A New Computational Method for Linearized Dynamic Models for Robot Manipulators

Abstract

In this paper, a new computational method to derive linear ized dynamic models about a nominal trajectory for robot manipulators is developed from the "straightforward" La grangian formulation. This method is very simple and sys tematic. It can be applied to compute the feedforward control law (i.e., the generalized forces and torques) about the de sired nominal trajectory, and to design the feedback con troller that reduces or eliminates any deviations from the de sired nominal trajectory. The salient advantage of using this method is the amount of computation for deriving the com plete linearized dynamic model for a manipulator is small enough to make real-time computation on a mini- or micro computer possible. The whole computation for a manipulator with six degrees of freedom requires at most about 2427 multiplications and 1798 additions. Camputationally, this is the fastest of all the existing methods.