Emergency vehicle detection using acoustic source localization techniques

Everyday, drivers use and process acoustic alerts when driving; examples include the emergency siren from an ambulance or a fire truck, or the sounding of a horn. These sounds alert the driver to an emergency vehicle in their vicinity, or an impending collision. However, as driver preference is leading to increased acoustic isolation from their surroundings, it has become necessary to develop a platform where these acoustic alerts can still be delivered to a driver via other means. This paper presents results from an ongoing project that is developing an on-board system for the rapid detection of emergency vehicles using acoustic source localization techniques. It will take the form of a driver assistance platform that detects the presence of an emergency siren and the direction from which it is approaching. This will be useful to both human drivers enjoying increasingly quieter cars, as well as autonomous vehicles. Our system uses a multichannel cross-correlation algorithm for the estimation of the direction-of-arrival of a known acoustic source. We report results in both controlled and uncontrolled acoustic environments.