Map-Building and Localization by Three-Dimensional Local Features for Ubiquitous Service Robot
In this work, we propose a semantic-map building method and localization method for ubiquitous service robot. Our semantic-map is organized by using SIFT feature-based object representation. In addition to semantic map, a vision-based relative localization is employed as a process model of extended Kalman filters, where optical flows and Levenberg-Marquardt least square minimization are incorporated to predict relative robot locations. Thus, robust map-building performances can be obtained even under poor conditions in which localization cannot be achieved by classical odometry-based map-building. To localize robot position and solve kidnap problem, we also propose simple, but fast localization method with a relatively high accuracy by incorporating our semantic-map.
KeywordsScale Invariant Feature Transform Stereo Vision Stereo Camera Robot Position Robot Localization
Unable to display preview. Download preview PDF.
- 4.Doh, N., Choset, H., Chung, W.K.: Accurate Relative Localization Using Odometry. In: Proceedings of IEEE International Conference on Robotics and Automation, vol. 2, pp. 1606–1612. IEEE Computer Society Press, Los Alamitos (2003)Google Scholar
- 5.Harris, C., Stephens, M.J.: A Combined Corner and Edge Detector. In: Alvey Vision Conference, pp. 147–152 (1988)Google Scholar
- 6.Lowe, D.G.: Distinctive Image Features from Scale-Invariant Keypoints. International Journal of Computer Vision, 91–110 (2004)Google Scholar
- 7.Lucas, B., Kanade, T.: An Interative Image Registration Technique with an Application To Stereo Vision. In: Proceedings of DARPA IU Workshop, pp. 121–130 (1981)Google Scholar
- 8.Maybeck, P.S.: The Kalman Filter: An Introduction to Concepts. Springer, Heidelberg (1990)Google Scholar