Learning Hand/Eye Coordination by an Active Observer. Part 1: Organizing Centers.

Abstract

This report, the first of a three-part series, presents preliminary results in a study on the role of the active observer in the hand/eye coordination problem. It was shown in 11 that the hand/eye coordination problem can be represented, for a given pose of the observer, by the singularities of a surface, the PCS. Small changes in the pose of the observer generally produce smooth deformations of the PCS. There are configurations, however, from which arbitrarily small modifications of the point of view result in profound changes in the topological nature of the PCS. This paper studies these bifurcation configurations and, furthermore, investigates the possibility of determining a priori displacements of the observer that can achieve a desired effect on the PCS such as simplifying its topology or reducing the number of singularities separating the current configuration from a goal to be reached. The result of this analysis takes the form of the "family portrait" of all possible aspects of the PCS, indexed by the geometry of the manipulator and the pose of the observer relative to it. A hand/eye system is then completely coordinated-has "learned its PCS"-when a "portrait" has been matched with the experimental data gathered by the low-level controller 3.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1994
Accession Number
ADA289502

Entities

People

  • Jean-yves Herve

Organizations

  • University of Maryland

Tags

Communities of Interest

  • C4I

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Collision Avoidance
  • Computational Science
  • Computer Vision
  • Computers
  • Coordinate Systems
  • Displacement
  • Experimental Data
  • Geometry
  • Identification
  • Learning
  • Motion Planning
  • Observers
  • Robots
  • Topology
  • Two Dimensional
  • Visual Servoing

Readers

  • Computer Vision.
  • Systems Analysis and Design
  • Vision Science/Vision Psychology/Cognitive Neuroscience.