A Path Planning Method Based on Theta-star Search for Non-Holonomic Robots

Abstract

This study presents an improved path planning method for non-holonomic robots based on the Theta-star search. Theta-star search produces any-angle waypoints as long as each pair of waypoints satisfies line-of-sight constrain in grids with blocked and unblocked cells. However, the waypoints of Theta-star search are usually selected around obstacles/walls' edges, causing the robot to drive at a minimal collision-free distance from obstacles/walls. Therefore, we propose a new evaluation function that considers surrounding obstacles so that the generated waypoints are relatively equidistant to obstacles. On the other hand, straight lines connecting the waypoints in the Theta-star search cause sharp turns which are not feasible for non-holonomic robots. Thus, a smoothing method using ReedsShepp (RS) curves is utilized to produce kinematically feasible curves for non-holonomic robots. In the end, we demonstrate the proposed method in simulations via different scenarios. The method can be applied in diverse indoor applications, such as storage, indoor agriculture, and other applications. © 2022 IEEE.