Monocular Omnidirectional Vision based Robot Localisation and Mapping

Robot localisation is an important task in mobile robotics. In this paper a 2D simultaneous localisation and map- ping implementation that uses only monocular omnidirectional vision is presented. Experiments are performed using a novel hybrid-simulation method to evaluate the SLAM algorithm based on the sensor model and the layout of the environment. Tests are then carried out that show the advantage gained by using omnidirectional vision. The application of this work to a real robot is discussed.