Cyl-IRRT*: Homotopy Optimal 3D Path Planning for AUVs by Biasing the Sampling Into a Cylindrical Informed Subset
Fujie Yu, Yuan Chen
Abstract
In a complex 3-D environment, efficiently and safely reaching the target position is of great significance for autonomous underwater vehicles. This article proposes a cylinder-based informed rapid exploration random tree (Cyl-iRRT*) algorithm, which seeks to find the homotopy optimal path by focusing the search space on the designed gradually shrinking cylinder. The proportional shrinkage method and obstacle-based sampling strategy are presented to yield a fast convergence response and robust stability. Furthermore, the probabilistic completeness and homotopic optimality of Cyl-iRRT* are proven to be effective. Finally, both simulation and real-world experimental results reveal the superiorities of the Cyl-iRRT* algorithm.