Highway travel time estimation with data fusion

摘要

Travel time information is the key indicator of highway management performance and one of the most appreciated inputs for highway users. Despite this relevance, the interest of highway operators in providing approximate travel time information is quite recent. Besides, highway administrations have only recently begun to gather such information as a means to measure the accessibility service provided by the road, in terms of quality and reliability.\ud\udIn the last century, magnetic loop detectors played a fundamental role in providing traffic volume information and also, with less accuracy, information on average speed and highway occupancy. New traffic monitoring technologies (intelligent cameras, GPS or cell phone tracking, Bluetooth identification, new MeMS detectors, etc.) have appeared in the recent decades and permit considerable improvement in travel time data gathering. Some of the new technologies are cheap (Bluetooth), others are not (cameras); but in any case most of the main highways are still monitored by magnetic loop detectors. It makes sense to use their basic information and enrich it, when needed, with new data sources.\ud\udThis monograph presents a simple approach for the measurement and short term prediction of highway travel times based on the fusion of inductive loop detector and toll ticket data. The methodology is generic and it is not technologically captive: it could be easily generalized to other equivalent types of data.\ud\udBayesian analysis makes it possible to obtain fused estimates that are more reliable than the original inputs, overcoming some drawbacks of travel time estimations based on unique data sources. The developed methodology adds value and obtains the maximum (in terms of travel time estimation) of the available data, without falling in the recurrent and costly request of additional data needs.\ud\udThe application of the algorithms to empirical testing in the AP-7 toll highway in Barcelona proves that it is possible to develop an accurate real-time travel time information system on closed toll highways with the existing surveillance equipment. Therefore, highway operators could give this added value to their customers at almost no extra investment.