Multi-agent coverage search and task allocation path optimization methods, devices, and media

Abstract

This invention relates to a method, device, and medium for multi-agent coverage search and task allocation path optimization. The specific method includes: scenario construction and problem description, which describes the characteristics of the task area, the attributes of agents and targets, and clarifies the area coverage search problem and the target task allocation problem; area coverage model design, which establishes a minimum task time model objective function and constraints, and then solves the area coverage model; and task allocation strategy and path planning algorithm, which, considering agent speed and target attributes, combines a greedy strategy and a genetic algorithm to allocate target exploration tasks to agents, optimize target exploration paths, and reduce total task time. Compared with existing technologies, this invention improves the efficiency of marine area search and target exploration through multi-agent cooperation, achieving full area coverage search and accurate target exploration.

Description

Technical Field

[0001] This invention relates to the field of multi-agent path planning technology, and in particular to a method, device, and medium for multi-agent coverage search and task allocation path optimization. Background Technology

[0002] In recent years, with the rapid development of information and control technologies, intelligent agents have become a research hotspot. Multi-agent systems (MASs), as an important research direction in the field of intelligent agents, aim to solve complex tasks that are difficult for a single agent to handle. For example, researchers have proposed a method for searching maritime targets based on autonomous underwater robots to solve the problem of searching for moving targets and invisible objects. Intelligent algorithms such as ant colony optimization have also been widely applied to the search path design of unmanned aerial vehicles (UAVs) to ensure that moving targets are found in the shortest possible time. Multi-agent systems not only provide more solutions but also have higher efficiency, higher intelligence levels, and better fault tolerance. Multiple UAVs and multiple autonomous underwater robots have been widely used in fields such as marine engineering, maritime search and rescue, deep-sea resource surveys, scientific research, and naval defense.

[0003] However, several key challenges remain in the field of multi-agent cooperative search. One challenge is achieving full-coverage search path planning for a specific search area; efficiently covering the entire search region is a crucial issue in multi-agent system research. Another challenge is the allocation of multi-agent, multi-target search tasks; effectively allocating agents to complete multiple search tasks and achieving global point-to-point path planning. Multi-agent coverage search methods are primarily based on grid partitioning and are suitable for scenarios where agents have the same detection radius. However, this method is not applicable when agents have different detection radii. Research on target task allocation mainly focuses on scenarios where one or more agents perform a single visit to multiple targets, with less attention paid to multiple visits by multiple agents to multiple targets. However, in actual marine exploration, targets often require multiple visits to collect complete information, and sometimes effective monitoring within specific time intervals is necessary. Therefore, there is an urgent need for an efficient method for multi-agent coverage search and task allocation path optimization to improve search efficiency, achieve regional coverage search, and ensure accurate target detection, which has significant application value. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a multi-agent coverage search and task allocation path optimization method, device, and medium. Specifically, it includes scenario construction and problem description, regional coverage model design, task allocation strategy, and path optimization algorithm. The aim is to establish a complete, scientific, reasonable, and universal regional coverage model and task allocation path optimization method by studying the problem of multi-agents with different attributes performing target search and target exploration in unknown sea areas. This can provide a feasible method for solving the problems existing in real-world maritime target search.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] The first aspect of this invention provides a method for multi-agent coverage search and task allocation path optimization, comprising the following steps:

[0007] S1: Scenario Construction and Problem Description

[0008] Regarding the problem description:

[0009] The task of the multi-agent system is to perform a full-coverage search and exploration of multiple targets within a given area.

[0010] In terms of scene construction:

[0011] Set the task area.

[0012] Set the number and attributes of agents and targets;

[0013] S2: Regional Coverage Model Design

[0014] S21: Establish the model objective function, including calculating the time required for each agent to allocate different detection paths, taking the maximum detection time of all agents as the task time for full coverage search of the region, and taking minimizing the task time as the optimization objective.

[0015] S22: Establish model constraints, including agent region coverage constraints, agent detection time constraints, and agent scenario assumption constraints;

[0016] S23: Solve the region coverage model;

[0017] S3: Constructing task allocation strategies and path optimization algorithms

[0018] S31: Generate an initial optimal solution, which includes the target exploration path for each agent;

[0019] S32: Optimize the initial optimal solution, use a genetic algorithm for iterative optimization, and obtain the final multi-agent coverage search and task allocation path optimization strategy.

