Calibration and Parameter Estimation

Abstract

The geometric and inertial parameters of a robot are of crucial importance for the development of model-based control and validation of simulation results. These parameters are key parameters in the equations of motion. Most of the time, they are parameters provided by CAD data; however, experience has shown that CAD data are only a rough approximation of the true parameters because the CAD data do not take into account cables and small equipment, in addition the robot may be subject to several modifications and enhancements with time. This chapter describes the state of the art in kinematics calibration and dynamics identification for humanoid robots. After presenting the fundamental equations and the resolution of the problem, this chapter emphasizes the practical implementations to facilitate the identification and to guarantee a good accuracy of the results. In particular, this chapter emphasizes the three key aspects to perform accurate identification: (1) modeling; (2) generating motions for identification; (3) practical implementation. Experimental examples and results are used to illustrate their importance.