A Machine Learning Approach for Feature-Sensitive Motion Planning

Abstract

Abstract. Although there are many motion planning techniques, there is no method that outperforms all others for all problem instances. Rather, each technique has different strengths and weaknesses which makes it best-suited for certain types of problems. Moreover, since an environment can contain vastly different regions, there may not be a single planner that will perform well in all its regions. Ideally, one would use a suite of planners in concert and would 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 regions that are well suited to one of the methods in our library of roadmapbased motion planners. After the best-suited method is applied in each region, the resulting region roadmaps are combined to form a roadmap of the entire planning space. Over a range of problems, we demonstrate that our simple prototype system reliably outperforms any of the planners on their own.

Effective Metrics for Multi-Robot Motion-Planning

“…This implies that a set of metrics, each suitable for a different setting, can be combined in order to solve more diverse settings that consist of smaller, specific, (sub)settings. Morales et al [35] have the same observation that different portions of the C-space may behave differently. In our work we will also refer to the case where different metrics are more effective than others in different portions of the C-space.…”

Realtime Informed Path Sampling for Motion Planning Search

“…Another recent adaptive approach is to construct a meta-planner with several tools at its disposal; such planners employ multiple sampling strategies (Hsu et al, 2005) or multiple randomized roadmap planners (Morales et al, 2004), based on a prediction of which approach is most effective in a given setting.…”

SmartApps: Middle-ware for Adaptive Applications on Reconfigurable Platforms

“…The adaptive algorithm selection framework in STAPL is a generalization of previous work done by our group for the adaptive selection among multiple algorithmic options for parallel sorting [49,2], for selecting among different motion planning methods [33,35], and for performing parallel reduction operations [83,84]. To the best of our knowledge, our framework provides the first general methodology for automatically developing a model for selecting among multiple algorithmic options.…”