Robotic Rock Climbing using Computer Vision and Force Feedback

Abstract

Climbing robots that climb flat structures using suction cups or magnets are commonly described in the literature. However, robots that can autonomously find randomly placed handholds, and then plan and climb a route up the walls using those handholds, have not been described. A low cost robot has been designed by Dartmouth College students to climb a near-vertical indoor climbing wall using computer vision to locate handholds, and force feedback to maintain pressure on handholds. Using a variation of the probabilistic roadmap algorithm, a climbing route up the wall is planned and then executed.