Analysis of an on-line minimum-time velocity optimization algorithm

Examination of the control of a high-speed vehicle is essential to achieve autonomous navigation on high speed roads. Velocity profile generation is one of the challenging problems. In this paper we analyze the efficiency of an on-line velocity profiler algorithm. The presented method is based on a receding horizon implementation in which the vehicle executes part of the computed optimal trajectory, while simultaneously optimizing the path up to a new horizon. The search space at each optimization step is considerably reduced, which results in reduced computational cost. We present an analysis of the horizon length selection and the impact on the running times compared to the infinite-horizon solution.