[0020] Furthermore, in S1, the task area is a convex polygonal region, and is further approximated as a rectangle;

[0021] The process of setting the number and attributes of agents and targets includes:

[0022] There are n agents outside the mission area and m unknown targets within the mission area. Each agent has different attributes, including initial position, maximum speed, detection radius, detectable angle, and maximum detection speed.

[0023] Each target has different attributes, including location, number of times to be explored, exploration radius, exploreable angle, and target revisit time interval.

[0024] Furthermore, in S1, regarding the problem description:

[0025] The full-coverage search of the region is achieved by minimizing the detection task time by reasonably allocating the detection task area of ​​the intelligent agent, thereby achieving full-coverage search of the sea area and obtaining information on all unknown targets within the task area;

[0026] The multi-target exploration is a task allocation problem involving multiple agents exploring multiple targets multiple times. Specifically, after the area coverage search, exploration tasks need to be performed on the detected targets. By reasonably allocating the target exploration tasks for each agent, the exploration task time can be minimized, thereby improving exploration efficiency and accuracy.

[0027] Furthermore, in S1, the constraint on the target revisit time interval is:

[0028] When agent A enters the detection range of target B and completes a valid detection, it needs to leave the detection range of target B. If the time spent leaving the detection range of target B exceeds the revisit interval of target B, agent A can re-enter the detection range of target B and is considered to have conducted a second valid detection of target B.

[0029] The constraint on the target revisit time interval applies only to repeated explorations of a single target by a single agent; explorations of the same target by different agents are not restricted.

[0030] Furthermore, in S21, the objective function for minimizing the task time model is established as follows:

[0031] (1)

[0032] In the formula, T k The time taken for the k-th agent to complete the planned path.

[0033] Furthermore, in S22, the specific constraint on the coverage of the agent's region is as follows:

[0034] The area coverage adopts the channel division method. Based on the initial position, maximum sailing speed, and detection range radius of the agent, the number of channels for each agent is reasonably allocated to ensure that all channels cover the task area, and the channel coverage area shall not exceed the width of one detection radius of the task area. The full coverage search constraint formula (2) for the agent's task area is as follows:

[0035] (2)

[0036] In the formula,

[0037] k, the agent number, k∈(1,2,3,…,n)

[0038] The number of waterways that the k-th agent plans to explore. =1,2,3….,

[0039] The detection range radius of the k-th agent, in meters.

[0040] R, the maximum detection radius of all agents, in meters.

[0041] The difference between the maximum and minimum ordinates of the mission area. =max Lat – min Lat, unit: m;

[0042] The specific constraints on the agent's detection time are as follows:

[0043] The navigation time of the intelligent agent's route planning scheme is calculated based on the detection radius and maximum navigation speed of each agent. The constraint is shown in the following formula (3):

[0044] (3)

[0045] In the formula,

[0046] d1, the distance from the agent's initial position to the assigned starting point of the course, in meters;

[0047] d2, the path distance assigned to the agent's course, in meters;

[0048] The maximum speed of the kth agent;

[0049] The specific constraints of the agent scenario assumptions are as follows:

[0050] Based on the mission area, the detection radius of each agent, and the maximum navigation speed, the course range of each agent and the range of the total number of courses are determined, and scenario assumption constraints are established, as shown in the following formulas (4) and (5):

[0051] (4)

[0052] (5)

[0053] Furthermore, in S31, the process of generating the initial optimal solution specifically includes the following steps:

[0054] S31-1: In the first round of selection, the order of the agents is randomly shuffled, and each agent chooses the target closest to itself;

[0055] S31-2: In the nth round of selection, sort the total travel time of each agent to the target in the previous n-1 rounds. The agent with the longest total travel time has priority. Each agent selects the target closest to the current position in turn according to the total travel time in the previous n-1 rounds, and determines whether the time to travel to the next target is less than the target revisit time interval. If the time to travel to the next target is less than the target revisit time interval, then travel to the next target. Otherwise, select the current target, and the number of times the selected target needs to be explored is reduced by 1.

