VisLab and the Evolution of Vision-Based UGVs,

Thanks to the reduced costs of image acquisition devices and to the increasing computational power of current computer systems, Computer Vision has recently become a very popular method to sense the surrounding environment. This work presents a challenging application of machine vision to the automatic guidance of autonomous vehicles, discusses the key problems intrinsic to this field, and describes the solutions adopted in the development of different prototype vehicles worldwide. In the second part this paper focuses on the GOLD system, a stereo vision system developed at the University of Parma, Italy, for generic obstacle detection and lane localization, able to process images in real-time. GOLD was tested on the MOB-LAB experimental land vehicle for more than 3,000~km along extra-urban roads and freeways at speeds up to 80 km/h.