Machine learning for the improvement of fuel efficiency on hybrid vehicles

Anind K. Dey, Computer Science: Human Computer Interaction Institute

Hybrid vehicles have become and are the current next step for automotive systems. Current technologies report huge savings on fuel consumption in the city as well as on the road using a synergy between the electrical motor and the combustion engine through a continuous variable transmission in most cases. This synergy allows the use of the motor and engine in their most efficient operation areas. However further fuel efficiency could be achieved by implementing better control strategies that may help to increase the fuel savings by trying to predict the driver’s driving pattern, terrain conditions and traffic conditions. This exploratory project looks to bring Machine Learning to the automotive industry and generate new ways to improve current control strategies for the usage of the electrical motor and the engine based on personalized algorithms learned and improved on the fly for each driver.

Skills required:

We are looking for a highly motivated and energetic undergrad from electrical or mechanical engineering willing to do research in a multidisciplinary atmosphere, willing to learn new concepts from the machine learning and Human computer interaction fields. The successful candidate should have, or have interest in, some of the skills listed below:

• Matlab:   Basics on Simulink and Stateflow
• Control theory:   Basics, some knowledge of adaptive, stochastic and optimal control
• Automotive systems: Basics on hybrid vehicles (functioning, control strategies)