Affordable Access

Access to the full text

Virtual Training for a Real Application: Accurate Object-Robot Relative Localization Without Calibration

Authors
  • Loing, Vianney1
  • Marlet, Renaud2
  • Aubry, Mathieu2
  • 1 Ecole des Ponts ParisTech, UPE, Navier (UMR 8205), Champs-sur-Marne, France , Champs-sur-Marne (France)
  • 2 Ecole des Ponts ParisTech, UPE, LIGM (UMR 8049), Champs-sur-Marne, France , Champs-sur-Marne (France)
Type
Published Article
Journal
International Journal of Computer Vision
Publisher
Springer-Verlag
Publication Date
Jun 21, 2018
Volume
126
Issue
9
Pages
1045–1060
Identifiers
DOI: 10.1007/s11263-018-1102-6
Source
Springer Nature
Keywords
License
Yellow

Abstract

Localizing an object accurately with respect to a robot is a key step for autonomous robotic manipulation. In this work, we propose to tackle this task knowing only 3D models of the robot and object in the particular case where the scene is viewed from uncalibrated cameras—a situation which would be typical in an uncontrolled environment, e.g., on a construction site. We demonstrate that this localization can be performed very accurately, with millimetric errors, without using a single real image for training, a strong advantage since acquiring representative training data is a long and expensive process. Our approach relies on a classification Convolutional Neural Network trained using hundreds of thousands of synthetically rendered scenes with randomized parameters. To evaluate our approach quantitatively and make it comparable to alternative approaches, we build a new rich dataset of real robot images with accurately localized blocks.

Report this publication

Statistics

Seen <100 times