Obstacle Avoidance in Distributed Optimal Coordination of Multirobot Systems: A Trajectory Planning and Tracking Strategy

Abstract

This paper studies the problem of obstacle avoidance in the distributed optimal coordination (DOC) for a class of uncertain multi-robot systems. Due to the existence of the obstacle regions, the considered optimization problem is intrinsically non-convex, which will result in the generation of some unexpected equilibriums (local minima). The existing results lack systematic obstacle-avoidance trajectory planning approaches such that the robots potentially fall in the unexpected equilibriums and cannot reach the global optimal solution. To address it, a safe reference trajectory planning approach is first designed by online projecting the unsafe part of the existing distributed optimization trajectory into the peripheral boundary of the obstacle region. On the basis, a distributed backstepping tracking control scheme is proposed based on a novel multiplicity-integral-type Barrier-Lyapunov function. It is proved that all robot systems can keep away from the unexpected equilibriums and asymptotically reach the global optimal position while avoiding collisions with moving obstacles.