Identification of robot dynamic model and joint frictions using a baseplate force sensor

Dynamic models play an important role in robot control and applications. The accurate identification of dynamic models has become crucial to meeting increasing performance requirements. Owing to the inertial forces and the joint frictions coupling, the identification first requires a parametrized friction model. However, the joint frictions are strongly nonlinear and vary with many factors including posture, velocity and temperature. Hence, all friction models have some deviation from the real values, which reduces the identification accuracy. This paper proposes an identification approach using a baseplate force sensor. It identifies the inertial parameters first and then computes the joint friction values by subtracting the inertial torques from the joint torques. This method has the advantage that it does not require a priori friction model. It can choose or construct a proper model to fit the real values and is thus expected to achieve high performance. Experiments on a 6-DoF robot were conducted to verify the proposed method.