calculated by identifying the maximum gap. The original FGM and the gap angle φ gap_c 、Target angle φ goal , safety factor α, target weight coefficient β, and the minimum distance d between the pursuer and the adjacent obstacle min The formula for FGM is as follows:

[0152]

[0153] Step 3.1: Since FGM needs to calculate the clearance angle between each pursuer's adjacent obstacles, this angle is related to the obstacle radius. Considering the safety and reliability of the pursuer's obstacle avoidance, the obstacles need to be expanded, so a linear expansion coefficient is introduced. The linear expansion coefficient is negatively correlated with the distance between the pursuer and the adjacent obstacle, which is convenient for meeting the safety margin during actual movement. The linear expansion coefficient is as follows:

[0154] r'=r_min+(r_max-r_min) * (d-D_min) / (D_max-D_min) (8)

[0155] Among them, D_min=1.2, D_max=2.4.

[0156] Therefore, the clearance angle will change as follows:

[0157]

[0158] Where P oi and P oi+1 represents the coordinates of the tangent point between the center position of the pursuer at the optimal gap and the expanded obstacle, P e Indicates the current location of the escapee. Figure 5 Determine all obstacles within the pursuer's field of view, obtain the obstacle index, calculate the angles between the expanded obstacles and the pursuer in sequence, and sort the gap angles from large to small.

[0159] Step 3.2: When the pursuer is encircling the escapee, as the escapee moves quickly, the pursuer will change the current gap angle according to the changes in the adjacent obstacles. In order to adapt to the rapid changes in the environment, the pursuer's safety factor α needs to be adjusted to an adaptive safety factor α'. However, the setting of the safety factor in the existing method is a fixed value, which does not meet the performance requirements of real-time encirclement and increases the possibility of collision avoidance between the pursuer and the pursuer or between the pursuer and the obstacle. Therefore, the safety factor needs to be adjusted according to the real-time planned trajectory of the pursuer to ensure safety. Keep the pursuer in the central safe area of ​​the gap as much as possible. The adaptive variable coefficient safety factor α' is based on the current position P i 、The target position P of the pursuer gi and the center of the tangent point after adaptively expanding the obstacle (P Oi1 +P Oi2 ) / 2 Real-time calculation, such as Figure 5 As shown. i is the adaptive safety factor of pursuer i, and the adaptive safety factor α' can be calculated based on α′ i Sure.

[0160]

[0161] Step 3.3: Calculate the target heading angle φ goal , combined with formula (7) to determine the final heading angle φ of each pursuer at the next moment final .

[0162] Step 3.4: According to φ final Calculate the position of each pursuer at the next moment. If the position will be far away from the escaper, return to step 3.1 and take the clearance angle at the second largest gap as the optimal clearance angle. Repeat this process until the optimal clearance angle that is safe and suitable for capture is determined, and then the final heading angle is determined.

[0163] Step 4: Combine the ORCA algorithm for real-time speed control.

[0164] The improved FOGM algorithm only drives the pursuer at a constant speed at the maximum speed, and lacks the adjustment of the real-time speed domain. When there are many obstacles or other pursuers in the pursuer's field of view, collision or deadlock is very likely to occur, resulting in the failure of this round-up mission. In order to improve the real-time performance of the algorithm, FOGM is combined with ORCA, and ORCA is used to provide the real-time speed domain that FOGM lacks. The necessity of introducing ORCA is clarified by planning the real-time speed of two pursuers A and pursuer B or one escaper C. The corresponding radii of pursuer A and pursuer B are r respectively. A and r B , when pursuer A and pursuer B appear in their respective danger avoidance ranges, a speed obstacle domain will appear, and its formula is:

[0165]

[0166] Among them, P A is the position of pursuer A, P B is the position of pursuer B, when , pursuer A and pursuer B will collide within time τ. Therefore, it is necessary to obtain the allowed speed set of pursuer A: v AB is the relative speed between pursuer A and pursuer B.

