Distributed model predictive control and virtual force obstacle avoidance for formation of nonholonomic agents

Abstract

In this paper, the problem of formation control for a group of nonholonomic mobile robots is presented. Obstacle avoidance is also included in the problem formulation using virtual force method. An objective function is minimized to develop a distributed constrained model predictive control law for a formation of unicycle mobile robots. In the leader-follower formation under consideration, a desired trajectory is given to the leader. The followers keep a specified distance and orientation while following the leader to keep a certain formation. All the agents in the formation avoid obstacles by producing a local virtual force with respect to their distance from the obstacle. Simulations for nonholonomic mobile robots formation with obstacle avoidance are performed to test the effectiveness of the proposed distributed predictive control.