A Scalable Sampling-Based Optimal Path Planning Approach via Search Space Reduction

Many sampling strategies in Sampling-Based Planning (SBP) often consider goal and obstacle population and may however become less efficient in large and cluttered 3D environments with a goal distanced away. This paper presents a search-space-Reduced optimal SBP approach (RSBP) for a rigid body. This reduced space is found by a sparse search tree, which is enabled by a Metric Function (MF) built on a neural network. The offline-learnt MF estimates the minimum traveling cost between any two nodes in a fixed small workspace with various obstacles. It allows connections of two sparse nodes without path planning, where the connections represent the traveling costs (not paths). It is proven that the asymptotic optimality is preserved in the RSBP (assuming a zero-error MF) and the optimality degeneration is bounded (assuming a bounded-error MF). The computational complexity during planning is shown linear to the Lebesgue measure of the entire search space (assuming the same sampling density across environments). Numerical simulations have shown that in tested large and cluttered environments the RSBP is at least as fast as the bidirectional fast marching tree* and informed rapidly exploring random tree*, with planned paths of similar optimality. The results also have shown the RSBP's improved scalability to large environments and enhanced efficiency in dealing with narrow passages.

Abstract licence: CC BY-NC-ND

40 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.