[0056] S31-3: Repeat S31-1 and S31-2 until all targets have 0 exploration attempts, and obtain a series of initial solutions with the shortest target exploration path time for the agents.

[0057] Furthermore, in S32, optimizing the initial optimal solution specifically includes the following steps:

[0058] S32-1: Tournament selection is adopted. s individuals are randomly sampled from the population with replacement. The best individual is then selected to enter the next generation. Only when the fitness value of an individual is better than the other s-1 competitors can it win the tournament.

[0059] S32-2: Perform crossover operations on the individuals selected in S32-1 through the tournament. By exchanging and combining chromosomes, the superior characteristics of the parent string are inherited to the substring, thereby generating new superior individuals. The number of times each target needs to be explored is fixed. When performing the crossover operation, the target sequence number combination is inserted, and duplicate target sequence numbers are removed from the new solution after exchanging chromosomes.

[0060] S32-3: Perform mutation operations on the individuals selected in the tournament in S32-1. The algorithm achieves chromosome mutation by randomly swapping 1-2 target indices.

[0061] S32-4: Repeat S32-1 to S32-3, using a shorter task time to retain the solution in each iteration. By setting the population size, the global optimal solution for the target task allocation is finally obtained, serving as a multi-agent coverage search and task allocation path optimization strategy.

[0062] A second aspect of the present invention provides an electronic device, including a memory and a processor, wherein the processor is used to execute a program in the memory to realize the multi-agent coverage search and task allocation path optimization method described above.

[0063] A third aspect of the present invention provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, is used to perform the multi-agent coverage search and task allocation path optimization method described above.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] (1) Intelligent agents have different attributes, such as maximum sailing speed, detection radius, and maximum exploration speed, while targets also have different attributes, such as location, number of times to be explored, and range of detectable radius. Based on a comprehensive consideration of the diversity and differences between multiple intelligent agents and targets, this invention establishes a complete, scientific, reasonable, and versatile method for area coverage search and task allocation path optimization based on a time minimization model. This comprehensive consideration makes the method more applicable to various complex marine environments and provides a more flexible and adaptable solution.

[0066] (2) The task allocation strategy and path optimization method in this invention take into account the uncertainty of multiple visits to multiple targets, the determinacy of the number of visits, the number of agents and the exploration speed limit. Compared with the single number of visits of a single agent, the decision optimization of multiple visits of multiple targets by multiple agents is more complex. This algorithm avoids repeated exploration and improves the efficiency of target exploration by optimizing the mutual cooperation between agents, reducing search time and improving the efficiency of path planning.

[0067] (3) This invention introduces a target revisit time interval attribute, which requires the agent to fully consider the revisit requirements of different targets in task allocation and path planning to ensure timely and effective exploration of these targets. This consideration can improve the periodicity and accuracy of exploration work and help to better meet the needs of practical applications. Attached Figure Description

[0068] Figure 1 This is a schematic diagram of the correct exploration provided by the present invention;

[0069] Figure 2The multi-agent task area route division map provided by this invention;

[0070] Figure 3 This is a schematic diagram of the "probing line" provided by the present invention;

[0071] Figure 4 The optimized initial optimal solution diagram provided by this invention;

[0072] Figure 5 This invention provides a target task allocation graph for each agent.

[0073] Figure 6 A segment diagram of the intelligent agent operation simulation process provided by the present invention. Detailed Implementation

[0074] Overall, this invention proposes a multi-agent coverage search and task allocation path optimization method, specifically including: 1. Scene construction and problem description, which describes the characteristics of the task sea area, the attributes of agents and targets, and clarifies the area coverage search problem and the target task allocation problem; 2. Area coverage model design, which establishes the objective function and constraints for minimizing task time, and then solves the area coverage model; 3. Task allocation strategy and path planning algorithm, which, considering agent speed and target attributes, combines a greedy strategy and a genetic algorithm to allocate target exploration tasks to agents, optimize target exploration paths, and reduce total task time. This invention aims to improve the efficiency of marine area search and target exploration through multi-agent cooperation, achieving full area coverage search and accurate target exploration.

