Model Predictive Longitudinal Motion Control for the Unmanned Ground Vehicle with a Trajectory Tracking Model

Longitudinal motion controllers based on over- simplified models result in steady-state errors, oscillations, and overshoots of the velocity, all of which impair the unmanned ground vehicle (UGV) multiple objectives (trajectory tracking capability, energy economy, and ride comfort). While it is challenging for complicated methods to accomplish real-time control in the vehicle-mounted electronic control unit (ECU), which meets harsh working conditions but has limited computing power. This paper proposes a multi-objective model predictive longitudinal motion control approach for an ECU based on a trajectory tracking model. For the trajectory tracking model, we establish an internal powertrain/brake dynamics model and divide the external resistance into three components. They are identified using a variety of practical and straightforward methods. The control strategy is composed of a high-level multi-objective model predictive controller (MPC) that obtains the optimal internal acceleration and a low-level controller that calculates control inputs based on the powertrain/brake model. To implement real-time control on the ECU, the MPC optimization problem and its solver that combines offline and online calculation are elaborately designed. Finally, the method is validated through simulations in three typical driving scenarios and vehicle tests on a hybrid-electric sport utility vehicle (SUV) and a 6-speed dual- clutch transmission (DCT) fuel sedan. Simulations and experiments demonstrate that the proposed approach is superior to the other four commonly used methodologies in terms of achieving multiple objectives.