This paper describes a study of vehicle localization using its vibration due to road surface roughness. The authors suggest a vehicle localization scheme in which a vehicle can identify its location on the road. The vehicle localization can be carried out matching the current vehicle vibration with the prior observed vehicle vibration in the database as a vibration map. In this paper, the longitudinal vehicle localization is carried out by observing the vehicle vibration with a smart phone on the dashboard. First, the authors examine the effect of the matching duration of the current vehicle vibration data to the prior observed vibration for the precise of vehicle localization. It is shown that the long duration matching is effective in more precise vehicle localization. Next, the authors evaluate the performance of the vehicle localization scheme using current vehicle vibration data and prior vibration data in the database observed from the same vehicle, and using the ones from a vehicle and the ones from another vehicle. It is shown that the vehicle localization error is about 2.5 meter when using vibration data from the same vehicle, and about 3.5 meter when using vibration data from a vehicle and another vehicle.
展开▼
