A group of German researchers under the guidance of a lecturer at the University of Alcalá, UAH, Spain has developed a system that employs artificial vision to identify pedestrians walking ahead of vehicle.
Real example where the system detects a pedestrian (in red) within the detection area (in magenta). Image: CG Keller et al.
This innovative technology will shortly be introduced in the sophisticated Mercedes vehicles. The device consists of two cameras, along with a unit that is capable of real time processing of the information provided by all image points.
The installation of dense stereo system is the latest feature of the device. The term 'stereo' implies the inclusion of two cameras, which are arranged 30 cm apart in a structure beneath the rear-view mirror. Accurate real time recognition of the surroundings in front of the vehicle is facilitated by the dense stereo vision. Detection is done based on the variations in contact between the vehicle and the road, the presence of pot holes, and the shape of the road. Furthermore, the system allows instant detection of animate objects in less than 200 ms, for example, a child suddenly crossing the road.
The two cameras are linked to a processing unit, which implements the artificial vision algorithm. The hardware is based on field programmable gate array (FPGA) technology.
Images can be visualized through projections on the windscreen's inner side or on a front transparent screen. A variety of response elements such as alarms, break activation systems, and wheel-control systems have also been incorporated in order to warn and assist the driver, thereby improving the control of vehicle.
Single camera systems already exist in Class C Mercedes vehicles in order to detect road signs. Toyota anticipates launching the pedestrian detection stereo systems shortly.