Travel Time Estimation Using Speed Predictions

Advanced traveler information systems (ATIS) use sensing, communication and data processing technologies to collect and disseminate real time road traffic information. Route guidance systems utilize this data to estimate and optimize travel times. Paths with the shortest travel times are then computed to suggest to the user, an optimal path to reach their destination. In this study, we examine the accuracy of popular travel time calculation techniques, that use historical, instantaneous and predictive data to calculate travel time. A dynamic time calculation technique that accounts for time varying road speeds is used and the driver is considered to experience changes in the traffic pattern as the journey progresses. We base our study on real world traffic data collected from Singapore. We calculate the travel times for trips on major routes based on these techniques and compare them with the true average travel time on the particular route. We also analyze the variability in travel time that can be experienced by the user when guided by different route guidance system. The results show that dynamic predictive routing using multiple prediction horizons provides a better estimate of actual travel times as opposed to routing algorithms that only utilize real-time data about current traffic conditions.