Vision-based detection of personal rapid transit guideway

ULTra is a system of autonomous vehicles operating on a segregated guideway. The existing navigation system requires the use of raised kerbs to assess the vehicles position within the guideway. A vision-based system has the potential to be more flexible by sensing markings on the ground. It could also assist future obstacle detection systems. This paper implements three well known lane detection approaches developed for roads using images from a personal rapid transit (PRT) guideway. The Hough transform technique is found to be the least accurate, the LOIS algorithm is found to have reasonable performance and the RALPH algorithm is found to be the most accurate and robust to noise.