Robust Dense Visual Odometry For RGB-D Cameras In A Dynamic Environment

Abdallah Dib (1, 2) , François Charpillet (2, 1)

1 LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications
2 LARSEN - Lifelong Autonomy and interaction skills for Robots in a Sensing ENvironment

—The aim of our work is to estimate the camera motion from RGB-D images in a dynamic scene. Most of the existing methods have a poor localization performance in such environments, which makes them inapplicable in real world conditions. In this paper, we propose a new dense visual odometry method that uses RANSAC to cope with dynamic scenes. We show the efficiency and robustness of the proposed method on a large set of experiments in challenging situations and from publicly available benchmark dataset. Additionally, we compare our approach to another state-of-art method based on M-estimator that is used to deal with dynamic scenes. Our method gives similar results on benchmark sequences and better results on our own dataset.

Mots clés: vision; visual odometry; optical flow; camera; rgb-d; slam; computer vision; localization

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]; Traitement des images [eess.IV]; Robotique [cs.RO]; Intelligence artificielle [cs.AI]

Abdallah Dib, François Charpillet. Robust Dense Visual Odometry For RGB-D Cameras In A Dynamic Environment. International Conference on Advanced Robotics ICAR 2015, Jul 2015, Istanbul, Turkey. ⟨hal-01212043⟩