AAAI Publications, Twenty-Sixth International Conference on Automated Planning and Scheduling

Font Size: 
Strict Theta*: Shorter Motion Path Planning Using Taut Paths
Shunhao Oh, Hon Wai Leong

Last modified: 2016-03-30

Abstract


A common way to represent dynamic 2D open spaces in robotics and video games for any-angle path planning is through the use of a grid with blocked and unblocked cells. The Basic Theta* algorithm is an existing algorithm that produces near-optimal solutions with a running time close to A* on 8-directional grids. However, a disadvantage is that it often finds non-taut paths that make unnecessary turns. In this paper, we demonstrate that by restricting the search space of Theta* to taut paths, the algorithm will, in most cases, find much shorter paths than the original. We describe two novel variants of the Theta* algorithm, which are simple to implement and use, yet produce a remarkable improvement over Theta* in terms of path length, with a very small running time trade-off. Another side benefit is that almost all paths found will be taut, which makes more convincing paths.

Full Text: PDF