[0075] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms, and other features not explicitly described in this technical solution are considered common technical features disclosed in the prior art.

[0076] The technical solution adopted in this invention is a multi-agent coverage search and task allocation path optimization method, which specifically includes the following steps:

[0077] S1: Scenario Construction and Problem Description.

[0078] The complex and ever-changing marine environment significantly impacts the detection accuracy of traditional platform payloads. Generally, close-range target observation yields high target localization accuracy. Utilizing intelligent agents to assist in target detection in designated complex sea areas can effectively improve detection accuracy and reduce safety risks. In practical engineering, multi-agent dynamic planning is performed on designated irregular and enclosed sea areas, requiring the achievement of multi-agent area coverage search within the shortest possible mission time and completion of the detection task for already detected targets. This case study adheres to the WGS84 coordinate system.

[0079] Mission Sea Area Description

[0080] The mission area is an irregular convex polygon region, which is formed by connecting the 8 vertices in Table 1 in sequence. Its size is approximately a rectangle with a length of 7090 meters and a width of 12335 meters.

[0081] Table 1 Task Area

[0082]

[0083] Intelligent agent attribute description

[0084] There are eight agents in total, each with different attributes, including initial position, maximum speed, detection radius, detectable angle, and maximum exploration speed. These differences in attributes will affect the agents' search capabilities and strategy selection.

[0085] The agent-related variables are shown in Table 2 below. Here, `id` is the agent's ID, `lon` and `lat` are the agent's initial latitude and longitude coordinates, `detectRadius` represents the agent's detection radius, `detectAngle` represents the agent's detectable angle range, `sailLimit` represents the agent's maximum sailing speed, and `certainLimit` refers to the agent's maximum detection speed. For example, agent 1 has an initial latitude and longitude position of (121.65108, 38.825956), a detection radius of 250m, and a detectable angle range of [30, 150] (e.g., ...). Figure 1 The maximum sailing speed is 20 knots, and the maximum detection speed is limited to 10 knots. The detectable angle range is a symmetrical detection area with 0 degrees at the bow and 180 degrees at the stern.

[0086] Table 2 Agent Attribute Table

[0087]

[0088] Target attribute description

[0089] There are m unknown targets distributed within the mission area, which need to be found through a multi-agent region coverage search. Each target has different attributes, including location, number of times to be explored, exploration radius, threat radius, exploreable angle, and target revisit time interval. Each target needs to be explored correctly multiple times.

[0090] Target revisit interval: After agent A enters the detection range of target B and completes a valid detection, it needs to leave the detection range of target B. If the time spent leaving the detection range of target B exceeds the target B revisit interval, the agent will be terminated.

[0091] If agent A re-enters the exploration range of target B, it is considered to have conducted a second effective exploration of target B. This constraint only applies to repeated explorations of a single target by a single agent; explorations of the same target by different agents are not restricted.

[0092] Correct detection: A correct detection requires the following conditions to be met: Agent A enters the detection range of target B, target B falls within the detection range of agent A, and agent A travels at a speed less than the maximum detection speed. This is recorded as a correct detection. Figure 1 As shown.

[0093] The latitude and longitude coordinates of the task area and agent in the WGS84 coordinate system are converted to the projected coordinates in the EPSG:3857 projected coordinate system. At the same time, the agent velocity unit is converted from knots to m / s. The initialization information of the task area and agent is shown in Tables 3-4 below.

[0094] Table 3 Task Area Initialization Table (Unit: m)

[0095]

[0096] Table 4 Agent Attribute Initialization Table

[0097]

[0098] S2: Regional Coverage Model Design.

[0099] The main assumptions of the regional coverage model are as follows:

[0100] The agents were initially located outside the task area;

[0101] The agent model is assumed to be a point mass;

[0102] The detection radius of the intelligent agent is always greater than the detection radius of the target.

[0103] Targets are randomly assigned within the mission area, with any two adjacent targets being at least 200m apart and at least 150m away from the boundary of the mission area.

