LINEARIZED ROBOT MODELS IN JOINT AND CARTESIAN SPACES

Abstract

In this paper, linearization of the Newton-Euler formulation of robot dynamics is considered, and linearized robot models in joint and Cartesian spaces are presented. A new computational method for recursive evaluation of linearized dynamic robot models, about a nominal trajectory is given. The proposed algorithms are designed to evaluate the “sensitivity matrices” necessary to determine the linearised robot models and can be implemented numerically or symbolically. To achieve an efficient formulation for the coefficient sensitivity matrices, concepts from tensor algebra are used. A numerical implementation of the proposed linearized model shows that it has computational complexity of order N 2 .