The fusion of multiple image analysis algorithms for robot road following

Abstract

Over the past five years Bristol University has been conducting research into Vision Guided Vehicle Navigation, focusing both on hardware and software. At Bristol, a parallel computer architecture, based on the Inmos transputer has been developed that allows multiple processes to be run, each with access to video images digitised at real-time. Road following algorithms have been developed for this system, although to attain operation at near frame rate, they are neither as robust nor reliable as road finding algorithms that are free of any time constraints. This paper presents a method of executing a number these algorithms concurrently, fusing their outputs together into an accurate road model.