Identification of Friction Coefficient for Constrained Robotic Tasks

Correct modeling of friction forces during constrained robotic operations is critical to high-fidelity contact dynamics simulation. Such simulations are particularly important for the development, mission planning and operations analysis of space robotic systems. Most existing friction models employ the coefficient of friction to capture the relationship between the friction force and the normal load. Hence, accurate identification of this parameter is prerequisite to accurate simulation. This issue is particularly important for space robotic operations since friction characteristics of materials are very different in space. In this manuscript, the problem of identification of the coefficient of friction is investigated experimentally and numerically. The motivating application being space manipulator systems, our principal objective is to develop a practical off-line identification algorithm, requiring minimum number of measurements from sensors available on space robots. To this end, a strategy is proposed to determine the coefficient of friction by using only the measured end-effector forces. The key idea behind the method is that during one-point contact, these forces represent the contact force and hence, can be directly used to calculate the coefficient of friction. The proposed approach is tested with the experimental data from peg insertion experiments conducted on a planar robotics test-bed with a specially designed contact interface. The algorithm is generalized to arbitrary complex geometries and applied to identify the coefficient of friction for a simulated battery drop test.