Key points are not available for this paper at this time.
Systems already exist that reduce collision damage through collision-danger alarms and by operating such features as pre-tensioning seat belts and collision mitigation brakes when the system has determined the possibility of a collision by using millimeter-wave radar. Conversely, at present, carelessness in observing oncoming traffic accounts for a large percentage of head-on collisions. Timely collision warnings are effective in avoiding accidents and for mitigating the severity of the collision. However, warnings given too early even before the driver has had a chance to carry out normal evasive maneuvers, can annoy the driver. Accordingly, by adding a driver face direction sensor, the authors have developed in the present research a system that will only hasten the timing of warnings when the system has detected the direction of the driver's face and determined that they are not facing the front of the vehicle. The driver face detection sensor is comprised of a camera located in the vicinity of the steering wheel column and an image processing ECU. An algorithm was newly devised to lend strong reliability and high precision to the system under a variety of different circumstances. The system then uses this value to determine whether the driver is facing the front of the vehicle. By adding this sensor we have realized a pre-crash safety system (hereunder PCS) with a high level of precision whilst also reducing the number of nuisance warnings. The present paper will give an outline of the new PCS and give details of the algorithm used in the driver face detection sensor.
Ohue et al. (Mon,) studied this question.