Robot dynamic model: freudenstein-based optimal trajectory and parameter identification

Abstract

A thorough understanding of an industrial robot’s dynamic model is critical for practical robotic applications. An effective dynamic model is required for optimal controller design and trajectory planning. Robot manufacturers only provide kinematic data, which can only guarantee a certain level of positioner accuracy. The design of the trajectory-planning scheme, on the other hand, necessitates a thorough understanding of its dynamic features. The identification of dynamic parameters involves several procedures. This study used the Euler-Lagrangian equation to derive the robot dynamic model in its canonical form. To excite each link of the irb1600 robot industrial robot while avoiding displacement, velocity, and acceleration discontinuities at the start and endpoints, a Freudenstein 1-3-5 trajectory based on Fourier series expansion was used. The dynamic parameters were determined using the nonlinear least-squares approach based on the Levenberg-Marquart equation. The Savitzky-Golay smoothing filters improved the identification method by decreasing system noise.