Although “green” vehicles are attracting a lot of attention these days because of a growing interest in environmental issues, safety systems are even more important. Vehicles, often weighing more than 1 ton and operating at high speeds, depend on safety features to prevent accidents and decrease damage. Toward that end, Japanese car makers have developed new sensor technologies to monitor both the interior and the exterior of the cars that you drive.

With safety in mind, Japanese engineers have developed systems that use sensors to imitate the function of the human eye. FHI (Fuji Heavy Industries) in May launched one such system, Eyesight, in a model-year 2009 Subaru Legacy. Eyesight uses stereo cameras to enable a precrash brake, which decreases damages from a crash; a sensor that detects obstacles and prevents a car from starting; and a full-speed-range adaptive-cruise-control system that keeps vehicles apart from each other even when they are standing still. It consists of twin cameras, one on each side of the rearview mirror, that use humanlike stereoscopic vision to judge distances and generally keep tabs on the driver (Figure 1). The system can help you keep your distance from other cars on the highway, provides a lane-departure-warning system and a wake-up call should everyone pull away from the lights but you, and even keeps an eye out for pedestrians. In contrast to Eyesight’s camera-based system, precrash systems using radar detect only the car ahead. Eyesight can detect not only the vehicle directly ahead but also the vehicle diagonally ahead. It also detects bicycles, pedestrians, lane hogging, and swerving.

Eyesight’s stereo cameras integrate two monochrome CCD (charge-coupled-device) sensors with 0.3M-pixel resolution. The device uses processed images from the cameras to decipher a vehicle’s location. It then indicates the location to the sensor, which starts control of the engine and transmission using a 500-kbps CAN (controller-area network) for communication. Eyesight integrates the stereo cameras and an ECU (electronics-control unit) into one module, dramatically reducing system cost. The stereo-camera unit integrates an ASIC for image processing, a 32-bit microcontroller for image recognition, and another 32-bit microcontroller for control. According to FHI, Eyesight costs approximately 200,000 yen ($1850), which is less than half the price of the company’s other driving-support system.

CMOS image sensors for consumer products, such as digital cameras, achieve resolution of several megapixels or greater. On the other hand, the image sensors for monitoring the rear of vehicles use CCD sensors with only 0.2M- to 0.3M-pixel resolution. However, Japanese manufacturers are striving to develop even more advanced image sensors to improve safety. Nissan Motor recently demonstrated an example of such a sensor, the Around View Monitor, on the Elgrand minivan, which the company released in Japan in October 2007. The sensor eases parking by providing greater visibility of the surrounding environment. It synthesizes a bird’s-eye view of the vehicle and its surroundings using Sony 3000, ultrawide-angle, 1.3M-pixel cameras on the front, sides, and rear of the vehicle.

Mazda Motor introduced a rear-vehicle-monitoring system in its new Atenza (Mazda6 outside Japan), which it launched in January 2008. When the car is operating at speeds greater than 60 kph (37 mph), this system uses 24-GHz microwave radar to detect vehicles approaching from behind and alerts the driver to the danger of changing lanes. Two sensors near the rear of the vehicle monitor objects that enter the detection area, activating an icon in the driver’s or passenger’s side-view mirror. If the driver attempts a lane change when an object is in the detection area, the icon flashes, and an audible beep sounds.

In another development, Suzuki Motor and Keio University have co-developed a driver-monitoring system that extracts the feature points of a driver as a 3-D pattern using multiple cameras rather than the 2-D pattern and one camera that many other systems use. The 3-D system combines with a particle filter, which recognizes patterns in temporal sequence, so that it simultaneously detects both the orientation of the driver’s face and the direction of the driver’s eyes.