Research On The Method Of Target Search Task In Multi-robot System

Search and capture refers to the behavior of searching and capturing in some known or unknown areas.The problem of multi-agent target search and tracking has a long history.In recent years,it has been widely used in various fields with high influence,such as search and rescue operations in disaster scenarios,natural resource exploration,environmental monitoring,etc.For some search and capture tasks that can not be completed by human beings,multi robot system can be used to execute,which not only greatly improves the task efficiency,but also ensures human safety during the task execution.Cooperative search and acquisition of multi robots is also a typical problem to study the cooperation among distributed multi robot systems.Because multi robot systems have the advantages of parallelism,robustness and scalability,it has become an inevitable trend to replace human beings to complete repetitive,boring and dangerous tasks.In this paper,the pursuit problem of multi robot system is taken as the research goal,and three key steps in the pursuit problem are studied respectively: target search,task allocation,and the scheme of encirclement.Its main contents include:(1)The improved search strategy design based on multi robot cooperative search.It is easy for multi robot cooperative random search strategy to fall into local self-locking due to the search mechanism during the search period,which will lead to some problems such as incomplete search of complex terrain,low efficiency of autonomous search and so on.Based on this strategy,the relative positions of robots are optimized during the search,so as to find the best moving direction for robot groups in the map,avoid the search falling into local optimization,and make them adapt to more complex environment or changeable terrain.The simulation results show that when the search area is relatively simple,the random search strategy of multi robots can achieve better average effect,but when the obstacles in the search area increase or the terrain becomes relatively complex,the improved scheme is significantly better than the previous strategy,and the search coverage is increased by about 15% in the simulation.(2)Research on the approach and form of multi robot seizing.In the past,a single guidance and capture scheme was used in multi robot seizing.The traditional fixed pattern and traditional seizing scheme may bring many unnecessary constraints,and may not meet the needs of dynamic seizing.Therefore,the corresponding mechanism of robot entrapment is improved.The space between robots is controlled by the gravitational repulsion model.Different robots respond quickly to guide to the target position when considering the moving direction and tendency at the same time.After the completion of the guidance,in the stage of encircling,in the face of a single simple encircling target,the traditional round encirclingstrategy is adopted,the target is placed in the heart of the polygon,and the robot is in the fixed position of the polygon.Compared with other complex schemes,this method is the most simple and fast.For multiple complex moving targets,implicit function is used to construct the shape of the encircling graph.According to the task target and the number of robots involved,the response is more rapid to form an unfixed shape of the encircling.The simulation results show that the overall operation of the scheme can fully solve the problem of slow response of multiple robots in the operation period compared with a single round up strategy,and reduce the redundant operation of many robots when the complex round up task is executed,and the simple processing for the simple problem also improves the efficiency of the round up.(3)Design of multi robot task assignment scheme with variable communication range.During the task allocation period of multi robots,redundant information exchange often occurs,which not only increases the energy consumption of system operation,but also extends the time of task allocation in dynamic task allocation.In order to solve this problem,a task assignment scheme with variable communication range is proposed.During the task execution,the dynamic communication range is expanded and reduced,and the centralized and distributed mode is adopted.In the initial stage,tasks are divided into different groups by auction,and robots are divided into different groups according to the number of tasks.In the process of task execution,tasks can be quickly distributed within and among small groups according to the dynamic change of group communication range,so as to improve the effect of task execution and reduce the allocation time.In addition,when some groups have too many tasks,tasks can be divided into different groups By expanding the communication range and reallocating,the local optimal situation can be avoided.The simulation results show that the local self-organizing communication mode with limited communication range can reduce the allocation time by about 30% compared with the basic auction allocation,at the same time,it can effectively reduce the information throughput and energy consumption.