Distributed Model Predictive Control w-ith Obstacle Communication

This paper presents a distributed model predictive control approach for multiagent systems, where the agents are vehicles required to move from given initial positions to individual target points in the presence of obstacles. A global cost function is approximated such that each agent minimizes a local cost function based on locally available information. Within a given communication distance agents exchange their position and target information together with information about detected obstacles, which are represented by ellipses. Simulation studies illustrate the proposed method.