[0167] and It can be expressed as:

[0168]

[0169] Therefore, combining formula (13) and formula (14), the optimal new speed of the pursuer is

[0170]

[0171] in, represents the deviation speed of the pursuer A, d A|B (v) means from v to The distance of the symbol from any region edge.

[0172] Step 5: Design a strategy for the escapee to escape.

[0173] The complexity of the environment requires the evader to adopt various breakout strategies and choose the best one according to the current situation.

[0174] (1) When there is no pursuer within the escaper's line of sight, the escaper can move randomly. If a pursuer appears, the escaper will move away from the pursuer.

[0175]

[0176] (2) When there are multiple pursuers around the evader, the evader will move away from the geometric center of all pursuers.

[0177]

[0178] (3) When multiple pursuers surround a fugitive, the fugitive will choose to break out from the bisector of the angle with the largest gap between the pursuers.

[0179]

[0180] The escapee adopts multiple strategies to move in a dense environment, effectively verifying the scalability of the method mentioned in this application.

[0181] Step 6: Update the motion status of the pursuer and the escaper, and determine the capture stage based on the positional relationship between the pursuer and the escaper and the capture success conditions. If the capture is still in progress, the pursuer needs to complete dynamic task redistribution and distributed motion planning in real time; if the capture fails or succeeds, the capture mission is over.

[0182] The present application obtains the optimal solution for task allocation in real time by solving the conflict problem of path point intersection when allocating tasks to multiple pursuers; determines the current optimal allocation relationship by combining historical optimal allocation data to ensure the consistency of the optimal allocation solution at any time; improves the original following gap method, introduces a linear expansion coefficient and an adaptive safety factor to determine a safe and fast optimal encirclement gap, and smoothes the motion trajectory of the pursuer; combines FOGM with ORCA to solve the problem of FOGM's missing speed domain, and at the same time solves the deadlock problem of ORCA in complex environments; combines real-time dynamic task allocation with distributed motion planning to construct a hierarchical method to quickly solve the problem of multiple pursuers surrounding the escapee; and solves the problem of initial constraints of pursuers in dense environments based on the proposed hierarchical method.

[0183] The present application discloses a new hierarchical framework for multiple robots to surround a fugitive, which effectively solves the encirclement problem of multiple pursuers in dense environments without any pre-assigned initial position constraints. In order to obtain an approximate global optimal solution in dynamic task redistribution, a dynamic task allocation algorithm based on dynamic redistribution review consensus DRRC is proposed to resolve conflicts in all tasks. Each pursuer can dynamically calculate the total marginal reward through the relationship between the allocation schemes at the previous moment and the current moment. The allocation process is highly correlated with the previous moment to ensure the consistency of the best results of each allocation, greatly improving the allocation efficiency. A distributed motion planning algorithm based on the optimal gap following method FOGM integrates an adaptive safety factor and optimal reciprocal collision avoidance ORCA. Through the geometric analysis of pursuers and escapers in dense environments, linear expansion coefficients and adaptive safety factors are introduced to control the curvature of the pursuer's obstacle avoidance, smooth the motion trajectory of the pursuer, and solve the speed domain missing problem in FOGM. The encirclement trajectory of multiple pursuers is planned in real time, and finally an efficient encirclement of an escaper is achieved.

[0184] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 6As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a multi-robot distributed encirclement and layering method is implemented.

