Real time traffic states estimation on arterials based on trajectory data

• We propose a method for traffic states estimation on arterials using trajectory data.
• The method provides good results with very limited amount of demand data.
• The method is able to predict traffic states, queue lengths and travel times.
• Analysis shows the good accuracy of the method, especially during congested and transition periods.
• Computational time is low, so the method is feasible to implement in real time.

New technologies able to register vehicle trajectories, such as GPS (Global Position Systems)-enabled cell phones, have opened a new way of collecting traffic data. However, good methods that convert these data into useful information are needed to leverage these data. In this study a new method to estimate traffic states on arterials based on trajectory data is presented and assessed. The method is based on the Lighthill–Whitham–Richards (LWR) theory. By using this theory, traffic dynamics on arterials can be better captured by extracting more information from the same piece of data. Trajectory data used consist of the trajectory of the latest equipped vehicle that crossed the segment under study. Preliminary analysis based on micro-simulation suggests that this method yields good traffic state estimates both at congested and uncongested situations, even for very low penetration rates (1%). The method is also able to forecast queue length at intersections and travel times along a road section.