Decentralized Model Predictive Control for Multi-Robot Coordination

This is a path planning problem, where the task of the robots is to safely reach a target area located on the right border of the screen. We present a decentralized approach for coordinating multiple robots with possibly conflicting interests over a continuous state space. It is an open-loop receding horizon control scheme for planning and acting under uncertainty. Based on an alternating improvement of the agents' plans it is real-time capable and provably converges to a Nash equilibrium plan.