Brake-to-steer: Differential braking-based discrete-valued predictive path tracking control for autonomous ground vehicles

Differential braking has the potential for path tracking control (PTC) for autonomous ground vehicles. Taking advantage of the discrete ON/OFF nature of the inlet and outlet valves used for wheel cylinder pressure regulation in the hydraulic braking system, this article proposes a discrete-valued predictive PTC (DV-P-PTC) solution using rear differential braking. Firstly, defining the finite control combinations of inlet and outlet valves, the control-oriented path tracking model with a discrete-valued input is built. Secondly, the DV-P-PTC solution is presented. The pressure, yaw rate, and yaw angle predictions are performed by applying all candidate control combinations to the discrete-valued path tracking model. A predefined cost function, including pressure cost, yaw rate cost, and yaw angle cost, evaluates these predictions. The optimal prediction and corresponding control combination can thereby be determined. With the above formulation, the PTC has been converted to an optimization problem with a discrete-valued decision variable. The sequential evaluation strategy is adopted to avoid weighting factors tuning. The three sub-cost functions are evaluated in a given order and rule. Finally, a hardware-in-the-loop test bench is built to implement the proposed DV-P-PTC solution. The experimental results in the double lane change scenarios validate its feasibility and effectiveness.