Remote vehicle manoeuvring using augmented reality

In this paper we present an application of augmented reality aimed for remote vehicle navigation and manoeuvring. The concept is used to aid the operator in manoeuvring a vehicle in the case of temporary loss of vision or poor visibility conditions, relying entirely on the graphics. The approach is based on the continuous acquisition of vehicle position and orientation in the world reference frame, combined with a near-scene reconstruction obtained from a laser-scan, in real-time. An algorithm to register real imagery from an on-board stereo camera system with a generated graphical scene representation has previously been developed and implemented. A successful integration of graphical and sensory information in real-time was achieved combining a digital elevation model of the terrain with accurate registration techniques. A fully functional demonstrator has been developed for a one-dimensional test-case, while the concept is derived for an arbitrary motion pattern over a flat surface.