Detecting Electric and Hybrid Vehicles Using a Smartphone

Pedestrians have difficulty in noticing electrical vehicles (EVs) and hybrid vehicles (HVs) approaching from behind quietly. We propose a vehicle detection scheme using a smartphone carried by a pedestrian. We exploit a high frequency switching noise generated by a motor unit in HVs and EVs. In this paper, we propose an approach of machine learning which is robust over the ambient noise, vehicle type and vehicle speed. In our evaluation, a J48 classifier implemented on the smartphone can tell whether an EV or a HV is approaching or not in the accuracy of 92% and 82% respectively. The first alarm was issued as early as 11.6 seconds before the vehicle approaches the observer the most. The scheme can also tell the vehicle speed and vehicle type.