Numerical Optimized Command Shaping with Constrained Peak Input Acceleration for Minimizing Residual Vibration

Bowen Wu, Purdue University

Abstract

The fast point-to-point motion of a robotic system can be greatly affected by its joint flexibilities, which introduce vibrations during and after motion. While numerous techniques were developed to minimize the residual vibration of the flexible robotic system, this work studies the command shaping method. This method constructs an input profile that aims to reduce excitation energy around the natural frequencies to minimize residual vibration. This work modifies the command shaping method by applying a numerical optimization technique with a constraint on peak input acceleration. A framework has been constructed for the combination of these two modifications, which proved their benefits compared to previous research. This method is then implemented in simulation and experimentally validated a two-link flexible-joint robotic arm with ramped sinusoid basis function. The performance metrics have been defined to evaluate the level of residual vibrations for result analysis. The influence of weighting factor and peak input acceleration has been investigated for the command shaping input profile. Lastly, the results from both simulations and experiments show better performance at minimizing residual vibrations from the numerical approach than from the previous analytical approach.

Degree

M.S.

Advisors

Meckl, Purdue University.

Subject Area

Robotics

COinS