Multivehicle Flocking with Collision Avoidance Via Distributed Model Predictive Control
Yang Lyu,Jinwen Hu,Ben M. Chen,Chunhui Zhao,Quan Pan
DOI: https://doi.org/10.1109/tcyb.2019.2944892
IF: 11.8
2021-01-01
IEEE Transactions on Cybernetics
Abstract:Flocking control has been studied extensively along with the wide applications of multivehicle systems. In this article, the distributed flocking control strategy is studied for a network of autonomous vehicles with limited communication range. The main difference from the existing methods lies in that collision avoidance is considered a necessary condition while the vehicles are driven to follow a common desired trajectory under the proximity network. The sufficient conditions for system feasibility and stability are given by the proposed strategy. First, a centralized standard model predictive control (MPC) scheme is adopted to formulate the multivehicle flocking control problem by setting collision avoidance as an optimization constraint under the proximity network. Further, an equivalent distributed MPC (DMPC) is developed based on the consensus of local controllers under the existing framework of the alternating direction method of multiplier (ADMM). However, it may require infinite time to achieve consensus for all vehicles and, thus, the local controllers resulting in a limited number of ADMM iterations may not satisfy the given constraints. The constraints for each local controller are then modified so that the collision between vehicles is avoided all of the time. The feasibility and stability of the proposed method are analyzed under practical conditions. Simulation and experimental results show that the flocking of vehicles can track the common desired trajectory stably with no collisions by the proposed method.