A Vision-Based System for Robot Localization in Large Industrial Environments

In this paper, we propose a vision-based system to localize mobile robots in large industrial environments. Our contributions rely on the use of fisheye cameras to have a large field of view and the associated algorithms. We propose several calibration methods and evaluate them with a ground-truth obtained by a motion capture system. In these experiments, we also evaluate the influence of the parameters as the number of points used for calibration, or the influence of the accuracy of these points. Our system is then experimented in a real industrial environment, where we localize an ROV (Remotely Operated underwater Vehicle) in a basin. As shown is this paper, this system can also localize wheeled mobile robots in the same way.