[0104] S21: Establish the objective function for the model;

[0105] Based on the maximum navigation speed and planned exploration path of each agent, the time required to allocate different exploration paths to each agent is calculated. The maximum exploration time of the agents in the path allocation scheme is taken as the total task time of the scheme. The task time minimization model is shown in the following formula (6):

[0106] (6)

[0107] In the formula, Tk The time taken for the k-th agent to complete the planned path.

[0108] By establishing a time minimization model and performing multiple iterative calculations, the path allocation scheme with the shortest total task duration is obtained.

[0109] S22: Establish model constraints, including agent region coverage constraints, agent scenario assumption constraints, and agent detection time constraints.

[0110] Agent region coverage constraints

[0111] Based on the initial position, maximum speed, and detection radius of the agent, the number of channels for each agent is reasonably allocated to ensure that all channels cover the mission area, and the channel coverage area shall not exceed the width of one detection radius of the mission area. The full coverage search constraint formula (7) for the agent's mission sea area is as follows:

[0112] (7)

[0113] In the formula,

[0114] k, the agent number, k∈(1,2,3,…,n)

[0115] The number of waterways that the k-th agent plans to explore. =1,2,3…;

[0116] The detection range radius of the k-th agent, in meters;

[0117] R represents the maximum detection radius of all agents, in meters (m).

[0118] The difference between the maximum and minimum ordinates of the mission area. =max Lat – min Lat, unit: m;

[0119] Agent detection time constraints

[0120] The sailing time of the intelligent agent's route planning scheme is calculated based on the route assigned to each intelligent agent and the maximum sailing speed. The constraint is shown in the following formula (8).

[0121] (8)

[0122] In the formula,

[0123] d1, the distance from the agent's initial position to the assigned starting point of the course, in meters (m).

[0124] d2, the path distance assigned to the agent's course, in meters;

[0125] , the maximum speed of the kth agent.

[0126] Intelligent agent scenario assumptions and constraints

[0127] Based on the mission area, the detection radius of each agent and the maximum sailing speed, the course range of each agent and the range of the total number of courses are determined, and scenario assumption constraints are established, as shown in the following formulas (9) and (10).

[0128] (9)

[0129] (10)

[0130] S23: Solve the region coverage model.

[0131] The algorithm for solving the region coverage model is as follows:

[0132] Step 1: Initialize parameters, including mission area, maximum speed and initial position of the agent.

[0133] Step 2: Determine the maximum difference in ordinates within the task area. Based on constraints (4) and (5), calculate the range of the number of channels ck for each agent, and the range of the total number of channels.

[0134] Step 3: Based on constraint (2), generate all possible course allocation combinations {c1, c2, ..., ck} for the multi-agent team.

[0135] Step 4: Iterate through each channel and assign combinations {c1, c2, ..., c} k}. Calculate the detection time for each agent according to formula (3) to determine the final task duration. In each iteration, retain the allocation scheme with the shorter task time. Set the number of iterations, repeat this step, and finally obtain a sufficient combination of course allocations.

[0136] This embodiment uses PyCharm Community Edition 2022.1.3 to implement the above algorithm and solve the region coverage problem. The runtime environment is Python 3.11. In this embodiment, the sea area to be detected is approximated as a rectangle. Based on the region coverage model calculation results in step S2, the sea area is divided into 22 approximate rectangles. The agent moves along the center line of the rectangle, and each agent detects 2 or 3 small rectangular regions. Figure 2The planned detection trajectory for the agents is defined as follows: agents 1, 4, 5, 6, 7, and 8 detect three approximately small rectangular regions, while agents 2 and 3 detect two small rectangular regions, achieving region coverage search (see...). Figure 2 The model calculated the initial estimated detection time for each agent during the coverage process, as shown in Table 5. The final coverage search task took 39.87 minutes, and a total of 35 targets were found. Furthermore, agents that complete the coverage search task ahead of schedule can explore the target closest to their current location instead of waiting in place. Step S3 is executed once all agents have completed their search tasks.

[0137] Table 5 Multi-agent Region Coverage Duration

[0138]

[0139] Figure 2 The relative positions of targets obtained during the coverage search are displayed. The target information is initialized and organized into the form shown in Table 5 below. The first 35 rows are target information, and the last 8 rows are agent information. id is the agent and target number, x_coord and y_coord are the projected coordinates under EPSG:3857, demand is the number of times to be explored, speed is the agent's maximum travel speed, rest_time is the agent's revisit time interval, name and seq_no are the new numbers for all agents and targets; angel is the range of detection or exploration angles.

