The mileage of a vehicle is the sum of the distance traveled within a defined period of time. This parameter is particularly essential for quantifying the strain on infrastructure and the environmental impact of traffic. In their work, Dr. Naghmeh Niroomand from the ZHAW School of Management and Law and Christian Bach from Empa are changing the traditional approach to estimating vehicle mileage by using artificial intelligence (AI) and deep learning techniques. By employing the innovative method, the researchers were able to classify the passenger cars and thus enable a more precise assessment of the average vehicle mileage. The new method is presented in the paper “Estimating Average Vehicle Mileage for Various Vehicle Classes using Polynomial Models in Deep Classifiers”(PDF 2,8 MB).

The increasing electrification of road transport poses a challenge when assessing CO₂ emissions. “While road traffic has traditionally operated within its energy system, which made assessing CO₂ emissions relatively straightforward, the increasing electrification of road traffic poses challenges in distinguishing energy consumption between road traffic and other stationary energy uses” explains Niroomand, who is responsible for the project. In the past, it was easy to determine CO₂ emissions from cars because they could be estimated based on fuel consumption or distance traveled. In the area of electrification, this is more complex because the electricity distributed via the power grid is used for many other things in addition to charging an electric car. “Given the global trend to-wards electric vehicles, the accurate estimation of mileage and emissions models is more important than ever. Our AI driven approach not only provides a precise mathematical model to estimate the average mileage of vehicles, but also lays the foundation for informed decisions in sustainability initiatives,” says Niroomand.