A Hierarchical Motion Control Framework for Reference Trajectory Tracking of Autonomous Vehicles Considering Decoupling Vehicle Kinematics and Dynamics

With the development of automated driving intelligence, autonomous vehicles (AVs) have been developed to partially take the place of the advanced driver assistance system (ADAS), especially under closed or low-interactive scenarios, such as navigation pilot on the highway and automated valet parking, having the advantage of safety guarantee and driver workload reduction. Trajectory tracking control, one of the most crucial issues to be solved for AVs, is influenced by the vehicle system’s nonlinearities and uncertainties. Recently, there have been a series of studies that focus on this issue. However, trajectory tracking considering the highly nonlinearity is rarely investigated in the previous literatures. A hierarchical framework for decoupling vehicle kinematics and dynamics is developed in this work to handle the model’s highly nonlinearity. The upstream kinematics-based controller is designed based on the trajectory error dynamics and sliding mode control with a conditional integrator. The downstream dynamics-based controller is designed based on the vehicle error dynamics and model predictive control (MPC) algorithm. The trajectory tracking performance is demonstrated through simulations under medium-speed conditions and vehicle experiments under low-speed conditions, including various vehicle speeds and various path curvatures. Results under the typical situations show that the maximum longitudinal speed error is less than 0.5 m/s, the maximum lateral error is less than 0.2 m and the maximum heading error is less than 4°, indicating a satisfactory trajectory tracking performance.