Looking back at the history of machine vision for road vehicle guidance, the experience gained and progress achieved with increasing microprocessor power together with the new approaches discussed above, allow predicting a bright future for the sense of vision in vehicles. The advantages to be gained with respect to safety are so large that in the not too far future the sense of vision will be similarly wide spread in vehicles as it is among animals right now. However, the time scale under discussion is decades and not years if performance levels of humans are taken as yardstick. Nonetheless, simple systems for driver assistance will keep creeping in over the next years. After 'lane-departure warning' and support for improvements in convoy driving (radar plus vision), automatic 'Stop & Go'-assistance at lower speeds may be the next. Superiority of machine vision will first show up where boring tasks tend to annoy the human driver. 'Lane change'- assistants already are rather complex since they have to check the traffic situation in the entire lateral hemisphere of change direction. In unidirectional traffic (on high-speed roads) fast vehicles approaching from behind have to be recognized early (at large distance). In bi-directional traffic, oncoming vehicles have large relative speed since the magnitudes of the speed of both vehides add up. In addition, in order to be of real help, these assistants have to observe the traffic signs regulating lane change and passing. Driving on country roads, with turnoffs at junctions and many different types of traffic participants, e.g. kinds of animals and humans bicycling or carrying strange loads, is still much more difficult than driving on freeways. It is, however, most likely that machine vision will develop to a level where all of this can be achieved in a safe way. Most of the corresponding knowledge bases still have to be developed. Explicit representations of perceptual and behavioral capabilities have just been started. This is considered to be a precondition for autonomous adaptability and learning capabilities without which human-like performance levels will hardly be achievable. Flexible human machine interaction, its reliable realization and easy as well as pleasant partnership between two or more cognitive agents will not be possible without these features. There is plenty of room for future developments and improvements over decades to come. Spatio-temporal models of objects and subjects, including their wide variety of perceptual and behavioral capabilities, are at the core oft he knowledge bases required.


    Access

    Access via TIB

    Check availability in my library


    Export, share and cite



    Title :

    Future perceptual capabilities of vehicles with a sense of vision


    Additional title:

    Zukünftige Wahrnehmungsfähigkeiten sehender Fahrzeuge


    Contributors:


    Publication date :

    2002


    Size :

    12 Seiten, 7 Bilder, 1 Tabelle, 14 Quellen


    Type of media :

    Conference paper


    Type of material :

    Print


    Language :

    English




    Future perceptual capabilities of vehicles with a sense of vision

    Dickmanns,E.D. / Univ.d.Bundeswehr Muenchen,DE | Automotive engineering | 2002


    EMS-Vision: a perceptual system for autonomous vehicles

    Gregor, R. / Lutzeler, M. / Pellkofer, M. et al. | IEEE | 2000


    EMS-Vision: a perceptual system for autonomous vehicles

    Gregor, R. / Lutzeler, M. / Pellkofer, M. et al. | IEEE | 2002


    EMS-Vision: A Perceptual System for Autonomous Vehicles

    Gregor, R. / Lutzeler, M. / Pellkofer, M. et al. | British Library Online Contents | 2002


    EMS-Vision: A Perceptual System for Autonomous Vehicles

    Gregor, R. / Lutzeler, M. / Pellkofer, M. et al. | British Library Conference Proceedings | 2000