Automatic State Space-Based Dynamic Characterization of Industrial Machining Robots

Abstract

Robotic manipulators play a crucial role in precision machining, and understanding their dynamic behavior is essential for optimizing performance. This study presents the dynamic characterization of a six-degree-of-freedom serial robotic manipulator for precision machining operations. A simplified dynamic model is proposed using Frequency Response Functions (FRFs), pole-zero estimation, and regression analysis. The robot's dynamic behavior is represented through the mechanical admittance matrix, linearized by transfer functions dependent on joint states. Experimental data were collected through automated torque impacts within a defined Region of Interest and analyzed using Fast Fourier Transforms. The resulting FRFs guided the estimation of transfer functions, optimizing model complexity to balance accuracy and computational efficiency. Regression analysis indicated that cubic models with interaction terms best capture the system's dynamics. The identified models closely approximate experimental data, though minor discrepancies suggest potential areas for refinement. This methodology has may enable improvements in the precision and dynamic accuracy of robotic manipulators in machining tasks, as well as broader applications in automated manufacturing.