Constraint-tree-driven modeling and distributed robust control for multi-vehicle cooperation at unsignalized intersections

Abstract This paper proposes a multi-vehicle cooperation method based on the virtual platoon concept for unsignalized intersections. The method consists of two levels, i.e. constraint-tree-driven modeling (CTM) strategy for the supervision level and distributed robust control (DRC) method for the execution level. After projecting the approaching vehicles to a virtual lane, the CTM strategy determines the conflict-free passing sequence and driving mode and forms multiple local virtual platoons (LVP). A constraint tree is then constructed, in which constraints with a unified equality formation are utilized to describe the geometry topology of all vehicles, regardless of leader or member. Then, a class of robust control is proposed to render each vehicle to follow its prescribed constraint. The control is designed based on third-order nonlinear vehicle dynamics. The uniform boundedness, uniform ultimate boundedness performance of the entire system and string stability of LVP have been proven by leveraging the Lyapunov stability theory. The simulation results confirm that the proposed method functions well under time-varying traffic volume. Furthermore, the proposed method significantly improves the fuel economy, driving comfort and safety under low traffic volume compared with baseline methods.