Landmark-based Robot Localization Using Stereo Camera Systems

This paper proposes the method of 2D localization using a stereo webcam. The navigating space of robot with landmarks for robot localization is an indoor environment. Therefore, the main task is that features of landmarks that are unchanged in scales and directions are determined from the left and right images. The algorithm of the Scale-Invariance Feature Transform (SIFT) is applied for transformation of all features of the landmark image to multi-dimensional vectors. Moreover, the pixel locations of the detected landmarks on the left and right images are transformed to distances using the geometric projections. The coordinates of each detected landmark and the distances to the robot are finally combined using Euclid distance equations for estimating the robot location. Results are shown to illustrate the effectiveness of the proposed stereo webcam system with the consuming time and the low cost, but it still satisfies the localization task.
Bạn đang xem trang mẫu tài liệu này.