Autonomous Flight with Robust Visual Odometry under Dynamic Lighting Conditions
- Authors
- Kim, Pyojin; Lee, Hyeonbeom; Kim, H. Jin
- Issue Date
- Aug-2019
- Publisher
- SPRINGER
- Citation
- AUTONOMOUS ROBOTS, v.43, no.6, pp 1605 - 1622
- Pages
- 18
- Journal Title
- AUTONOMOUS ROBOTS
- Volume
- 43
- Number
- 6
- Start Page
- 1605
- End Page
- 1622
- URI
- https://scholarworks.sookmyung.ac.kr/handle/2020.sw.sookmyung/146887
- DOI
- 10.1007/s10514-018-9816-4
- ISSN
- 0929-5593
- Abstract
- Sensitivity to light conditions poses a challenge when utilizing visual odometry (VO) for autonomous navigation of small aerial vehicles in various applications. We present an illumination-robust direct visual odometry for a stable autonomous flight of an aerial robot under unpredictable light condition. The proposed stereo VO achieves robustness with respect to the light-changing environment by employing the patch-based affine illumination model to compensate abrupt, irregular illumination changes during direct motion estimation. We furthermore incorporate a motion prior from feature-based stereo visual odometry in the optimization, resulting in higher accuracy and more stable motion estimate. Thorough analyses of convergence rate and linearity index for the feature-based and direct VO methods support the effectiveness of the usage of the motion prior knowledge. We extensively evaluate the proposed algorithm on synthetic and real micro aerial vehicle datasets with ground-truth. Autonomous flight experiments with an aerial robot show that the proposed method successfully estimates 6-DoF pose under significant illumination changes.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - 공과대학 > 기계시스템학부 > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.