Decentralized flocking of multi-agent system based on MPC with obstacle/collision avoidance

Most of the literature on flocking of multi-agent system with obstacle/collision avoidance make essential distinction between agents and obstacles. This paper proposes a decentralized model predictive flocking algorithm, and addresses the problems by defining obstacles as special agents, which enables agents to effectively evade external obstacles and to avoid colliding with neighbors while forming a stable a-lattice structure. We not only numerically verify the effectiveness of the algorithm with 50 agents in the MATLAB, but also achieve obstacle avoidance of ten UAVs on a software-in-the-loop simulation platform.