Locomotion System Considering Foothold Information Based on Vision and Force Sense

Abstract

Humanoid robots have potensials to walk on uneven ground unlike wheeled robots. However, it is difficult to reach a given destination without falling down based on only visual sensors. To reach the destination safely, we propose the autonomous safe locomotion system applying vision and sole reaction force to the footstep planning. Moreover, the robot reuses the results of the motion by saving them in the database with the foothold label given by the visual classifier. Experimental results using a real bipedal robot CHIDORI with the propsed system are shown at last.