[0140] Table 6 Target Requirements Table

[0141]

[0142] To meet the requirement of "correct exploration," this example designs "exploration lines" to adjust the agent's driving angle and speed. For example... Figure 3 As shown, the "exploration line" refers to the line connecting intersection point C and intersection point D. Intersection point C is the intersection of the angle bisector of the target's probing angle range and the target's exploration radius circle, while intersection point D is the intersection of the angle bisector rotated 15° clockwise and the target's exploration radius circle. During the target exploration process, (1) if the target point the agent wants to go to next is not the target point where the agent is currently located, then the agent travels to point C at the maximum speed according to the planned exploration route, and then travels from point C to point D at the exploration speed. This design satisfies the requirement of correctly exploring each target. (2) If the target point the agent wants to go to next is the target point where the agent is currently located, then the agent travels around the target at a specified speed (< the maximum exploration speed) in a direction away from the target, and the travel time is the revisit time interval of the target.

[0143] S3: Task allocation strategy and path optimization algorithm.

[0144] Based on the target attribute information obtained in step S2, in order to reasonably allocate target exploration tasks to multiple agents, this step designs a task allocation strategy and path optimization algorithm, namely the HGALSA algorithm, considering the characteristics of multiple target visits and target revisit time intervals. The HGALSA algorithm combines an improved greedy algorithm and an improved genetic algorithm, with the optimization objective of minimizing the total exploration task time. The specific steps are as follows:

[0145] S31: Generate the initial optimal solution:

[0146] Step 1: In the first round of selection, the order of the agents is randomly shuffled, and each agent selects the target closest to itself.

[0147] Step 2: In the nth round of selection, sort the total travel time of each agent in the previous n-1 rounds to the target. The agent with the longest total travel time has priority. Each agent selects the target closest to its current position in turn according to the total travel time in the previous n-1 rounds, and checks whether the time to reach the next target is less than the target revisit interval. If the time to reach the next target is less than the target revisit interval, then proceed to the next target; otherwise, select the current target. The number of times the selected target needs to be explored is reduced by 1.

[0148] Step 3: Repeat steps 1 and 2 until all targets have been explored 0 times, and obtain a series of initial solutions with the shortest target exploration path time for the agents.

[0149] S32: Optimize the initial optimal solution

[0150] Step 1: Use tournament selection. Randomly sample s individuals from the population with replacement. Then select the best individual to enter the next generation. Only when the fitness value of an individual is better than the other s-1 competitors can it win the tournament.

[0151] Step 2: Perform a crossover operation on the individuals selected in Step 1, exchanging chromosomes to inherit the superior features of the parent string into the substring, thus generating new superior individuals. In this algorithm, the number of times each target needs to be explored is fixed. During the crossover operation, target indices are inserted, and duplicate target indices are removed from the new solution after chromosome exchange. Figure 4 As shown in (a).

[0152] Step 3: Perform mutation operations on the individuals selected in Step 1 (tournament selection). The algorithm achieves chromosome mutation by randomly swapping one target index. Figure 4 As shown in (b).

[0153] Step 4: Repeat steps 1-3, using a shorter task time to retain the solution in each iteration. By adjusting the population size, the globally optimal solution for the target task assignment is finally obtained.

[0154] S33: Implement the path optimization algorithm in programming software.

[0155] The path optimization algorithm proposed in this step has been implemented in Python software. The following code is the pseudocode of the main function of the path optimization algorithm, where targets_demand is from Table 6.

[0156]

[0157] The algorithm for generating the initial optimal solution for the path is quite complex, and its pseudocode is shown below:

[0158]

[0159] After running the code, the target exploration tasks and task durations assigned to each agent are shown in Table 7 below. The initial task allocation path is as follows: Figure 5 As shown, the initial estimate of the final target exploration mission duration is 38.59 minutes.

