Research On Adaptive Formation Strategy In Cooperative Multi-robot Capture

In the rapid development of robotics technology,collaborative multi-robot technology has attracted more and more attention from people.It plays very significant practical role in environmental detection,target searching,tracking,rounding,robot soccer,and national defense military,among which the formation rounding is one of the key issues for people to discuss multi-robot technology research.Formation rounding of collaborative multi-robot refers that a group of robots with cooperative mechanisms successfully capture the target through a certain formation strategy together with optimization algorithms.The core of this issue is to design intelligent formation algorithms and apply them to specific hunting scenes to rounding targets in a “Rapid and Low consumption” way.Rigid structure algorithm is one of the typical algorithms for formation problems.It is used to coordinate the overall structure of each robot unit to maintain a stable,and it solve the problem of target following and formation consistency effectively,playing an important and applicable role in maintaining and stabilizing the formation.However,the rigid structure causes a dead end to a certain extent.Event trigger mechanism can decrease the sampling frequency of multi-robot system and reduce communication consumption.It is widely used in multi-robot sampling communication problems,but this mechanism only reduces the number of sampling,ignoring the optimization problem of the sampling results,thus causing a large number of redundant invalid information exiting,which is called the non-intelligence of the trigger mechanism.A lot of research has been conducted at home and abroad for rounding,but the following problems remain to be solved due to the rigidity of the rigid structure and the non-intelligence of the event trigger mechanism:(1)It excessively focuses on stability and neglects the flexibility of the formation,which is the contradiction of rigid structure algorithm.A stable and rigid structure forms between every hunter,and causes the hunting system dead ends,making it impossible for hunters to complete the hunting accurately and efficiently.In this case,rigid formation will cause great waste of resources and even result in endless loops.(2)The traditional event trigger mechanism lacks intelligence.Although the frequency of sampling is reduced to a certain extent in a trigger condition,the results are not optimized directly.Thus,thereis still redundant and useless sampling information.(3)At present,the escaping goal generally has a low intelligence,and also the pursuing speed is often greater than the escaping speed(the optimal situation is keeping the equation of two speed),which is not persuasive for the effectiveness of hunting experiments.In view of the above disadvantages,the main research content and innovation are as follows:(1)Aiming at the problem of hunting dead angle of rigid structure algorithm,an adaptive rigid formation algorithm is proposed.In the algorithm,the hunter can adaptively adjust the formation to complete the rounding according to the location of the target and boundary and other environmental conditions.By studying the structure formation center controller and the formation controller,the formation center gradually approaches the goal in order to find out the encirclement.Meanwhile,the formation controller can dynamically adjust the formation according to the position of the target and avoid the occurrence of dead angle.(2)For the current event triggering mechanism,the accuracy of the sampling results cannot be improved accurately,so the communication volume cannot be well reduced,and the effectiveness of the information cannot be improved.This paper designs an event triggering mechanism based on prediction and strives to improve the effectiveness of sampling and decrease the communication cost.(3)To overcome the low energy problem of the goal,this paper designs an intelligent escape program that avoids a single escape direction,and the target can escape towards the direction without catchers when detecting a catcher.In addition,we discuss the speed relationship between the hunter and the target by Apollonius,and make the hunting speed less than the target's escaping speed within a reasonable range,increasing the intelligence and hunting difficulty of the target.(4)We performs a multi-robot pursuit simulation platform based on Netlogo.The algorithm and strategy can be simulated and verified on the platform,and the experimental process and results can be displayed intuitively.