Mobile Tolling - A Geotracking Capable Mobile Application for Toll Free Collection

Abstract

The traditional toll collection process uses two major charging techniques: manual collection, where the vehicle is forced to stop at toll plazas, is a reliable but both time consuming and costly method, and electronic collection, where the vehicles do not need to stop at toll plazas and are identified by an On-Board Unit (OBU), which still carries production and maintenance costs and forces the vehicles without an OBU to use the manual collection method. The necessity to improve the amount of time lost, the operational costs with manual toll operators as well as the need to implement tolling in already built roads, gave way to the creation of an automatic license plate recognition (ALPR) method so that multi-lane free flow scenarios were possible. This removed the time consumption constraint from the process but since the state of the art in ALPR still only achieves 85% reliability using OCR (Optical Character Recognition) engines and 96% reliability using Fingerprint engines, global operational costs processing the unidentified cars, which have to be manually validated, are still very high when we talk of handling millions of image recognitions per day. The goal of this internship is not to replace the OBU, but to study a method of improving the success rate of the automatic vehicle recognition by using mobile phones geolocalization services to track the vehicle position as a mean of identifying that the vehicle passed through a certain toll point. Using the events from both the ALPR and the Mobile Tolling application, the back-office can correlate the two, improving the confidence level for the identification close to 100%, reducing the operation costs with negative and false positive identifications.