Comparison of omnidirectional and conventional monocular systems for visual SLAM

Publication Type: Conference Paper
Publication: Workshop on Omnidirectional Vision, Camera Networks and Non-classical Cameras (OMNIVIS) (Best paper award) (2010)

The SLAM (Simultaneous Localization and Mapping) problem is one of the essentials challenges for the current robotics. Our main objective in this work is to perform a comparison between a visual SLAM system using monocular omnidirectional and conventional vision. Our approach is based on the Extended Kalman Filter (EKF). We use the Spherical Camera Model to obtain geometric information from the images wich is valid for both types of cameras. To integrate this model in the EKF-based SLAM we linearize the direct and the inverse projections. The approach also uses the inverse depth parameterization of the measurements and SIFT points as measurement image points. We perform experiments with images of both systems corresponding to the same challenging trajectories. This experimentation confirms that the omnidirectional system gives a much better tra- jectory and orientation estimation than the conventional vision system.