A simple procedure for modeling and identication of a test bench 4-DOF manipulator

Abstract

This work proposes a characterization procedure for modeling and identification of a workbench 4-DOF manipulator. The robot is a three link planar manipulator with joints connected through 4 servomotors. A decentralized approach is employed leading to four second order discrete-time linear transfer functions. Identification procedure applies amplitude modulated pseudo binary random signal (APRBS) as excitation input due to the nonlinear behavior of the system. Sensing for dynamic variables are gyroscope and accelerometers connected through an I2C protocol to a launchpad MSP432 from Texas Instruments, assembled with an ARM M4F processor. The embedded system allows sensor fusion and input signal generation such that a gray box identification may be performed. The procedure is validated through experimental data which shows the root mean square error (RMSE) with magnitude of 10⁻⁵ and the adjusted determination coefficient R²aⱼ greater than 0.92 for each system. These are findings that show the effectiveness of the presented procedure herein.