Radar systems have been chosen by industry in the 1990’s to introduce “Adaptive Cruise Control” (ACC) into the car market; these systems were intended to work above a certain lower cruise speed on highways keeping a speed-dependent distance to the vehicle ahead.
- They cannot detect roads and lane markings in the general case,
- their lateral resolution is poor, and
- they may generate false alarms from multi-path reflections.
Therefore, combinations of these types of systems with vision have been investigated in the late 1990’s. Together with an industrial partner, the combination of radar and visual interpretation sketched in the figure (at left) has been investigated with the test vehicle VaMP.
- All hypotheses generated from radar are checked by vision; those finding no support from vision are deleted.
- Vision adds information on lanes and on the width of the vehicle observed, and
- it perceives both horizontal and vertical curvature of the road (the latter one is coded in the yellow bars moving vertically).
Hofmann U, Rieder A, Dickmanns, ED (2000). EMS-Vision: Application to Hybrid Adaptive Cruise Control. Proc. Int. Symp. on Intelligent Vehicles (IV’2000), Dearborn, MI. pdf
Rieder A (2000). Fahrzeuge sehen – Multisensorielle Fahrzeugerkennung in einem verteilten Rechnersystem für autonome Fahrzeuge. Dissertation, UniBwM / LRT. Kurzfassg
Hofmann U, Rieder A, Dickmanns ED (2001). Radar and Vision Data Fusion for Hybrid Adaptive Cruise Control on Highways. Proc. ICVS, Vancouver, July
Hofmann U, Rieder A, Dickmanns ED (2003). Radar and Vision Data Fusion for Hybrid Adaptive Cruise Control on Highways. Int. J. ‘Machine Vision and Applications’, Vol. 14(1), Springer-Verlag, pp. 42-49 Abstract
Hofmann U (2004). Zur visuellen Umfeldwahrnehmung autonomer Fahrzeuge. Dissertation, UniBwM / LRT, Kurzfassg
Dickmanns ED (2007). Dynamic Vision for Perception and Control of Motion. Springer-Verlag, (Section 14.6.3) Content