Visual Tracking Based Multi-Robot Cooperative Localization And Formation

Multi-robot system is a complex system which could perform information sharing and cooperation through a series of communication and collaborative algorithm.Compared with single robot system,it has wider application in many aspects and can solve lots of problems that can't be solved by a single robot with unparalleled advantages.The localization and cooperative control are the first two problem to be solved in the multi-robot system.In this paper,a multi-robot system with Leader and Followers is designed to study the localization and formation problem based on the experimental requirements.A follower is divided into sensing layer,driver layer,network communication layer and arithmetic control layer.The follower owns self-localization and cooperative localization functions and information can be shared through the wireless network among the robots.The follower is equipped with a camera to perceive the external environment,using the color feature to detect the moving target in a dynamic background and projection to determine its image coordinates.In order to keep the target in the center of the image,a double closed loop PID visual tracking controller is designed with the image coordinates as feedback signal and the controller parameters are set on-line.After the camera is aimed at the target,the follower can obtain the relative distance and angle between the follower and the moving target via specific sensors.In order to determine the relative position between the multi-robot group and the external landmark,these local information in environments is used to locate the robots by Kalman filter.This paper study the self-localization method and cooperative localization method in a multi-robot team.The former is easy to implement and expand,but carrying a cumulative error caused by the leader locating itself by dead reckoning.The latter makes better use of the observation information of multiple robots without accumulate error,but takes up more computing and communication resources especially in a centralized implementation.A distributed computing and control method can reduce the occupation of resources and is easy to transplant on a new follower,but the cost is the loss of localization accuracy.After obtaining the pose of the robots in the environment,Leader-Follower method and virtual structure method is researched to complete the multi-robot formation control problem.The Leader-Follower select the l-? strategy to establish the error model and design its formation controller.The virtual structure method transform the formation problem in multirobot team into a trajectory tracking problem in a single robot,then design the trajectory tracking controller by utilizing the Backstepping way to construct Lyapunov functions.Simulation tests on Matlab show that the latter formation method has a better performance.Finally,the virtual structure formation method is employed on the practical system under the self-localization method,showing good formation efforts.