Estimation of Unobserved Vehicles in Congested Traffic from Probe Vehicle Samples

Probe vehicle data is increasingly becoming widely available thanks to advancements in connected and automated vehicles (CAVs) and the prevalence of personal mobile devices such as smartphones. As a result, monitoring and estimating different network conditions such as speed and travel times based on this data have been proven to be greatly beneficial. However, there have been comparatively limited studies on extracting other critical traffic flow variables such as density and flow. These limited number of studies mostly develop algorithms to extract and exploit the somewhat regular patterns solely in fully stopped regions of the congested traffic. This paper proposes a more generalized microscopic approach that is applicable to a wider range of traffic conditions. A naïve Bayes model is implemented to estimate the number of unobserved vehicles in between a set of probe vehicles. The parameters needed for the naïve Bayes include means and standard deviations for the probability density functions (pdfs) for the distance headways between vehicles. The proposed model is tested based on the trajectory data collected from US 101 and I-80 in California for the FHWA’s NGSIM (next generation simulation) project. Under the various traffic conditions analyzed, the results show that the number of unobserved vehicles between two probes can be predicted with reasonably high accuracy for mixed traffic conditions.