Image processing for robot road following

The goal of the project at the University of Bristol is to investigate algorithms and transputer based architectures for real-time road following along initially simple, uncluttered roads. A trials vehicle has been prepared to test some of the road following algorithms. A number of computer vision algorithms are described that identify the road in an image sequence by means of tracking the major road edges and finding the road surface. Initial work on identifying the presence and location of potential obstacles in the path of a mobile robot is outlined. The results of the first outdoor trials of a small tracked vehicle are presented. The algorithms described have been coded in OCCAM to run on transputer hardware designed and built at Bristol University. The system comprises a video interface board that provides digitised video data from camera or VCR to a number of framestores. These are accessible by a number of 20 MHz T414 32-bit transputers via a VME-like bus. An IBM compatible PC with plug-in B004 is used for program development. A number of sequences of road images have been collected from a video camera mounted on the roof of a car travelling at an average speed of 20 mph around public roads in the Bristol area