[2O6-GS-13] AI application: Traffic

[2O6-GS-13-04] Traffic anomaly detection using ETC2.0 probe data

〇Masaki MATSUDAIRA¹, Sachia FUJITA¹ (1. Oki Electric Industry)

Keywords:ETC2.0 probe data, traffic anomaly detection, traffic flow model

Detecting traffic anomalies such as accidents and obstacles on freeways is one of serious problems for traffic management. In this paper, we propose the algorithm for detecting automatically traffic anomalies using ETC2.0 probe data. Our algorithm is based on some features: distance from jam head position to recovered speed position, acceleration at jam head, timespan from free traffic flow to jam traffic flow, and time difference from jam to free flow at 2 positions on downstream, which is calculated using probe data. We evaluated our algorithm with 400 jam data on 2 roads by precision and recall score. The result is good, precision 81.0% and recall 85.0%.