Abstract
This paper presents a methodology, based on the estimation of the optical flow, to detect static obstacles during the motion of a mobile robot. The algorithm is based on a correlation scheme. At any time, we estimate the position of the focus of expansion and stabilize it by using the Kalman filter. We use the knowledge of the focus position of the flow field computed in the previous time to reduce the search space of corresponding patches and to predict the flow field in the successive one. Because of its intrinsic recursive aspect, the method can be seen as an on-off reflex which detects obstacles lying on the ground during the path of a mobile platform. No calibration procedure is required. The key aspect of the method is that we compute the optical flow only on one row of the image, that is relative to the ground plane.
Acknowledgements: this paper describes research done at the Robotic and Automation Laboratory of the Tecnopolis CSATA. Partial support is provided by the Italian PRO-ART section of PROMETHEUS.
Chapter PDF
Similar content being viewed by others
References
Little J., Bulthoff H. and Poggio T.: Parallel Optical Flow Using Local Voting. IEEE 2nd International Conference in Computer Vision, 1988
Sandini G. and Tistarelli M.: Robust Obstacle Detection Using Optical Flow. IEEE Workshop on Robust Computer Vision, 1–3 October 1990, Seattle-USA
Ancona N.: A First Step Toward a Temporal Integration of Motion Parameters. IECON'91, October 28 1991, Kobe-Japan
Enkelmann W.: Obstacle Detection by Evaluation of Optical Flow Fields from Image Sequences. First European Conference on Computer Vision, April 1990, Antibes-France
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1992 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Ancona, N. (1992). A fast obstacle detection method based on optical flow. In: Sandini, G. (eds) Computer Vision — ECCV'92. ECCV 1992. Lecture Notes in Computer Science, vol 588. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-55426-2_30
Download citation
DOI: https://doi.org/10.1007/3-540-55426-2_30
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-55426-4
Online ISBN: 978-3-540-47069-4
eBook Packages: Springer Book Archive