Acoustic Localization of Emergency Vehicle Sirens in Traffic
* Präsentierender Autor
Zusammenfassung:The presence of emergency vehicles poses a considerable challenge to assistiveand autonomous systems developed to assist in traffic-related processes. In order to enable such systems to correctly recognize and categorize the appearanceof high priority vehicles, knowledge of their siren signals' position or directionof arrival can be a great advantage in handling cases of emergency. The field ofdigital array processing has access to several methods for acoustic source localization, some of which are evaluated in this work regarding their applicabilityfor siren localization. For signal-specific pre- and postprocessing, a range ofsteps are considered. Localization methods are then directly compared in realnoise scenarios, with selected pre- and postprocessing steps applied to them.Robustness against source distance is analyzed and an experiment regardingmoving siren localization with emergency vehicles is concluded. Results showthat multiple methods perform sufficiently robust within the considered scenarios.