Evaluating a Signalized Intersection Performance Using Unmanned Aerial Data

TL;DR

This study introduces a drone-based method to measure intersection performance metrics in real-time, providing detailed vehicle behavior analysis and supporting traffic management improvements.

Contribution

It presents a novel approach using drone-collected data and shockwave analysis to estimate multiple MOEs at a signalized intersection, enabling real-time traffic assessment.

Findings

Drone data effectively estimates MOEs like queue length and delays.

Real-time MOE estimation from drone data is feasible.

Microscopic models capture transient vehicle behaviors.

Abstract

This paper presents a novel method to compute various measures of effectiveness (MOEs) at a signalized intersection using vehicle trajectory data collected by flying drones. MOEs are key parameters in determining the quality of service at signalized intersections. Specifically, this study investigates the use of drone raw data at a busy three-way signalized intersection in Athens, Greece, and builds on the open data initiative of the pNEUMA experiment. Using a microscopic approach and shockwave analysis on data extracted from realtime videos, we estimated the maximum queue length, whether, when, and where a spillback occurred, vehicle stops, vehicle travel time and delay, crash rates, fuel consumption, CO2 emissions, and fundamental diagrams. Results of the various MOEs were found to be promising, which confirms that the use of traffic data collected by drones has many applications. We also demonstrate that estimating MOEs in real-time is achievable using drone data. Such models have the ability to track individual vehicle movements within street networks and thus allow the modeler to consider any traffic conditions, ranging from highly under-saturated to highly over-saturated conditions. These microscopic models have the advantage of capturing the impact of transient vehicle behavior on various MOEs.