Localization of intelligent ground vehicles in outdoor urban environments using stereovision and GPS integration

We propose an approach for vehicle localization in dense urban environments using a stereoscopic system and a GPS sensor. The stereoscopic system is used to capture stereo video flow, to recover the 3D environments and to estimate the vehicle motion on the basis of feature detection, matching and triangulation of stereo pairs. Then reliable GPS positions are used as measurements to update the vehicle motion predicted by stereovision. The proposed method is tested on real data acquired by the SeT laboratory experimental vehicle. The obtained results show that even if the only stereovision based approach is not sufficient to provide accurate localization in long term, the proposed method can efficiently support GPS in short term. The stereovision based method effectively permits to provide the localization service during GPS outages (e.g., GPS jumps, GPS mask) due to multi-path problem or other noises.