[0185] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0186] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0187] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0188] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0189] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0190] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0191] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0192] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0193] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A multi-robot distributed encirclement layering method, characterized in that: The multi-robot distributed encirclement layering method comprises: Acquire information data; the information data includes: motion data of the pursuer, motion data of the escaper and motion data of the obstacle; the motion data includes: position and speed; the initial position in the motion data is randomly generated based on the motion model; the motion model is a mathematical model constructed based on the motion data; the speed of the obstacle is always zero; Construct a pursuer task sample; the pursuer task sample is determined based on the number of pursuers, the speed thresholds of the pursuers and the escapers, the capture radius of the escapers, and the successful capture condition; the successful capture condition is: the distance between all pursuers and the designated capture target point is less than the set distance, and the capture target point will be evenly distributed in the capture area with the escaper as the center and the capture radius Rc; the speed thresholds of the pursuers are not all greater than the speed threshold of the escaper; According to the pursuer task sample and the information data, tasks are assigned to the pursuer, and it is determined whether to perform secondary assignment to obtain an assignment result; wherein the determination condition for performing secondary assignment includes: the existence of a path intersection; Using a distributed motion planning algorithm based on a follow-the-best-gap method, a final heading angle of the pursuer is determined according to the allocation result; Based on the final heading angle, an ORCA algorithm is used to perform speed regulation processing to obtain an adjusted speed; Updating the position information of the escaper and the pursuer; the position information of the pursuer is determined based on the final heading angle and the adjusted speed; the position information of the escaper is determined based on the escaper's breakout strategy; According to the position information, determining whether the hunting is in the hunting stage based on the hunting success condition, and obtaining a determination result; If the determination result is yes, then return to "allocate tasks to the pursuer according to the pursuer task sample and the information data, and determine whether to perform secondary allocation to obtain the allocation result"; If the determination result is no, the roundup mission ends.

2. The multi-robot distributed encirclement layering method according to claim 1, characterized in that: The motion model specifically includes: P=[x,y] T ; Where P is the position vector; x is the horizontal coordinate corresponding to the position vector; y is the vertical coordinate corresponding to the position vector; T is the transpose; is the first-order derivative of x; is the first derivative of y; v is the linear velocity; w is the angular velocity; θ is the direction angle; is the first-order derivative of θ.

3. The multi-robot distributed encirclement layering method according to claim 1, characterized in that: According to the pursuer task sample and the information data, tasks are assigned to the pursuer, and it is determined whether to perform secondary assignment to obtain an assignment result, which specifically includes: According to the pursuer task sample and the information data, the pursuer is initially assigned tasks based on the overall marginal reward function to obtain an initial assignment result; the overall marginal reward function is determined based on the distance discount factor and the relative speed factor using a dynamic reallocation review consistency algorithm; Determine whether there is an intersection in the initial allocation result; the intersection is a cross path between pursuers or between pursuers and a capture target point; If yes, the assigned capture target point is pre-exchanged, and the overall marginal gain after the pre-exchange is compared with the overall marginal gain before the pre-exchange to obtain a comparison result; According to the comparison result, determining whether to perform secondary distribution and obtaining the distribution result; If the comparison result is that the overall marginal gain increases after the exchange, the exchange process is performed; otherwise, the corresponding allocated capture target point is not exchanged; the overall marginal gain is determined according to the overall marginal reward function.

4. The multi-robot distributed encirclement layering method according to claim 1, characterized in that: A distributed motion planning algorithm based on the best gap following method is used to determine the final heading angle of the pursuer according to the allocation result, specifically including: Determining a clearance angle between adjacent obstacles within the pursuer's field of view based on a linear expansion coefficient, wherein the linear expansion coefficient is negatively correlated with a distance between the pursuer and the adjacent obstacle; Determine adaptive safety factors; Determining a target heading angle according to the clearance angle and the adaptive safety factor; Determine whether the pursuer is close to the escapee and obtain a determination result; the determination result is based on the criterion that the distance between the pursuer and the escapee at the next moment is not greater than the distance between the pursuer and the escapee at the current moment under the target heading angle; If the judgment result is no, return to the step of "determining the clearance angle between adjacent obstacles within the pursuer's field of view based on the linear expansion coefficient"; If the judgment result is yes, the final heading angle is determined according to the target heading angle; the calculation formula of the final heading angle is: Among them, φ final is the final heading angle; α is the adaptive safety factor; d min is the minimum distance between the pursuer and the adjacent obstacle; φ gap_c is the clearance angle; β is the target weight coefficient; φ goal is the target heading angle.

