Sliding Local Planners for Sampling-Based Path Planning

Abstract

Sampling-based path planning algorithms based on straight-line local planners, even in combination with advanced sampling strategies, occasionally perform poorly when a rigid robot needs to pass through narrow passages in the C-space. In order for a rigid robot to effectively navigate C-space narrow passages, we present two simple sliding local planners for sampling-based path planning. These planners either slide to the workspace medial axis or to the workspace boundary. In addition, we also propose a parallelized bidirectional RRT, a highly efficient global path planner. Our proposed local planners when integrated within our proposed global path planner can solve some of the benchmark basic motion planning problems more efficiently than possible with a straight-line local planner in combination with advanced sampling strategies. We observed only small variations in solution time when our global planner integrated our local planners with advanced sampling strategies. Our experimental results underscore the effectiveness of our proposed local planners in solving some of the basic motion planning problems.