摘要: |
By 2020, traffic delay is froecasted to cost 8.4 million hours for society and result in a fuel waste of 4.5 billion gallons in the United States (U.S.). Besides the wasted time and fuel, incidents also cause local pollution (due to higher levels of emissions), and injuries/fatalities. Roadway accidents are responsible for the majority of this high toll a 57.9%. If an incident is not cleared in a timely fashion, the queue back-up due to incidents can further block nearby ramps or intersections, causing additional delays. Early incident detection is also reported to save lives by increasing the survival probability of an injury accident victim.
In order to reduce these negative impacts, government agencies invest in Intelligent Transportation Systems (ITS) infrastructure (such as traffic sensors and cameras) for better management of traffic, including roadway incidents. ITS infrastructure includes and array of information collection systems to share real-time data with integrated traffic control systems and advisory alerts designed to manage traffic, detect incidents, and provide travelers with current route information. A wide range of statistical inference methods and algorithms as well as commercialized products are suggested for efficient and accurate detection. However these technologies heavily depend on input from ITS infrastructure, i.e. magnetic loop detectors, Bluetooth readers, traffic cameras, etc. Consequently, such sensor dependent incident management (IM) strategies come with substantial infrastructure and maintenance costs. Incident detection methods with lower costs can yield a very high cost-benefit ratio. Use of social media feeds to detect traffic indidents is one such approach which does not require any infrastructure investment, yet has shown to exhibit a strong potential for effectiveness. |