A trajectory reconstruction approach for leader-following of multi-robot system

Abstract

A trajectory reconstruction approach is proposed for accurate leader-following with a desired separation distance on the trajectory. The approach aims to reconstruct the relative trajectory of the leader within a sliding time window based on the on-board sensors. The needed trajectory information of the leader relative to the follower is acquired by combining the noisy and intermittently available visual measurement with the odometry information. The situation that the leader is out of filed of view (FOV) is also taken into consideration. Moreover, instead of following the current pose of the leader, the trajectory information of a historical point on the trajectory is used as the reference input for controller, thereby avoiding the cutting-corner issue. Simulation results demonstrate the effectiveness of the proposed trajectory reconstruction approach.