In this paper we consider the detection of small surface features, such as ridges and bumps, on the surface of an object during dextrous manipulation. First we review the representation of object surface geometry and present de nitions of surface features based on local curvature. These de nitions depend on the geometries of both the robot ngertips and the object being explored. We also show that the trajectory traced by a round ngertip rolling or sliding over the object surface has some intrinsic properties that facilitate feature detection. Next, several algorithms based on the feature de nitions are presented and compared. Finally, we present simulated and experimental results for feature detection using a hemispherical ngertip equipped with an optical tactile sensor.
Allison M. Okamura, Mark R. Cutkosky