Distributed Patrol Algorithm Design And Experimental Verification For Multi-Robot System

Security has always been one of the most important factors affecting the stable development of human society.Patrol missions for security purposes play a vital role in public areas such as markets,warehouses,stations,and even defense areas such as borders and sea areas.Due to the parallelism,robustness,scalability and other advantages of multi-robot system,it has become an inevitable trend that using multi-robot replace humans to complete repeating,boring,dangerous patrol mission.Coordination and cooperation between robots are the most decisive issue in patrolling because robots must effectively coordinate their actions to obtain information about the surrounding environment so as to ensure the safety of the whole patrol area.For the multi-robot coordination patrol problem,the main research contents are as follows:(1)A simulation experiment platform multi_robot_patrol of multi-robot patrol system was established based on ROS.The platform is suitable for any environment and any number of robot teams.In addition,in order to test the multi-robot patrol algorithm in the real environment,the physical experiment platform multi_turtlebot3 was established by using the physical robot TurtleBot3.The programs tested on the simulation experiment platform can be directly transplanted into the physical experiment,which ensures that the results obtained in the simulation environment can be verified in the real environment at any time.(2)Based on the global average idleness of the system,a distributed online patrol algorithm(EGAI)is designed to solve the multi-robot collaborative patrol problem.The robot pre-estimates the global average idleness of the system after reaching a neighbor point when making the decision,and moves to the direction that can reduce the global average idleness.To avoid the local optimum,the algorithm considers the instantaneous idleness of the neighbor points.The algorithm also considers the intention information of other robots,which can effectively avoid conflicts between robots.The results of simulation and physical experiments show that with the increase of the number of robots,the global average idleness is gradually reduced,and the performance of the system gradually improved.Comparison with five distributed state-of-the-art algorithms shows that EGAI algorithm can perform better and is more suitable for patrolling tasks with a large-scale robot team.(3)A new distributed patrol algorithm(CVFV)is designed for the problem that the EGAI algorithm has a large difference in access frequency between patrol vertices when the number of robots is small.The algorithm considers the importance of vertices based on the EGAI algorithm.The results of simulation and physical experiments show that the improved CVFV algorithm has a significant effect on improving the access frequency of patrol vertices,and finally make each vertex be accessed at nearly the same frequency.In addition,when the average point frequency is used as the evaluation criterion,compared with five state-of-the-art algorithms,the CVFV algorithm has the best performance when the number of robots is small,while the EGAI algorithm has the best performance when the number of robots is large.