Motion planning has received much attention over the past 40 years. More than 15 years have passed since the introduction of the successful sampling-based approach known as the Probabilistic RoadMap Method (PRM). PRM and its many variants have demonstrated great success for some high-dimensional problems, but they all have some level of difficulty in the presence of narrow passages. Recently, an approach called Toggle PRM has been introduced whose performance does not degrade for 2-dimensional problems with narrow passages. In Toggle PRM, a simultaneous, coordinated mapping of both C free and C obst is performed and every connection attempt augments one of the maps – either validating an edge in the current space or adding a configuration ’witnessing’ the connection failure to the other space. In this paper, we generalize Toggle PRM to d-dimensions and show that the benefits of mapping both C free and C obst continue to hold in higher dimensions. In particular, we introduce a new narrow passage characterization, α-ε-separable narrow passages, which describes the types of passages that can be successfully mapped by Toggle PRM. Intuitively, α-ε-separable narrow passages are arbitrarily narrow regions of C free that separate regions of C obst , at least locally, such as hallways in an office building. We experimentally compare Toggle PRM with other methods in a variety of scenarios with different types of narrow passages and robots with up to 16 dof.
Bibliographical noteKAUST Repository Item: Exported on 2020-10-01
Acknowledged KAUST grant number(s): KUS-C1-016-04
Acknowledgements: This research supported in part by NSF awards CRI-0551685, CCF-0833199, CCF-0830753, IIS-096053, IIS-0917266by THECB NHARP award 000512-0097-2009, by Chevron, IBM, by Award KUS-C1-016-04, made by King Abdullah University ofScience and Technology (KAUST). Denny supported in part by an AFS Merit Fellowship,Texas A&M University.
This publication acknowledges KAUST support, but has no KAUST affiliated authors.