Advanced passive safety system via prediction and sensor fusion

This paper describes a new method that employs a proximity sensor to detect a frontal collision by which the reliability of the whole automotive air bag system is expected to be greatly improved. To build the highly accurate and highly reliable systems, we must consider the following problems: 1) the air bag system can detect accurately whether the automobile collides to an obstacle; 2) the system can judge quantitatively whether the air bag should be triggered; and 3) when the air bag must be triggered, the system can expand the bag at the optimal timing. To solve the above problems, the following five processing blocks are prepared: 1) accurate detection of frontal collision; 2) estimation of both the acceleration and jerk from noisy measured acceleration by Luenberger observer; 3) filtering to estimate the velocity and displacement of the occupant's motion; 4) prediction of the velocity and displacement of the occupant's motion; and 5) judgement to trigger the air bag.<>