Object surface classificaiton based on friction properties for intelligent robotic hands

Object surface properties are among the most important information for intelligent robotic grasping and manipulation. This paper presents a new object surface classification approach based on frictional properties. The idea is to use a robotic finger to rub over an object surface with a low acceleration and identify the frictional properties using measured friction force and sliding velocity. A quasi-static LuGre model is used to characterise the relationship between friction force and sliding velocity, and the generalized Newton-Raphson method is applied to estimate unknown frictional coefficients of this model. Since the frictional coefficients of the quasi-static LuGre model are closely related to the material physical properties, object surfaces can be classified using a naïve Bayes classifier with the identified frictional coefficients. Test results show that the proposed approach can achieve a high correctness in object surface classification.