Friction compensation for control of drawing an object by a dual-arm robot using fuzzy-neuro

Kazuo Kiguchi, Toshio Fukuda

Research output: Contribution to journalArticlepeer-review

Abstract

Manipulating an object is one of the most important tasks of robots. Numerous studies have been carried out on object manipulation algorithms and optimizing control force. When robots try to manipulate an object on some environments, in other words, when robots try to draw an object, they must take the effect of friction into account in order to control the object smoothly. In this paper, a friction compensation method is proposed for control of drawing an object using a specialized neuron which is connected to a fuzzy neural network. The specialized neuron learns the friction effect only when the object does not move against applied force. The fuzzy neural network is used for object trajectory control. The controller consists of a main fuzzy neural network controller which includes a neuron for friction compensation for object trajectory control and sub-fuzzy neural network controllers for manipulator force control to the object. Furthermore, modified Delta-Bar-Delta learning rate adaptation method is introduced for the main neural network controller. Computer simulation was performed to evaluate these proposed methods.

Original languageEnglish
Pages (from-to)997-1004
Number of pages8
JournalNippon Kikai Gakkai Ronbunshu, C Hen/Transactions of the Japan Society of Mechanical Engineers, Part C
Volume62
Issue number595
DOIs
Publication statusPublished - 1996
Externally publishedYes

All Science Journal Classification (ASJC) codes

  • Mechanics of Materials
  • Mechanical Engineering
  • Industrial and Manufacturing Engineering

Fingerprint Dive into the research topics of 'Friction compensation for control of drawing an object by a dual-arm robot using fuzzy-neuro'. Together they form a unique fingerprint.

Cite this