English

Robust Real-time RGB-D Visual Odometry in Dynamic Environments via Rigid Motion Model

Robotics 2019-07-22 v1 Computer Vision and Pattern Recognition

Abstract

In the paper, we propose a robust real-time visual odometry in dynamic environments via rigid-motion model updated by scene flow. The proposed algorithm consists of spatial motion segmentation and temporal motion tracking. The spatial segmentation first generates several motion hypotheses by using a grid-based scene flow and clusters the extracted motion hypotheses, separating objects that move independently of one another. Further, we use a dual-mode motion model to consistently distinguish between the static and dynamic parts in the temporal motion tracking stage. Finally, the proposed algorithm estimates the pose of a camera by taking advantage of the region classified as static parts. In order to evaluate the performance of visual odometry under the existence of dynamic rigid objects, we use self-collected dataset containing RGB-D images and motion capture data for ground-truth. We compare our algorithm with state-of-the-art visual odometry algorithms. The validation results suggest that the proposed algorithm can estimate the pose of a camera robustly and accurately in dynamic environments.

Keywords

Cite

@article{arxiv.1907.08388,
  title  = {Robust Real-time RGB-D Visual Odometry in Dynamic Environments via Rigid Motion Model},
  author = {Sangil Lee and Clark Youngdong Son and H. Jin Kim},
  journal= {arXiv preprint arXiv:1907.08388},
  year   = {2019}
}
R2 v1 2026-06-23T10:25:01.205Z