Skip to Main Content
In this paper, we present a visual navigation algorithm by combining visual localization with the extraction of valid planar regions from a single camera of an indoor mobile robot. Only two pairs of natural line and point are used for the visual localization to take the advantage of fast detection. To track a given landmark model, Lucas-Kanade optical flow algorithm is applied by using gradient descent. We use the odometer data combined with visual information to determine the height of the landmark features. On-ground image features are used to calculate the homography between two image frames and to detect the planar region for navigation. Experimental results show the robustness of the method with respect to image illumination and noises. The performance in indoor environments shows the feasibility of the proposed visual navigation algorithm in realtime.