Transportation accounts for over a quarter of U.S. energy use and greenhouse gases as well as hundreds of thousands of premature deaths annually from toxic emissions such as Nitrous oxides. Therefore, reducing harmful vehicle emissions and energy consumption are important goals for our society and transportation science. A key challenge is the limited understanding of emissions and energy-consumption during real-world driving. This project investigates the potential of emerging vehicle big data to further the understanding of emissions and energy consumption during real-world driving. Currently underutilized by vehicle manufacturers and regulatory agencies, vehicle big data details emissions and energy use at high frequency and spatial resolution. It has rich information to help identify patterns of unacceptably high emissions or energy use as well as associated vehicle properties or road features. Such patterns will be used to improve prediction of emissions and energy use during real-world driving. In doing so, the research will lead to improved vehicle design and operation practices to reduce future emissions and energy use to save lives by improving air-quality and dampening climate change. It will also improve education through a creative eco-driving challenge to maximize distance travelled for a fixed energy (or emission) budget in a driving simulator environment.
The goal of this project is to build next-generation spatio-temporal informatics (STI) tools to analyze emerging vehicle big data such as on-board diagnostics data to further the understanding of real-world emissions and energy consumption. The specific aims are to explore a set of concepts and develop a set of spatio-temporal informatics tools to: (a) provide a mapping between the concepts in transportation science and current informatics methods, (b) conveniently represent common patterns of interest to transportation scientists and practitioners, (c) efficiently mine novel, useful and interesting spatio-temporal patterns from vehicle big data, (d) use mined patterns to improve the physical science models of real-world vehicle emissions and energy use, and (e) integrate research results in education via eco-driving activities. The project will advance STI knowledge and understanding in multiple ways. For example, it will probe new algorithms to detect statistically-significant linear hotspots of high emissions or energy inefficiency even if these are not along shortest paths by considering simple paths in a transportation network. Furthermore, it will design new strategies to efficiently mine spatio-temporal co-occurrence patterns even when those are not prominent globally over the entire road network. The project will broaden STI's focus from simple GPS-trajectory data to multi-attributed trajectory data such as vehicle on-board diagnostics data with hundreds of physical variables and constraints. It will also enrich current laboratory and test-track focused transportation science by improving understanding of real-world energy-use, emissions, and physical science models used to predict these factors.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
