Vision-Based Leader–Follower Formation Control of Multiagents With Visibility Constraints

Abstract

In this brief, a vision-based tracking scheme is studied for the leader–follower formation control of multiagent system under visibility constraints without communication among agents. A kinect is the sole sensor installed on each agent to estimate the pose, including relative orientations and relative positions of the local leader. A dipolar vector field is established to generate and maintain a desired formation given the desired displacement on positions among agents. The visibility constraints of kinect are converted into input constrains given the pose of the local leader, and an auxiliary system is designed to analyze the effect of visibility constraints. Combining the state of the auxiliary system with the vector filed, the proposed control protocol is able to achieve the formation tracking without global or local communication among agents. Simulation studies on the robot operating system using the Gazebo simulator and experimental results on Turtlebot2 are presented to illustrate the performance of the proposed approaches.