Basics and Challenges in Acoustic Vehicle Sensing Using Sidewalk Microphones

Shigemi Ishida,Shigeaki Tagashira,Akira Fukuda

DOI: https://doi.org/10.1007/978-3-030-42234-9_7

2020-01-01

Abstract:Vehicle detection is one of the fundamental tasks in the ITS (intelligent transportation system). Although automatic vehicle sensors have already been deployed to retrieve real-time traffic data, these sensors suffer from high deployment costs, resulting in limited deployment at high traffic roads. This paper presents a new approach of vehicle sensing using sidewalk microphones. We install stereo microphones at one side of a road and draw a sound map, i.e., a time-difference map of vehicle sound on the two microphones. A state-machine based algorithm analyzes vehicle passing drawn a sound map to detect vehicles. A simple image processing technique is applied to a sound map prior to analysis to reduce false negative detections. Experimental evaluations demonstrated that our vehicle sensing system successfully detected vehicles with an F-measure of 0.92. This paper also presents big challenges we are facing to extend our system to multi-lane high traffic roads.