Command shaping with constrained peak input acceleration to minimize residual vibration in a flexible-joint robot

Yumeng Wu, Purdue University

Abstract

Rapid point-to-point motion is limited when flexibilities exist in the system. In order to minimize the vibrations related to joint flexibilities, much work has been done, including modifying the system so that vibrations can be damped out more quickly, calculating the inverse dynamics of the system and constructing shaped input profiles that avoid system natural frequencies. In this work, the earlier fixed-time command shaping method has been extended to a peak-acceleration-constrained approach with two basis functions, the ramped sinusoid function and the versine function, such that the maximum acceleration is guaranteed without overconstraining the input profiles. The approach is developed and then validated with a two-link flexible-joint robotic arm. The effect of peak input acceleration and weighting factor on residual vibrations has been studied. A performance metric has been developed to assess residual vibrations. Input profiles with two basis functions are compared with each other, as well as the results of a bang-bang profile. All simulations and experiments have shown the effectiveness of the command shaping method with constrained peak input acceleration on residual vibration reduction. In addition, the ability to weigh the trade-off between actuation time and settling time warrants the optimization of total move time. Lastly, there exists an optimal weighting factor for each peak input acceleration to minimize the total move time and the total move time decreases with higher peak input acceleration.

Degree

M.S.M.E.

Advisors

Meckl, Purdue University.

Subject Area

Robotics

COinS