Use of PSO in Parameter Estimation of Robot Dynamics; Part Two: Robustness

TL;DR

This paper evaluates the robustness of a PSO-based parameter estimation method for robot dynamics, comparing it with traditional estimation techniques under noisy conditions through simulation.

Contribution

It introduces robustness enhancements to PSO-based parameter estimation and compares its performance with classical methods using simulated noisy data.

Findings

PSO-based method shows improved robustness over traditional methods.

Robustness is tested with noisy samples in simulated robot system.

Comparative analysis highlights strengths and weaknesses of each approach.

Abstract

In this paper, we analyze the robustness of the PSO-based approach to parameter estimation of robot dynamics presented in Part One. We have made attempts to make the PSO method more robust by experimenting with potential cost functions. The simulated system is a cylindrical robot; through simulation, the robot is excited, samples are taken, error is added to the samples, and the noisy samples are used for estimating the robot parameters through the presented method. Comparisons are made with the least squares, total least squares, and robust least squares methods of estimation.