High-resolution CMOS video sensors are on their way to replace or complement other surround sensors in a vehicle such as ultra sonic, LIDAR and RADAR. Due to a lack of suitable automotive high-rate transmission media traditional data fusion processors had to be located close to the sensor unit or the data fusion itself was confined to the object level. The availability of a new Ethernet variant that allows transmission over unshielded twisted pair cables, as proposed by the OPEN consortium, introduces a new architecture in which raw camera pixel data with more than 1 Mega Pixel resolution can be transmitted to a central data fusion unit. We can now implement a variety of Advanced Driver Assistance Systems (ADAS) on a single highly integrated multi-core platform, e.g. Panorama View Park Assistant, Blind Spot Detection, Intersection Management or 360° Collision Warning. In this paper we describe a Multi-Camera Platform for surround sensing based on Ethernet. We show a compact multi-core data fusion computer with GPU-based video processing facilities. As an example a 3D Panorama View Park system will be described that can be used also at higher speeds.
Multi-camera surround sensing for driver assistance systems
2012
7 Seiten, 6 Bilder, 4 Quellen
Aufsatz (Konferenz)
Englisch
Multi camera surround sensing for driver assistance systems
Kraftfahrwesen | 2012
|Multi-Camera Surround Sensing for Driver Assistance Systems
British Library Conference Proceedings | 2012
|3-D Surround View for Advanced Driver Assistance Systems
IEEE | 2018
|