pith. sign in

arxiv: 1607.06374 · v2 · pith:BAHZ27X3new · submitted 2016-07-21 · 💻 cs.RO

Interleaving Optimization with Sampling-Based Motion Planning (IOS-MP): Combining Local Optimization with Global Exploration

classification 💻 cs.RO
keywords motionglobaloptimizationplanningexplorationios-mplocallocally
0
0 comments X
read the original abstract

Computing globally optimal motion plans for a robot is challenging in part because it requires analyzing a robot's configuration space simultaneously from both a macroscopic viewpoint (i.e., considering paths in multiple homotopic classes) and a microscopic viewpoint (i.e., locally optimizing path quality). We introduce Interleaved Optimization with Sampling-based Motion Planning (IOS-MP), a new method that effectively combines global exploration and local optimization to quickly compute high quality motion plans. Our approach combines two paradigms: (1) asymptotically-optimal sampling-based motion planning, which is effective at global exploration but relatively slow at locally refining paths, and (2) optimization-based motion planning, which locally optimizes paths quickly but lacks a global view of the configuration space. IOS-MP iteratively alternates between global exploration and local optimization, sharing information between the two, to improve motion planning efficiency. We evaluate IOS-MP as it scales with respect to dimensionality and complexity, as well as demonstrate its effectiveness on a 7-DOF manipulator for tasks specified using goal configurations and workspace goal regions.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.