Identification of Flexible Joint Robot Inertia Matrix Using Frequency Response Analysis

This paper presents a novel, nonlinearity robust identification method for deriving the inertia matrix of multi- DOF Flexible Joint Robots (FJR), utilizing resonance and anti- resonance frequencies in the Frequency Response Functions (FRF). Our proposed method overcomes the limitations of conventional approaches, which are susceptible to mechanical nonlinearities, leading to inaccurate models. By leveraging fre- quency domain techniques, our approach effectively mitigates the influence of nonlinear characteristics, providing a more accurate and reliable means of robot control. Furturmore, the paper highlights the benefits of frequency domain system identification, including nonlinear robustness and the ability to decompose the flexible joint into motor and load components. Finally, a novel sequential excitation algorithm is proposed to obtain the inertia matrix of a multi-DOF robot manipulator without relying on complex theories or optimizations. The effectiveness of the proposed algorithm is verified through simulation and experiment.