Identification of Robot Dynamics: A Parallel Processing Approach

Knowledge of the exact characteristics of robot manipulators is one of the most significant factors in designing motion control systems, since the control performance is directly dependant upon the accuracy of the dynamic model. Dynamic models normally have complicated behaviour, including varying inertia depending upon the arm configuration, uncertain load effects, non-linear effects such as the Coriolis and Centripetal forces and interactions among joints. Unless these characteristics are included in the manipulator dynamics exactly, the performance of the controller is not expected to meet the given requirements. This necessitates the development of an efficient method to identify the dynamic parameters of robot arms. This paper will describe the on-line estimation of the link inertial parameters using a semi-customised symbolic representation of the dynamic equations based on the Lagrangian formulation........