Computer vision algorithms and architectures for an autonomous guided road vehicle

The goal of the work described has been to develop a parallel computer architecture, based on the Inmos transputer, which is capable of real-time analysis of video images obtained from a camera mounted on the front of a vehicle. The algorithms investigated are parallelised to a sufficient degree to allow operation at near frame rate (10-20 frames per second). The two principal approaches adopted are: road edge detection and surface segmentation. Both techniques are capable of navigating the vehicle for some distance along a variety of roads but each is liable to failure in different circumstances. Methods to fuse information from the various algorithms to produce a more reliable and robust system are being considered