DocumentCode :
38509
Title :
C-FOREST: Parallel Shortest Path Planning With Superlinear Speedup
Author :
Otte, Michael ; Correll, Nikolaus
Author_Institution :
Lab. for Inf. & Decision Syst., Massachusetts Inst. of Technol., Cambridge, MA, USA
Volume :
29
Issue :
3
fYear :
2013
fDate :
Jun-13
Firstpage :
798
Lastpage :
806
Abstract :
C-FOREST is a parallelization framework for single-query sampling-based shortest path-planning algorithms. Multiple search trees are grown in parallel (e.g., 1 per CPU). Each time a better path is found, it is exchanged between trees so that all trees can benefit from its data. Specifically, the path´s nodes increase the other trees´ configuration space visibility, while the length of the path is used to prune irrelevant nodes and to avoid sampling from irrelevant portions of the configuration space. Experiments with a robotic team, a manipulator arm, and the alpha benchmark demonstrate that C-FOREST achieves significant superlinear speedup in practice for shortest path-planning problems (team and arm), but not for feasible path panning (alpha).
Keywords :
manipulators; path planning; sampling methods; search problems; trees (mathematics); C-FOREST; alpha benchmark demonstrate; avoid sampling; better path; irrelevant portions; manipulator arm; multiple search trees; parallel shortest path planning; parallelization framework; prune irrelevant nodes; robotic team; shortest path-planning problems; single-query sampling-based shortest path-planning algorithms; superlinear speedup; trees configuration space visibility; Algorithm design and analysis; Computer architecture; Path planning; Planning; Robots; Vegetation; Distributed computing; efficiency; parallelization; path planning; robots; superlinear speedup;
fLanguage :
English
Journal_Title :
Robotics, IEEE Transactions on
Publisher :
ieee
ISSN :
1552-3098
Type :
jour
DOI :
10.1109/TRO.2013.2240176
Filename :
6425493
Link To Document :
بازگشت