Model approximation of an arm of the NAO™ robot using system identification

Abstract

This paper proposes a methodology to obtain a model approximation for the SoftBank NAO™ robot. It is worth to stress that the NAO™ robot is a closed commercial platform, namely, most of the parameter values are not available. Then, to improve the knowledge about dynamic behavior and to carry out further analysis, a mathematical model is necessary. The identification is achieved by basic knowledge form undergraduate control engineering courses, such as system identification, classic control, state space and robot dynamics. Model identification allow us to obtain unknown parameters of a first-order approximation of the mechanical dynamics for the NAO™ robotic arm. Additionally, a theoretical dynamic model is obtained by means of the Euler-Lagrange approach. Both viewpoints are complemented in order to compute the missing model parameters and to obtain a satisfactory approach. This demonstrates that even simple models can replicate without significant error the system dynamics.