A Hierarchical Extension of the D ∗ Algorithm

Abstract

In this paper a contribution to the practice of path planning using a new hierarchical extension of the D ∗ algorithm is introduced. A hierarchical graph is stratified into several abstraction levels and used to model environments for path planning. The hierarchical D∗ algorithm uses a downtop strategy and a set of pre-calculated trajectories in order to improve performance. This allows optimality and specially lower computational time. It is experimentally proved how hierarchical search algorithms and on-line path planning algorithms based on topological abstractions can be combined successfully.