Improving Vehicle Localization in a Smart City with Low Cost Sensor Networks and Support Vector Machines

A smart city's main purpose is to provide intelligent responses to different problems of the rapid urban population growth. For instance, integrating fleet management solutions into intelligent transportation systems (ITS) can efficiently resolve transportation problems relying on each vehicle information. Usually, the position estimate is ensured by the integration of the Global Positioning System (GPS) and Inertial Navigation Systems (INS). For multisensor data fusion, the Extended Kalman Filter (EKF) is generally applied using the sensor's measures and the GPS position as a helper. However, the INS are expensive and require more complex computing which induces restrictions on their implementation. Furthermore, the EKF performance depends on the vehicle dynamic variations and may quickly diverge because of environmental changes. In this paper, we present a robust low cost approach using EKF and Support Vector Machines (SVM) to reliably estimate the vehicle position by limiting the EKF drawbacks. The sensors used are a GPS augmented by a low cost wireless sensor network. When GPS signals are available, we train SVM on different dynamics and outage times to learn the position errors so we can correct the future EKF predictions during GPS signal outages. We obtain empirically an improvement of up to 94% over the simple EKF predictions in case of GPS failures.