Safe and Efficient Autonomous Exploration Technique for 3D Mapping of a Complex Bridge Maintenance Environment

Abstract

This paper describes a technique for autonomously exploring a complex steel bridge environment using a 6DOF anthropomorphic robotic arm, instrumented with a laser range scanner. Potential knowledge ained from a 3D range scan at an end-effector position and orientation (pose) is estimated, then arm configurations which avoid obstacles and unknown areas are computed using an optimisation approach. Safe pose solutions are compared in terms of potential gain of new weighted-information and minimal joint movement. Both simulations and robotic platform results show exploration of unknown areas occurs in a consistent and timely manner - taking an average of 4.5secs to calculate the next safe valid robot arm poses. Complex environments, typical in bridge maintenance, can be explored using an anthropomorphic arm equipped with this technique.