A multi-layered path planning algorithm for truss climbing with a biped robot

Abstract

Autonomous climbing is an essential function for a climbing robot to be applied in practical high-rise work. And for autonomous climbing, perception of climbing environment and planning of climbing path are two basic issues. Truss-style environment modeling and global path planning are still open for a biped climbing robot. In this paper, RGBD-SLAM is used to model truss-style climbing environment, in terms of pointcloud. Based on the model in pointcloud format, a multi-layered algorithm is presented to plan a global path along the poles in a truss. The algorithm can solve the problem of local optima and the discrete caused by pointcloud. The effectiveness of the modeling method and the planning algorithm is verified by experiments, where the time taken in the planning is less than 0.25s in a pointcloud model with 454411 nodes.