Localization of Moving Robots By Using Omnidirectional Camera in State Space Framework

Abstract

This paper proposes a localization method of mobile robots with omnidirectional camera. An omnidirectional camera captures images of 360^◦, hence the view is invariant with respect to the direction of the robot’s sensor if the image is processed into correlation. The localization algorithm consists of two steps: regression from the correlation image into the position, and applying Kalman filter by incorporating the dynamic model of the mobile robot. The experiment shows the effectiveness of the proposed approach.