Research On Multiple Robot System Formation Control

As an important part of robotics research field, the research of multi-robot systemcooperation based on single robot, but it has the incomparable advantages than singlerobot, such as the ability of adapting to the complex environment, work efficiency ofthe system and flexibility and robustness and so on. Therefore, as a hot topic,multi-robot cooperative technology causes the attention of scholars both at home andabroad.Formation control belongs to the topic of multi-robot cooperation research field,which has been widely used in industrial, civil and military, police and aerospace fields.The content of the study includes cooperative localization, autonomous obstacleavoidance, formation keeping and control. This paper focuses on several technologiesof multi-robot formation control.First of all, location problem is the most basic and important component offormation control technology, whose positioning precision will impact the quality ofthe formation directly. In view of cooperative localization problems of the robot, thispaper proposes a kind of particle filter algorithm of particle swarm optimization basedon the Gaussian distribution. This algorithm controls particle update speed using theGaussian distribution, by driving particle distribution move to the large area ofposterior probability value and thus improves the positioning accuracy.Secondly, based on the cooperative localization, for the problem of obstacleavoidance of formation control, this paper proposes Kernel-based least squares policyiteration (KLSPI algorithm). This algorithm introduces Gaussian kernel function asprimary function in the least squares policy iterative algorithm (LSPI algorithm), whichsolves feature selection and nonlinear space generalization problem of LSPI algorithmfor artificial selection linear basis function produced and reduces the complexity of thealgorithm and thus improved its generalization performance, ensure that mobile robotcan avoid obstacles in the process of movement safely and reach target pointsuccessfully, and at the same time obtain ideal optimal trajectory.Finally, based on the above research of two contents, for the problem of formation forming, this paper proposes an improved formation forming algorithm of multi-robot.This algorithm overcome shortcomings the mid-point method resulting in inconsistentthe centre of a circle, which is a kind of formation algorithm dynamic target the centerof gravity method obtained is introduced into the algorithm of behavior decomposition.Simulation results show that the proposed approach can form our desired formationeffectively; Based on closed loop l control and KLSPI obstacle avoidance algorithm,for the problem of formation control, this paper proposes a kind of formation controlmethod for hybrid multi-robot based on Leader-Follower, which solve the problem ofthe traditional leader-follower without formation feedback mechanism and improve thestability of the formation. Simulation results are given to demonstrate the effectivenessof the proposed formation control method.