Measurement of ground surface displacement using stereo vision and mechanical sensors on humanoid robots

This paper presents a system for measuring ground surface displacement using stereo vision and mechanical sensors on a humanoid robot. The stereo vision system generates a set of range data for each view by a correlation method as ground surfaces can be regarded as smoothed surfaces with random texture patterns. The mechanical sensors are encoders, gyro sensors and acceleration sensors that measure the angles of joints, angular velocity and acceleration of the robot. With those angles and the link geometry, the transfer matrices between different views are generated and used to register sets of range data.