Probabilistic Roadmaps: A Motion Planning Approach Based on Active Learning

Motion planning is the ability that an autonomous robot must possess to compute its motions, in order to perform tasks such as navigating from one location to another, assembling a product, fetching an object, building a map of an environment, inspecting a structure, tracking an un-predictable target, or climbing a cliff. A new motion-planning approach - Probabilistic RoadMap (PRM) planning - has emerged, which takes advantage of such techniques. The talk will discuss how a better understanding of these properties is already making it possible to design faster PRM planners capable of solving increasingly more complex problems