Although there are many motion planning techniques, there is no single one that performs optimally in every environment for every movable object. Rather, each technique has different strengths and weaknesses which makes it best-suited for particular types of situations. Also, since a given environment can consist of vastly different regions, there may not even be a single planner that is well suited for the problem. Ideally, one would use a suite of planners in concert to solve the problem by applying the best-suited planner in each region.
In this paper, we propose an automated framework for feature-sensitive motion planning. We use a machine learning approach to characterize and partition C-space into (possibly overlapping) regions that are well suited to one of the planners in our library of roadmap-based motion planning methods. After the best-suited method is applied in each region, their resulting roadmaps are combined to form a roadmap of the entire planning space. We demonstrate on a range of problems that our proposed feature-sensitive approach achieves results superior to those obtainable by any of the individual planners on their own.