[0160] Table 7 Target Task Allocation Table

[0161]

[0162] This example runs well during simulation, such as Figure 6 As shown, the eight agents successfully entered their respective planned routes and navigated along the predetermined paths. When a target was detected, they conducted effective exploration according to the designed steps. All agents navigated smoothly, and the program ran smoothly. According to the simulation results, the multi-agent region coverage search and task allocation path optimization algorithm achieved the following: agent region coverage rate of 100%, target correct detection rate of 100%, and total task completion time of 76 minutes.

[0163] This embodiment also proposes a multi-agent coverage search and task allocation path optimization device. This device includes a processor and a memory, which are coupled. The memory stores program instructions, and when these instructions are executed by the processor, the aforementioned task management method is implemented. The processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The memory may include Random Access Memory (RAM) or Non-Volatile Memory, such as at least one disk storage device. The memory can be an internal memory of the Random Access Memory (RAM) type. The processor and memory can be integrated into one or more independent circuits or hardware, such as an Application Specific Integrated Circuit (ASIC). It should be noted that when the computer program in the aforementioned memory is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention.

[0164] This embodiment also proposes a computer-readable storage medium storing computer instructions for instructing a computer to execute the aforementioned multi-agent coverage search and task allocation path optimization method. The storage medium can be an electronic medium, magnetic medium, optical medium, electromagnetic medium, infrared medium, or a semiconductor system or propagation medium. The storage medium may also include semiconductor or solid-state memory, magnetic tape, removable computer disk, random access memory (RAM), read-only memory (ROM), hard disk, and optical disc. Optical discs may include optical disc-read-only memory (CD-ROM), optical disc-read / write (CD-RW), and DVD.

[0165] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A multi-agent coverage search and task allocation path optimization method, characterized in that, Includes the following steps: S1: Scenario Construction and Problem Description Regarding the problem description: The task of the multi-agent system is to perform a full-coverage search and exploration of multiple targets within a given area. In terms of scene construction: Set the task area. Set the number and attributes of agents and targets; S2: Regional Coverage Model Design S21: Establish the model objective function, including calculating the time required for each agent to allocate different detection paths, taking the maximum detection time of all agents as the task time for full coverage search of the region, and taking minimizing the task time as the optimization objective. S22: Establish model constraints, including agent region coverage constraints, agent detection time constraints, and agent scenario assumption constraints; S23: Solve the region coverage model; S3: Constructing task allocation strategies and path optimization algorithms S31: Generate an initial optimal solution, which includes the target exploration path for each agent; S32: Optimize the initial optimal solution, use a genetic algorithm for iterative optimization, and obtain the final multi-agent coverage search and task allocation path optimization strategy; In S22, the specific constraints on the coverage of the agent's region are as follows: The area coverage adopts the channel division method. Based on the initial position, maximum sailing speed, and detection range radius of the agent, the number of channels for each agent is reasonably allocated to ensure that all channels cover the task area, and the channel coverage area shall not exceed the width of one detection radius of the task area. The full coverage search constraint formula (2) for the agent's task area is as follows: (2) In the formula, k, the agent number, k∈(1,2,3,…,n) The number of waterways that the k-th agent plans to explore. =1,2,3…., The detection range radius of the k-th agent, in meters. R, the maximum detection radius of all agents, in meters. The difference between the maximum and minimum ordinates of the mission area. =max Lat – min Lat, unit: m; The specific constraints on the agent's detection time are as follows: The navigation time of the intelligent agent's route planning scheme is calculated based on the detection radius and maximum navigation speed of each agent. The constraint is shown in the following formula (3): (3) In the formula, d1, the distance from the agent's initial position to the assigned starting point of the course, in meters; d2, the path distance assigned to the agent's course, in meters; The maximum speed of the kth agent; The specific constraints of the agent scenario assumptions are as follows: Based on the mission area, the detection radius of each agent, and the maximum navigation speed, the course range of each agent and the range of the total number of courses are determined, and scenario assumption constraints are established, as shown in the following formulas (4) and (5): (4) (5)。 2. The multi-agent coverage search and task allocation path optimization method according to claim 1, characterized in that, In S1, the task area is a convex polygonal region, and is further approximated as a rectangle; The process of setting the number and attributes of agents and targets includes: There are n agents outside the mission area and m unknown targets within the mission area. Each agent has different attributes, including initial position, maximum speed, detection radius, detectable angle, and maximum detection speed. Each target has different attributes, including location, number of times to be explored, exploration radius, exploreable angle, and target revisit time interval.

