Recognizing surface properties using impedance perception

When the end-effector of a robot is slid on a flat surface, the stiffness matrix obtained by linear fitting of position and force contains information not only on the stiffness coefficient and on the normal direction, but also on the friction coefficient. This paper proposes a new method for extracting information on those properties from the stiffness matrix provided by the impedance perception, which the authors previously proposed. The proposed method can be implemented as an encapsulated module for perception, which is separated from any control and planning methodologies. Therefore, this can be used for both autonomous and remote controlled robots, and even for monitoring a manipulation tasks performed by humans. Results of preliminary experiments are presented.