Safe Navigation of a Car-like Robot within a Dynamic Environment

This paper addresses the problem of navigation of a carlike robot in dynamic environments. Such environments impose a hard real time constraint. However, computing a complete motion to the goal within a limited time is impossible to achieve in most real situations. Besides, the limited duration validity of the model used for planning requires the model and therefore the plan to be updated. In this paper, we present a Partial Motion Planning (PMP) approach as the answer to this problem. The issue of safety raised by this approach is addressed using the Inevitable Collision State formalism and effectiveness of the approach is demonstrated with several simulation examples. The quality of the generated trajectories is discussed and continuous curvature metric is integrated as a mean to improve it.