Reinforcement learning for optimising EV climate control systems

Presenter: Dr James Brusey Professor of Computer Science
Topic: Reinforcement learning for optimising climate control systems
Date: Thursday October 19th 2023
Time: 18:30 for 18:40 lecture start
Location: Webinar – recording available from here


The EU H2020 DOMUS project aimed to develop and demonstrate a revolutionary approach to the design of electric vehicles (EVs) from a user-centric perspective for optimal efficiency. The project’s main objective was to reduce the energy consumption of future EVs in order to increase their range by 25%, without compromising passenger comfort and safety.

In this talk, we will look at one aspect of the project, which was to virtualise the car cabin and then use Reinforcement Learning to develop an optimal controller. The saving from this method alone was 20%. A key innovation in this work was the use of machine learning to develop fast simulations based on Computational Fluid Dynamics (CFD) models of the thermal environment within the cabin. While this technology is almost ready to be used in passenger vehicles, on-going studies will be looking at ensuring that the improved performance translates smoothly to real world vehicles and that the approach can be used to personalise comfort and optimise according to specific vehicles and climates.

James Brusey is a Professor of Computer Science and the lead for AI for Cyber Physical Systems with the Centre for Data Science at Coventry University. His current research is in applied wireless networked sensing, including projects in residential building energy and comfort sensing, algorithms for network packet reduction, decision support for buildings monitoring, car cabin comfort sensing and control, elderly / infirm falls and near-falls sensing.

James Brusey received his BAS with distinction and PhD from RMIT University in 1996 and 2003, respectively. His PhD won the Australian Computer Science Association award for Best Thesis in 2004. He has over 15 years experience in the IT industry, part of which was as an independent consultant. Since 2007, James has worked at Coventry University at the Cogent Computing Applied Research Centre. During this period, he has helped to establish Cogent as a world class research centre with a broad portfolio of successful, industry-sponsored projects related to wireless sensor networks. In 2012, he was awarded a readership in Pervasive Computing and in 2013, he took on the directorship of the Cogent Computing Applied Research Centre, and was awarded a professorial chair in computer science in 2018.

His current research interests include exploring issues with the practical deployment of wireless sensor networks, thermal comfort in buildings and car cabins, reinforcement learning, and bayesian methods.

Comments are closed.