5. The multi-robot distributed encirclement layering method according to claim 1, characterized in that: The expression of adjusting speed is: in, To adjust the speed; is the lateral velocity of the pursuer A; v is the linear velocity; is the set of speeds allowed for the pursuer A within time τ; d A|B (v) is from v to The distance of the symbol from any region edge.

6. The multi-robot distributed encirclement layering method according to claim 1, characterized in that: The escapee's escape strategy specifically includes: If there is no pursuer in the escaper's sight, the escaper can move randomly; if there is a pursuer, the escaper will move away from the pursuer; at this time, the corresponding first escaper's escape strategy is: If there are multiple pursuers around the evader, the evader will move away from the geometric center of all pursuers; at this time, the corresponding second evader's breakout strategy is: When multiple pursuers surround the escaper, the escaper breaks out based on the angle bisector; at this time, the corresponding third escaper's breakout strategy is: Among them, φ goall is the target heading angle of the escapee corresponding to the first escapee’s breakout strategy; P te is the relative position of the escaper and the pursuer corresponding to the first escaper's breakout strategy; P e for the location of the escapee; is the geometric position center of multiple pursuers; P te ′ is the relative position of the escaper and the pursuer corresponding to the second escaper’s breakout strategy; r e is the radius of the escaper; x pi is the horizontal coordinate of the position of the pursuer i; pi is the ordinate of the position of the pursuer i; dir is the direction vector from the escaper to the center of the pursuer group; is the number of escapees; i is the pursuer ID index; φ goal2 is the target heading angle of the escapee corresponding to the second escapee’s breakout strategy; φ goal3 is the target heading angle of the escapee corresponding to the third escapee’s breakout strategy; α i is the clearance angle; β i is the overlap angle.

7. A multi-robot distributed encirclement layering device, characterized in that: The multi-robot distributed encirclement and layering device comprises: An information data acquisition module is used to acquire information data; the information data includes: the movement data of the pursuer, the movement data of the escaper and the movement data of the obstacle; the movement data includes: position and speed; the initial position in the movement data is randomly generated based on the movement model; the movement model is a mathematical model constructed based on the movement data; the speed of the obstacle is always zero; A sample construction module is used to construct a pursuer task sample; the pursuer task sample is determined based on the number of pursuers, the speed thresholds of the pursuers and the escapers, the capture radius of the escapers, and the successful capture condition; the successful capture condition is: the distance between all pursuers and the designated capture target point is less than the set distance, and the capture target point will be evenly distributed in the capture area with the escaper as the center and the capture radius Rc; the speed thresholds of the pursuers are not all greater than the speed threshold of the escaper; The allocation module is used to allocate tasks to the pursuer according to the pursuer task sample and the information data, and determine whether to perform secondary allocation to obtain an allocation result; wherein the judgment condition for performing secondary allocation includes: the existence of a path intersection; a final heading angle determination module, configured to determine the final heading angle of the pursuer according to the allocation result by adopting a distributed motion planning algorithm based on a follow-up optimal gap method; An adjustment module, configured to perform speed adjustment processing using an ORCA algorithm based on the final heading angle to obtain an adjusted speed; An updating module, used for updating the position information of the escaper and the pursuer; the position information of the pursuer is determined based on the final heading angle and the adjusted speed; the position information of the escaper is determined based on the escaper's breakout strategy; A determination module, configured to determine whether the hunting is in the hunting phase according to the position information and the hunting success condition, and obtain a determination result; A return module, used for returning to the "allocation module" when the determination result is yes; The end module is used to end the roundup mission when the determination result is no.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the multi-robot distributed encirclement layering method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the multi-robot distributed encirclement layering method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the multi-robot distributed encirclement layering method according to any one of claims 1 to 6 is implemented.

Images