Compact camera systems are making possible two advances in automotive safety. Engineers at Volvo have introduced the technology to monitor the blind spots near the rear of a car on either side of the vehicle, and developers at Delphi (Kokomo, IN) are preparing for rollout next year of an eye tracker to detect drowsiness in drivers.
A camera system mounted on the sideview mirror will be available on 2005 Volvo models to give drivers warning of cars in the areas that are often not visible in either the rearview or sideview mirrors (see figure). The blind-spot information system (BLIS) uses National Semiconductor (Santa Clara, CA) CMOS imaging-array cameras rather than radar. Transmitting radar systems can have electromagnetic-interference concerns, according to project engineer Bjorn Hellden. A radar system would have been more difficult for the engineers to package in the sideview mirrors, he adds. An extra radar sensor would also be needed in the rear bumper to calculate relative vehicle speeds.
The cameras, which are located at the base of each sideview mirror, stare into an area called the detection “window” or warning zone 9.5 m long by 3‑m wide. This field of view was determined by calculating the blind-spot area for a driver in the worst seating position using a flat U.S.-regulation driver’s-side mirror (European mirrors have a convex outer area).
Each camera images at a rate of 25 frames/s. Array resolution is 640 × 480 pixels with the working area of the array 320‑×‑240 pixels. A dedicated Texas Instruments (Dallas, TX) 350-MHz digital signal processor brings image-processing algorithms to bear on the image data. Hellden says the pattern-recognition algorithms look for the same pattern (including ones for motorcycles) in the warning zone in three consecutive images. The rate of change between pictures is used to determine the location and relative speed of any vehicle within the window, according to Lars Erik Lundin, vice president and general manager of Volvo’s Monitoring and Concept Center (Camarillo, CA). At night, the algorithms detect headlights rather than vehicle-shape patterns.
If a vehicle is within the detection area, the BLIS processor illuminates an arrow-shaped orange light-emitting-diode (LED) lamp on the inside window frame next to the mirror. The system software does not issue alerts for vehicles overtaking at greater than 70 km/h (43 mph), which transit the blind spot quickly. Nor is warning given if the BLIS-equipped car passes another vehicle at more than 20‑km/h (12 mph), where the driver sees the slower car quickly passing through the warning zone. In addition, the controller does not activate the system until the car is moving at speeds greater than 10 km/h (6 mph).
The eyes have it
Another vision system, developed by the Electronics and Safety operation of Delphi, points its cameras inside the vehicle. Here, the driver’s eye movements are tracked to determine various behaviors that may affect driving ability and safety.
Based on university research, eye tracking is a nonintrusive way to measure physiological states such as blood-alcohol content or intoxication, notes Richard Lind, director of advanced engineering. He points out that modern, affordable processing technologies and computer power allow the number crunching to track eye movement and determine characteristic behavior patterns. Ten years ago, such vision-based systems may have required multiple cameras, Lind notes.
The first application of the Delphi system will be to determine drowsiness of truck drivers. “We think eye tracking is a significant way to tell if a driver is drowsy by monitoring blinking, stare, gaze, and where the driver is looking,” says Lind. With auto-initialization and calibration, no driver interaction is required for the system to work. The company is reluctant to provide further system details at this time, but notes it is designed to function under all lighting conditions and the “normal range” of glasses worn by drivers.
The Delphi eye-tracking system is expected to be available as an aftermarket trucking-industry technology in the fourth quarter of 2005. While fatigue monitoring will be the first product feature, other applications will likely follow if this one is successful.
The company is also looking into vision systems to track road markings to warn the driver if the vehicle is departing its lane without the turn signal being on. But still to be resolved are issues including indistinct or confusing road markings, such as in construction areas. Performance during rain or on snow-covered roads is also a concern.