Indoor Robot localization using Adaptive Omnidirectional Vision System

The typical omnidirectional vision sensor predicts the location of mobile robots based on the information of the wall. With this method, following the location of landmarks is available and it is possible to predict where the target is in broad areas, but the location error increases according to the increasing distance from the feature point. In this paper, we suggest the new omnidirectional vision system using a convex lens to remove such a weak point. The system makes it possible to acquire the image information on a ceiling and a wall at the same time by a single camera. We realize the suggested vision system and should check whether the location accuracy is improved through the driving test.