Omnidirectional Visual Servoing using the Normalized Mutual Information

Geometrical features have been a key element of visual servoing for several years. Recently, different works have shown how using all the information within the images can lead to successful servoing tasks. In particular, works using mutual information have been proposed and tested on perspective cameras. In this paper we propose to adapt this approach to vision systems following the unified sphere model for central cameras using a normalized version of the mutual information. This will permit to apply the technique to large fields of view with a more reliable similarity function. Several experiments are performed on a cartesian robot with a fisheye camera to validate our approaches.