Distributed Cooperative Control of Autonomous Vehicle Chassis Based on Multiagent Theory

Abstract

To solve the problems of centralized control for vehicle chassis, such as a high computational load, the difficulty of calculating a global optimal solution, and low safety redundancy, this article proposes a distributed cooperative control method for autonomous vehicle chassis based on a multiagent system (MAS) and graph theory. The control inputs of a single agent are calculated using distributed model predictive control (DMPC), which reduces the controller’s computational load. Aiming at the unknown DMPC terminal control and status sequences, this article uses the local static feedback method to predict the terminal sequence by solving the algebraic Riccati equation. In addition, the DMPC cost function is designed rationally, and the theory of consistency control is introduced. The Laplacian matrix determines the weights between various agents, simultaneously converging adjacent agents’ errors to zero and preventing system oscillation. Finally, the simulation results indicate that the proposed method has superior trajectory tracking performance and stability compared to the centralized control method. If a single agent fails, the distributed control vehicle still maintains a specific capability for trajectory tracking, there is no system failure, and safety redundancy is high.