An Adaptive Smoothing RRT Method for Path Planning of Concentric Cable-Driven Manipulators

Abstract

Concentric cable-driven manipulators (CCDMs) are dexterous enough to be widely used in confined space. While how to adaptively plan a smooth end path for CCDMs has become a key issue. In this paper, an Adaptive Smoothing Rapidly exploring Random Trees (AS-RRT) method is proposed for path planning of CCDMs. Firstly, the binocular vision is used to detect the target node and obstacles to further establish complete coordinates of oral environment. Secondly, the sampling convergence optimization strategy and the target gravitational bias strategy are detailed to adaptively optimize the target orientation and convergence speed. Thirdly, the polynomial smoothing optimization function is used to prune redundant branch paths and improve the smoothness of planned paths. Finally, experiments are carried out to verify the proposed method. Results show that errors between the actual path and the planned path of CCDMs are less than 1.055 mm. In that case the feasibility of the AS-RRT method for path planning of CCDMs is verified. In addition, the method is applicable not only to CCDMs, but also to many cable-driven manipulators with similar configurations.