Robust object tracking of irregular terrain vehicle

Tracking a target robustly by utilizing vision is very difficult for a mobile robot running over the irregular terrain of a natural environment, due to the image deformation caused by the rolling and pitching of the camera, as well as due to the relative movement between the target and the camera. One approach for coping with such problems is matching the target image with many affine-transformed candidate images while tracking. However, when the number of candidate images increases significantly, such an approach is not practical to real-time tasks due to the high computational cost involved. In the present paper, we propose a new system called Robustness Analysis for Tracking (RAT) that improves tracking ability. RAT is an analysis based on the features of the object image, in which three parameters: 'Detectability', 'Robustness for Depth (RBD)', and 'Robustness for Rotation (RBR)' are defined. Many more robust templates can be found by analyzing the object image using RAT before the tracking task is performed. The experimental results are shown in order to verify the effectiveness of this method.