Input Command Shaping Using the Versine Function with Peak Acceleration Constraint and Numerical Optimization to Minimize Residual Vibration

Pratheek Patil, Purdue University

Abstract

Dynamic systems and robotic manipulators designed for time-optimal point-to-point motion are adversely affected by residual vibrations introduced due to the joint flexibility inherent in the system. Over the years, multiple techniques have been employed to improve the efficiency of such systems. While some techniques focus on increasing the system damping to efficiently dissipate the residual energy at the end of the move, several techniques achieve rapid repositioning by developing cleverly shaped input profiles that aim to reduce energy around the natural frequency to avoid exciting the resonant modes altogether. In this work, a numerical framework for constructing shaped inputs using a Versine basis function with peak acceleration constraint has been developed and improvements for the existing numerical framework for the Ramped Sinusoid basis function have been made to extend the range of values of the weighting function and improve the computational time. Performance metrics to evaluate the effectiveness of the numerical framework in minimizing residual vibrations have been developed. The effects of peak input acceleration and weighting function on the residual vibration in the system have been studied. The effectiveness of the method has been tested under multiple conditions in simulations and the results were validated by performing experiments on a two-link flexible joint robotic arm. The simulation and experimental results conclusively show that the inputs developed using the constrained numerical approach result in better residual vibration performance as compared to that of an unshaped input.

Degree

M.Sc.

Advisors

Meckl, Purdue University.

Subject Area

Robotics|Energy|Design|Mathematics

COinS