The broader impact/commercial potential of this Small Business Innovation Research (SBIR) project is to democratize access to scientific techniques in vehicle energy modeling, making them intuitively available to all car buyers across the country. Unfortunately, the typical car buyer does not currently have access to information to determine their fuel consumption, costs, and range viability for different vehicles they may be considering for purchase on their own driving conditions. Further, car buyers typically do not have measurements of their mobility patterns in their current vehicle, making it further difficult to compare cars. Thus, car buyers have limited ability to understand the economic value in choosing a fuel-efficient vehicle which may cost more upfront but save them significant money in the long run. By providing greater access to information on the fuel consumption and costs that car buyers will experience in any vehicle they are considering, this project can accelerate the uptake of fuel-efficient vehicles. By accelerating the uptake of fuel-efficient cars, the team projects this project can enable up to 30-50 billion gallons of avoided petroleum use, and up to $450-680 billion of avoided fueling costs. This SBIR Phase 1 project proposes to apply data science, machine learning, and convex optimization techniques to develop and apply vehicle energy models in circumstances where only sparse and disparate sources of data are available. These circumstances represent use cases that are typically encountered by the vast majority of car buyers. To overcome the challenges posed by only sparse and disparate sources of data being available during the car comparison process for car buyers, this project will develop techniques for formulation and calibration of vehicle energy models using time-resolved, trip-resolved, and tank-resolved fuel consumption data. Further, this project will develop probabilistic techniques for trip profile generation to create speed/terrain profiles for given trips using origin-destination-departure time data or intermittent measurements of speed-position along a trip. These probabilistic techniques for trip profile generation can be combined with vehicle energy models to allow car buyers to compare any car they are considering for purchase, on their own driving conditions. The techniques developed will be made available for use by car buyers through implementation in a smartphone app and web-based tools that are easy and intuitive for car buyers to use during their car shopping process. 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.