3. The multi-agent coverage search and task allocation path optimization method according to claim 1, characterized in that, In S1, regarding the problem description: The full-coverage search of the region is achieved by minimizing the detection task time by reasonably allocating the detection task area of ​​the intelligent agent, thereby achieving full-coverage search of the sea area and obtaining information on all unknown targets within the task area; The multi-target exploration is a task allocation problem involving multiple agents exploring multiple targets multiple times. Specifically, after the area coverage search, exploration tasks need to be performed on the detected targets. By reasonably allocating the target exploration tasks for each agent, the exploration task time can be minimized, thereby improving exploration efficiency and accuracy.

4. The multi-agent coverage search and task allocation path optimization method according to claim 2, characterized in that, In S1, the constraint on the target revisit time interval is: When agent A enters the detection range of target B and completes a valid detection, it needs to leave the detection range of target B. If the time spent leaving the detection range of target B exceeds the revisit interval of target B, agent A can re-enter the detection range of target B and is considered to have conducted a second valid detection of target B. The constraint on the target revisit time interval applies only to repeated explorations of a single target by a single agent; explorations of the same target by different agents are not restricted.

5. The multi-agent coverage search and task allocation path optimization method according to claim 1, characterized in that, In S21, the objective function for minimizing the task time model is as follows: (1) In the formula, T k The time taken for the k-th agent to complete the planned path.

6. The multi-agent coverage search and task allocation path optimization method according to claim 2, characterized in that, In S31, during the process of generating the initial optimal solution, specifically... Includes the following steps: S31-1: In the first round of selection, the order of the agents is randomly shuffled, and each agent chooses the target closest to itself; S31-2: In the nth round of selection, sort the total travel time of each agent to the target in the previous n-1 rounds. The agent with the longest total travel time has priority. Each agent selects the target closest to the current position in turn according to the total travel time in the previous n-1 rounds, and determines whether the time to travel to the next target is less than the target revisit time interval. If the time to travel to the next target is less than the target revisit time interval, then travel to the next target. Otherwise, select the current target, and the number of times the selected target needs to be explored is reduced by 1. S31-3: Repeat S31-1 and S31-2 until all targets have 0 exploration attempts, and obtain a series of initial solutions with the shortest target exploration path time for the agents.

7. The multi-agent coverage search and task allocation path optimization method according to claim 2, characterized in that, In S32, optimizing the initial optimal solution specifically includes the following steps: S32-1: Tournament selection is adopted. s individuals are randomly sampled from the population with replacement. The best individual is then selected to enter the next generation. Only when the fitness value of an individual is better than the other s-1 competitors can it win the tournament. S32-2: Perform crossover operations on the individuals selected in S32-1 through the tournament. By exchanging and combining chromosomes, the superior characteristics of the parent string are inherited to the substring, thereby generating new superior individuals. The number of times each target needs to be explored is fixed. When performing the crossover operation, the target sequence number combination is inserted, and duplicate target sequence numbers are removed from the new solution after exchanging chromosomes. S32-3: Perform mutation operations on the individuals selected in the tournament in S32-1. The algorithm achieves chromosome mutation by randomly swapping 1-2 target indices. S32-4: Repeat S32-1 to S32-3, using a shorter task time to retain the solution in each iteration. By setting the population size, the global optimal solution for the target task allocation is finally obtained, serving as a multi-agent coverage search and task allocation path optimization strategy.

8. An electronic device, comprising a memory and a processor, characterized in that, The processor is used to execute the program in the memory to implement the multi-agent coverage search and task allocation path optimization method as described in any one of claims 1 to 7.

9. A storage medium containing computer-executable instructions, characterized in that, When executed by a computer processor, the storage medium containing the computer-executable instructions is used to perform the multi-agent coverage search and task allocation path optimization method as described in any one of claims 1 to 7.

Images