An RRT* path planning for kinematically constrained hyper-redundant inpipe robot

Abstract

This paper suggests an efficient RRT* path planning for hyper-redundant inpipe robot. We solve the path planning problem for the inpipe robot in two-dimensional pipeline where the robot has in total 12 degree-of-freedom, 9 pitching angles and the pose of the reference module of the robot. Such a robot inevitably suffers from a narrow passage problem when the robot passes through a miter, a T-branch or an elbow pipe. Moreover, it is desired that the driving modules travel along the axis of pipe so that the driving wheels generate enough tracking force by making contact with the pipe inner surface. We suggest using RRT* for the path planing of the inpipe robot with the inverse kinematics to the joint location for the node extension and a cost optimization about travel distance of modules to smooth out the zigzag paths of RRT* planner. Also, we use sliding windows for random sampling in the configuration space to take advantage of pipeline topologies. The simulation results show that the proposed approach finds drivable paths more efficiently than conventional methods.