DR-Informed-RRT* Algorithm: Efficient Path Planning for Quadruped Robots in Complex Environments.

Haoyu Jiang,Qiao Zhou,Yulong Zhang,Xungao Zhong,Xunyu Zhong

DOI: https://doi.org/10.1007/978-981-99-6501-4_8

2023-01-01

Abstract:Quadruped robotics system possesses superior trafficability. However, path planning for these robots in complex environments continues to present challenges. We propose a DR-Informed-RRT* (Dual-attached and reconstructed informed-RRT*) algorithm to address issues such as poor search optimality, low sampling efficiency, and excessive planning time. The path planning process is divided into two parts: initial path generation and path optimization. In the initial path generation phase, a dual sampling point approach with multiple sampling probabilities is employed to sample random points, reducing the number of futile node expansions and providing directionality to the growth of the random tree, thereby improving the algorithm’s efficiency. Furthermore, the reconstruction of parent nodes and the elimination of redundant nodes are incorporated in the path optimization phase, introducing different types of parent nodes to optimize the path and reducing redundancy in the planned path. Our algorithm showed promising performance in simulations and successfully validated its feasibility on a quadruped robot.