Improving Stereo Vision Odometry for Mobile Robots in Outdoor Dynamic Environments

Dan Pojar, Sergiu Nedevschi

2012

Abstract

This article presents a method for localization able to provide the pose in 3D using stereo vision. The method offers a better and inexpensive alternative to classical localization methods such as wheel odometry or GPS. Using only a calibrated stereo camera, the method integrates both optical flow based motion computation and SURF features detector for stereo reconstruction and motion computation. Robustness is obtained by finding correspondences using both feature descriptors and RANSAC inlier selection for the reconstructed points. Least squares optimization is used to obtain the final computed motion. World scale pose estimation is obtained by computing successive motion vectors characterized through their orientation and magnitude. The method involves fast algorithms capable to function at real time frequency. We present results supporting global consistency, localization performance and speed as well as the robustness of the approach by testing it in unmodified, real life, very crowded outdoor dynamic environments.

References

  1. Nister, D., Naroditsky, O., Bergen, J.,” Visual odometry for ground vehicle applications”, J. Field Robotics 23, 2006
  2. Howard, A., “Real-time Stereo Visual Odometry for Autonomous Ground Vehicles”, proc. Internation Conference on Robots and Systems (IROS), Sep. 2008
  3. Agrawal, M., and Konolige, K., “Rough Terrain Visual Odometry” In Proc. International Conference on Adranved Robotics (ICAR), Aug. 2007
  4. Hartley, R., Zisserman, A., “Multiple View Geometry in Computer Vision” Second Edition, C. U. Press, Ed. Cambridge, 2003.
  5. Shi, J., Tomasi, C., “Good Features to Track”, IEEE conference on Computer Vision and Pattern Recognition (CVPR94), 1994.
  6. Bradski, G., Kaehler, A., “Learning OpenCV: Computer Vision with OpenCV”, O'Reilly, 2008.
  7. Pojar, D., P. Jeong, Nedevschi S., "Improving localization accuracy based on Lightweight Visual Odometry", Intelligent Transportation Systems (ITSC), 2010, p. 641 - 646
  8. Scaramuzza, D., Fraundorfer, F., and Siegwart, R., “RealTime Monocular Visual Odometry for On-Road Vehicles with 1-Point RANSAC”, IEEE International Conference on Robotics and Automation (ICRA 2009), 2009
  9. Bay, H., Ess, A., Tuytelaars, T., Van Gool, L., "SURF: Speeded Up Robust Features", Computer Vision and Image Understanding (CVIU), Vol. 110, No. 3, pp. 346--359, 2008
Download


Paper Citation


in Harvard Style

Pojar D. and Nedevschi S. (2012). Improving Stereo Vision Odometry for Mobile Robots in Outdoor Dynamic Environments . In Proceedings of the 9th International Conference on Informatics in Control, Automation and Robotics - Volume 2: ICINCO, ISBN 978-989-8565-22-8, pages 476-480. DOI: 10.5220/0004043404760480


in Bibtex Style

@conference{icinco12,
author={Dan Pojar and Sergiu Nedevschi},
title={Improving Stereo Vision Odometry for Mobile Robots in Outdoor Dynamic Environments},
booktitle={Proceedings of the 9th International Conference on Informatics in Control, Automation and Robotics - Volume 2: ICINCO,},
year={2012},
pages={476-480},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0004043404760480},
isbn={978-989-8565-22-8},
}


in EndNote Style

TY - CONF
JO - Proceedings of the 9th International Conference on Informatics in Control, Automation and Robotics - Volume 2: ICINCO,
TI - Improving Stereo Vision Odometry for Mobile Robots in Outdoor Dynamic Environments
SN - 978-989-8565-22-8
AU - Pojar D.
AU - Nedevschi S.
PY - 2012
SP - 476
EP - 480
DO - 10.5220/0004043404760480