Active mass estimation with haptic vision

Abstract

Real-world objects exhibit rich physical interaction behaviors on contact. Such behaviors depend on how heavy and hard it is when hold, how its surface feels when touched, how it deforms on contact, etc. Recently, there are growing needs for haptic exploration to estimate and extract such physical object properties as mass, friction, elasticity, etc. In this paper, we propose an active mass estimation method based on haptic vision. We first observe an object with active vision to extract its 3D shape and posture. Next, we estimate a contact point and contact force and apply it to cause the object to move without rotation by a robot hand. We observe changes in the contact force using a force feedback sensor, and also observe its straight movement using a CCD camera. Then, we estimate the mass of the object using known static and dynamic friction coefficients. Experimental results show that the mass of solids such as wood, iron, and ceramic objects were estimated efficiently within 10% error bound.