A framework to transform real-time GPS data derived from transit vehicles to determine speed-flow characteristics of arterials

Transportation professionals are increasingly looking at technology as a way of improving safety and mobility of the traveling public. Over the past decade, a number of technology-based systems have been designed and built in major metropolitan areas all over the United States. In recent years, there has been a lot of interest in “real time” transportation monitoring systems. In particular, highway and traffic engineers have focused on real time traffic monitoring systems, and transit engineers and operators have focused on real time vehicle monitoring systems.; The development of these real time monitoring systems has come with high initial setup costs. Application development, related to the utilization of this monitoring data, has so far been limited in scope. Traffic monitoring systems have been used primarily in the generation of traffic flow maps and incident detection. On the other hand, transit-monitoring systems have been used primarily in the generation of on-time reports and overall system performance. Is there any way by which information gathered by one system (e.g., the transit monitoring system) can be used to also monitor and assess information for the other system (e.g., the traffic system)?; This research looks at the use of real-time travel time and location data obtained from transit vehicles fitted with GPS units to derive relationships with average roadway speeds and levels of congestions. On-road travel time was obtained by using a test vehicle that “floated” in traffic. Three different mathematical procedures are used to derive this associative relationship between bus speeds and on-road travel conditions—linear regression, multiple regression, and neural networks. The results of this research show that neural networks produce the best results of the three models and can estimate roadway travel conditions over seventy percent of the time.