LAUSR

Research on longitudinal motion control has so far mainly focused on high level planning and control techniques in high speed situations, for example highway driving, and the worst case scenarios there. Low level acceleration control is found only rarely in literature. With the development of automated vehicles, the controllers need to be able to handle the worst case scenarios at low velocities, such as traversing obstacles carefully. In this paper we present a novel approach for a low level acceleration controller that uses its knowledge of the road profile ahead to control engine and brakes proactively. The main contribution is the introduction of a suitable model predictive controller and its implementation in a real vehicle. In addition, the proposed control loop can be easily enhanced by employing existing control approaches. We present simulated data as well as experimental results for a state-of-the-art literature approach and our model predictive controller. Our data clearly shows that the control performance is significantly improved by our predictive solution.