Estimate Dynamic Road Travel Time Based on Uncertainty Feedback

With the development of a smart city and an Intelligent Transportation System, more and more sensors with heterogeneous features are being and will be deployed in cities. This development inevitably fosters demand for a tool which can provide consistent and comprehensive pictures of multiple types and handle large amounts of data. Data fusion is a necessary and sufficient technology for achieving these aims. However, adding increasing volumes of data from different sources associated with advanced traffic monitoring, along with other conventional traffic measurement instruments, into a data fusion system to improve data quality and reduce uncertainty is challenging. In this study, we propose a novel data fusion framework that fuses multiple data sources in a uniform Spatio-Temporal context. A novel fusion model, based on uncertainty feedback (UNIF), was developed to estimate road travel time based on measuring uncertainty. Our method was evaluated using four data sources of large-scale real-world traffic data. We obtained encouraging results for the performance indicators in urban traffic applications on a large scale.