Location: X4309, Jiuli campus
It is a longstanding goal for transportation researchers to model and understand how traffic distribute and propagate along road networks. In this research, a data fusion and information integration approach is proposed to model and interpret traffic along urban arterial corridors based on heterogeneous traffic data. The method attempts to deduce the “most probable” explanation that can integrate fixed-location data and mobile sensing data and match the vehicle records at upstream
with those collected at downstream sensor locations. To make the probabilistic model more realistic, traffic knowledge such as lane choice decision, traffic merging and travel time information are calibrated using the historical dataset and then integrated into the model. By doing so, the model can obtain individual travel times of the matched data pairs, which can be directly used to estimate corridor travel times of individual vehicles. The paper also explores some practical issues related to the use of heterogeneous traffic data, such as “inaccurate data”, and “detector failure” problem.Results from the method can be further applied to estimate vehicle trajectories along arterial corridors, estimate individual vehicle-based fuel consumption/emissions, and help infer real-time queuing processes at signalized intersections.