Visual Servoing of a Mobile Robot in Presence of Tilt Disturbances Using a Central Catadioptric Vision System

Catadioptric vision systems have been studied in the realm of visual servoing of mobile and articulated robots for the past decade. They are considered superior to a single camera system because of their larger field of view and higher resolution. This paper presents a new calibration method to estimate the tilt of a mobile robot equipped with a central catadioptric vision system moving on rough terrains. The proposed approach is based on the spatial projection of a set of parallel lines onto the image plane of an orthographic camera linked with a parabolic mirror. A methodology was developed to estimate 3-axis orientation of the catadioptric camera mounted on a mobile robot through information obtained from the 2D image. Furthermore, a visual servoing strategy was developed to drive a mobile robot from a given configuration to a target configuration while accounting for tilt disturbances imposed by the terrain. The performance of the proposed method is evaluated via computer simulations.