摘要：AbstractIn this paper we deal with model-based control design in the presence of uncertainties. We use Stochastic Dynamic Programming to solve two problems, called longitudinal and lateral vehicle control. The goal is to allow safe driving (navigation) of a moving vehicle in an environment with static obstacles. We show how to define the optimization problems given the stochastic driver behavior and environment. The vehicle dynamics model is deterministic (obeys physical laws) and is explicitly integrated into the optimization problem. In terms of results, the numerically computed control policies provide best-on-average performance (according to the expected value operator). In simulation, it is shown that the vehicle effectively avoids obstacles, thus ensuring a safe drive experience.
