Active Exploration of Surfaces for Legged Locomotion of Robots

This paper presents some results of an ongoing research project in the GRASP Lab in the area of active exploration and perception for the legged locomotion of robots. We propose an active perceptual scheme that is based on the ability of the robot to extract material properties from a surface during locomotion. This ability is provided to the robotic system through a compliant sensing device which is used to monitor the response of the surface when exploratory procedures are executed during the stepping and walking motions of the leg. Such a system will actively perceive changes in the surfaces properties and prevent the robot from slipping, falling, or sinking during locomotion. The paper describes the proposed perceptual scheme, the system set-up, and the implementation of the exploratory procedures.