Turstworthy AI for transport-related emission estimation and EU policy making

Andres L. Marin

Universitat Politècnica de València

Private vehicles have a significant negative impact on the environment, with transport being responsible for more than 20% of energy-related emissions. The EU’s objective is to reduce the emissions of CO2 in 2030 by 55% with respect to 2021. For that reason, a reduction of the emissions from vehicles is urgent. A significant reduction can be achieved by promoting more efficient driving strategies. This study aims to address the complex relationship between individual driving habits and energy consumption, and the role of drivers in causing emissions. With the proposed collaboration, we aim to apply trustworthy AI techniques to calibrate and explain the energy consumption predictions from the models based on driver behaviour. This could be used to provide feedback and recommendations to drivers on how to adopt more efficient driving strategies and policy makers to take appropriate measures. This is essential for the application of regulation or oficial recomendations to mitigate the impact of transport to CO2 emissions. The research will involve a visit to the University of Bristol, where the author will collaborate with Prof. Peter Flach’s group, who are experts in robust evaluation, calibration, and explainability of AI models.

Keywords: Trustworthy AI, Transparency, Explainability, AI evaluation, Transport, Artificial Intelligence, Human activity charaterisation

Scientific area: Artificial Intelligence applied to Energy & Transport

Biography: Andres L. Marin is a PhD student in computer science with the VRAIN group of Universitat Politècnica de València. Marin is also part of the Low Energy & GhG Emissions Neutral Transport group (Joint Research Center, European commission). While particularily nterested in energy consumption, human driver behavior and Deep Learning, Marin’s background is in Physics and Artificial Intelligence. Marin mainly work on artificial intelligence applications in transport for emissions reduction.

Visiting period: 01/03/2024 to 31/03/2024, at Intelligent Systems Laboratory, the University of Bristol