Locally optimal motion planning of nonholonomic systems

In this article, two algorithms for motion planning of nonholonomic systems based on the Hermes algorithm are presented, called a generalized algorithm and a modified generalized algorithm. The generalized Hermes algorithm gives a locally optimal trajectory while the modified generalized algorithm gives satisfactory results, in the sense of a cost criterion function, with moderate computational effort. The only assumption imposed on the system controlled is controllability. Simulation results presented confirm usefulness of these two algorithms in solving the motion planning task. (C) 1997 John Wiley & Sons, Inc.