[3M2-GS-10u] AI application

12:00 PM - 12:15 PM JST(3:00 AM - 3:15 AM UTC)

[3M2-GS-10u-05] A Traffic Demand Estimation Method under Observation Constraints during Congestion Using Historical Free-Flow Traffic Volume Data

〇Kazuki Yoshida¹, Kengo Okano¹, Hiroaki Koshikawa¹ (1. Oki Electric Industry Co., Ltd.)

Keywords:

Traffic Volume Prediction,Probe Data

Accurate comprehension of traffic demand is indispensable for enhancing the precision of congestion prediction. Typically, traffic demand is represented by traffic volume measured via vehicle detectors, and under free-flow conditions, they are assumed equivalent. However, when congestion imposes restrictions on the number of vehicles passing downstream, a discrepancy arises between demand and observed traffic volume. To address this, methods estimating traffic volume using spatiotemporal speed data from probe vehicles and the fundamental speed-volume relationship are well known. Nevertheless, spatiotemporal speeds during free-flow vary considerably, and minor speed fluctuations can cause substantial estimation errors. In this study, we propose a novel approach that leverages recent free-flow traffic volume data to estimate traffic volumes under congested and jammed conditions. Evaluation results demonstrate that the proposed method yields estimates that more closely align with actual measurements compared to conventional methods based on the speed-traffic volume relationship.