TABLE OF CONTENTS J.-P. Laumond, LAAS-CNRS, Toulouse, France
T. Sim on, LAAS-CNRS, Toulouse, France
This paper overviews the probabilistic roadmap approaches to path planning whose surprising practical performances attract today an increasing interest. We first comment on the configuration space topology induced by the methods used to steer a mechanical system. Topology induces the combinatorial complexity of the roadmaps tending to capture both coverage and connectivity of the collision-free space. Then we introduce the notion of optimal coverage and we provide a paper probabilistic scheme in order to compute what we called visibility roadmaps [26]. Reading notes conclude on recent results tending to better understand the behavior of these probabilistic path planning algorithms.
The framework of this work lies in the tentative to provide path planners working for large classes of mechanical systems. Such a generality is imposed by an increasing number of path planning applications that extend the robotics area where the research has been initially conducted [19].
In our case, we are interested in providing CAD systems with path planning facilities in the context of logistics and operation in industrial installations. A typical scenario is the following one. An operator has to define a maintenance operation involving the moving of an heavy freight. He has CAD software including all the geometric details and facilities to display the plant together with catalogue containing lists of available tools to perform the maintenance task. He should choose suitable handling devices among cranes, rolling bridges, carts... and validate his choice by simulating the task...
