Frequency-based Identification Routine for the Inertial Parameters of an Industrial Robot

Abstract

The frequency response of an industrial robot's (IR) speed-controlled system (SCS) provides valuable information about the inertial parameters of the respective manipulator. In addition, the frequency domain offers decisive advantages regarding parameter identification as opposed to the time domain. This paper presents an identification routine for the inertial parameters of a 6-degree-of-freedom (DOF) IR based on the frequency response measurements of the SCS and a genetic algorithm (GA) conditioning approach. These measurements are less susceptible to noise and offer the possibility to incorporate friction effects using phase information. In addition, the presented identification model is independent of the joint speed and acceleration, allowing for the optimization of the observation matrix exclusively through the pose. The GA is used to define an optimal set of poses while considering possible collisions at the same time.