Environmental impedance estimation and imitation in haptics by sliding mode neural networks

Baris Yalcin, Kouhei Ohnishi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

Due to the future perspective to reproduce highly nonlinear characteristics of the contacted environment exactly in the absence of environment, especially in haptics research, and also due to providing high robustness and stability of robot control systems during environmental contacts, ensuring precision in environmental impedance estimations and storing environmental impedances are imperative studies. In this paper impedance is considered as a nonlinear mapping from position and velocity to force. This paper utilizes a sliding mode control theory based neural network, which is proposed to be used as a fast and fussy online environmental impedance & stiffness estimator and imitator by relating position and velocity dimension to force dimension. In the end, validity of online impedance estimation method and how a neural network can turn to be the model of contacted environment (imitation) are going to be shown by the experimental results. As a future perspective, continuation of this research is going to result in exact environmental impedance reproduction.

Original languageEnglish
Title of host publicationIECON 2006 - 32nd Annual Conference on IEEE Industrial Electronics
Pages4014-4019
Number of pages6
DOIs
Publication statusPublished - 2006 Dec 1
EventIECON 2006 - 32nd Annual Conference on IEEE Industrial Electronics - Paris, France
Duration: 2006 Nov 62006 Nov 10

Publication series

NameIECON Proceedings (Industrial Electronics Conference)

Other

OtherIECON 2006 - 32nd Annual Conference on IEEE Industrial Electronics
CountryFrance
CityParis
Period06/11/606/11/10

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Environmental impedance estimation and imitation in haptics by sliding mode neural networks'. Together they form a unique fingerprint.

Cite this