Vehicle travel distance and time prediction using virtual detection zone and CCTV data

Abstract

This paper proposes a solution to traffic monitoring and planning by using nonintrusive traffic sensors only, namely Virtual Detection Zone (VDZ) and Closed Circuit Television (CCTV). We propose that our novel method, VDZ with CCTV snap shots, when combined as a system, can replace the traditional system using inductive loop detector sensor. The latter is in disadvantage because inductive loop detectors need to be physically embedded in to a highway-surface. Our previous works have shown that our system can provide the empirical data to construct Fundamental Diagrams and to calibrate a chosen model of the target highway, by using only seven VDZ agents which have carried GPS enabled smartphones, together with snapshots from fifteen existing CCTV, on Tangerang to Jakarta highway (21 km distance) over a period of three months. Our system is better than the traditional system, because it: can show zero speeds at totally jammed density, uses only one GPS coordinate to form a speed aware zone, costs less to build and is more practical to implement, especially for developing countries like Indonesia. By adopting and modifying Cell Transmission Model or CTM, a variant Macro model, our system can give short term predictions for travel time, and travel distance. Graphical and tabular comparisons are presented to highlight the worsening or improvement effects on traffic conditions when certain highway topology parameters have been changed for our target road. For example, simulated on our already calibrated model on our target highway, an addition of 3 off ramps at km 13, 15 and 17, would cause a reduction of vehicle travel time by up to 17.6% between minutes 75 to 230, which also mean a reduction of traffic jam. These results show that this system is potentially beneficial not only for commuters, but also for highway designers and government road authority.