Conference Papers Year : 1995

Multi-Layer Perceptrons for Task Visual Servoing in Robotics

Abstract

A six-degree-of-freedom robot with a CCD-camera at the extremity of its end-effector, is associated to a closed loop control system. An object moves freely in the 3D space (translation + rotation). The aim is to servo the camera on the object, so that the image of the object on the camera screen is always the "reference image", defined by an arbitrary reference position of the object versus the camera. Classical mechanics equations are first studied, in order to determine the significant parameters of the problem. Whereupon, a control system based on the use of a MultiLayer Perceptron (MLP) is built. Results and comparisons detailing the behavior of the system in different cases (uniform and non-uniform translations and rotations of the object) are presented, and show the interest of the approach vs classical methods.
Fichier principal
Vignette du fichier
1995_10_funchal_robot.pdf (357.52 Ko) Télécharger le fichier
Origin Files produced by the author(s)
Licence
Copyright

Dates and versions

hal-03222616 , version 1 (01-04-2023)

Licence

Copyright

Identifiers

  • HAL Id : hal-03222616 , version 1

Cite

Nadine Rondel, Gilles Burel. Multi-Layer Perceptrons for Task Visual Servoing in Robotics. International Workshop on Fuzzy Logic and Neural Networks in Engineering, co-located with EPIA’95, Oct 1995, Funchal, Madeira Island, Portugal. ⟨hal-03222616⟩

Collections

UNIV-BREST CNRS
29 View
14 Download

Share

More