Advanced automobile crash detection by acoustic methods Free

Today’s automobiles are equipped with a variety of safety features. On-board acceleration, rotation, steering angle, wheel speed, and throttle position sensors are widely incorporated in electronic stability control systems for traction and steering control. Cameras and even radars are increasingly employed for imminent crash detection and autonomous emergency braking whereas airbags are deployed during an actual collision. In contrast, use of acoustic signals for automobile pre-crash and crash detection has been somewhat limited even though sounds that occur during such events can offer valuable information. For example, the high-pitch squealing caused by tire skidding can provide advance warning especially if it is caused by an adjacent car. During collision, the acoustic waves traveling along the steel car frame are 17 times faster than the speed of sound in air, which can provide information more promptly than center-mounted acceleration sensors. To fully take advantage of the high-speed acoustic signals, a real-time, wavelet-based, acoustic signal processing method with sub-millisecond precision has been developed, offering distinct advantages in sudden onset detection, temporal localization accuracy, and computational cost over existing time- and frequency-domain methods. Demonstration results on different crash scenarios will be presented, which are indicative of a substantial enhancement of crash detection performance.