Urban Traffic Incident Detection using Qualitative Data and Temporal Intervals

摘要

In this paper, we explain a method for automatic bounded-time identification of urban traffic incidents. Several incident detection agents are being developed by analyzing TV traffic camera images. The method explained in this paper makes use of the most common data sensors available in an urban traffic scenario: the loop wire sensors. The method is an evolution of the method developed in the ESPRIT II EQUATOR project. It uses qualitative preprocessing wire sensor data and qualitative simulation data to figure out the incidents present at an urban traffic domain. The qualitative preprocessing data are built every time that actual sensor data are available. In the temporal intervals without actual sensor data, a qualitative simulation process is executed. These different data sources are compared to the diagnosis agent of the Intelligent Urban Traffic Control Multiagent System project.