Switched Model Predictive Controller for Cruise Control of Heavy Trucks with Heuristic Trajectory Planning

Abstract This paper introduces a model predictive control algorithm approach, which works as a cruise controller with automatic calculation of a speed level and trajectory. It aims on cost reduction for heavy trucks on long distance motorway operation compared with standard cruise controllers with driver adjusted speed level. It is assumed that the brake and the engine of the truck must be controlled by the predictive control algorithm but the robotized gear box does not need to be controlled. At this stage of development the approach is based on road topology information obtained through GPS positioning, 3D maps and a simplified linear model of the truck. It is assumed that there is no interaction with other traffic or the driver. The cruise control is split into a heuristic trajectory planning level and two real time capable MPCs. The heuristic module uses a simple nonlinear model of the truck and a slope map to calculate a limited horizon speed trajectory of the truck based on rules. The rules are acquired from driver training programs which approximate cost optimal driving. The lower level real time MPCs follow the trajectory considering the truck dynamic and the disturbance through slope profiles. Switching between engine and brake control is done by a switching logic. In contrast to other approaches we wanted to evaluate the cost saving potential with the simplest but therefore real time capable implementation on a standard automotive CPU like MPC5554. The approach considers only linearised models for unconstrained MPC but shows how to deal with limited control output. Also the switching problem between a MPC for the engine and a MPC for the brake will be described. A short comparison to other approaches regarding the saving potential, the real time capability and the robustness against parameter uncertainties are given.