Robust and Accurate Visual Odometry by Stereo

This work deals with the problem of estimating the trajectory of an autonomous rover by onboard passive stereo vision only (so-called Visual Odometry). The proposed method relies on bundle adjustment of tracked SURF feature points. Tests on real-world data show the effectiveness of this approach. In the case of cyclic trajectories, even greater accuracy is attained by periodically comparing the scene against a set of stored references.