Effective Vision Algorithms for Detection of Structured and Unstructured Roads

For an automatic co-pilot designed to assist the driver while operating a vehicle on wellmarked or unmarked roads algorithms were developed for recognizing and tracking the driving lane used by the vehicle. A video camera is used as a sensor. Each algorithm has been implemented on one parallel processor of the multiprocessor robot vision system BVV 3. The method of controlled correlation is used for a quick and robust feature extraction for well-marked roads. To further increase its robustness the program utilizes a 2D model representing knowledge of the appearance of roads and lane markers in the image of the camera. The algorithm has been tested successfully in simulation and in realworld experiments while driving on different types of roads at high speeds. A far lookahead distance of 100 m has been achieved, adequate for driving at high speed. For the more complex environment of an unstructured road an algorithm, based on the information of texture, is necessary. Depending on the texture transition it describes a pathway model for autonomous vehicles to operate on unstructured roads.