Predictive Energy Management in Uncertain Traffic with Learning-based Optimal Control

Job description

Master Thesis: Predictive Energy Management of Heavy Vehicle in Uncertain Traffic with Learning-based Optimal Control

 

Suggested Time Schedule: Jan 2025 – Jun 2025

 

Heavy-Vehicle transport is among the most significant contributors to CO_2-emissions and fuel consumption [1]. An important measure to elevate this problem is to increase the vehicles’ energy efficiency. Much effort has been made towards minimizing losses in the vehicle powertrain through incremental hardware improvements. An attractive, hardware independent, alternative is to apply principles from “Eco-driving” to operate vehicles more efficiently. Extensive prior work demonstrates that limiting rapid velocity changes and varying the vehicle speed based on the road topography, significantly reduces the energy consumption. To this end, Model Predictive Control (MPC) has been identified as a particularly well-suited approach, demonstrating efficiency improvements of up to 10% in highway driving simulations [2]. However, in real-driving scenarios the vehicle velocity is often restricted by the surrounding traffic, which can have a detrimental impact on the ability to track the optimal velocity. 

More recent work has investigated how the traffic uncertainty can be addressed in the MPC problem by learning the powertrain capabilities of the closest vehicle ahead (lead-vehicle) [4]. This approach was further extended to an adaptive setting using online parameter estimation [5]. While the idea is attractive, much work can still be done regarding how to estimate the parameters efficiently to obtain a solution that is realizable in practice.

 

The purpose of this thesis can be summarized as:

1.    Study relevant literature in Predictive Energy management and Learning-based Optimal Control.
2.    Design a Learning-based observer that estimates relevant properties of the leading vehicle using internal Volvo GTT data. 
3.    Formulate a Model Predictive Control problem that optimizes the vehicle speed for energy efficiency while using Machine Learning to handle the lead-vehicle state uncertainty. 
4.    Implement the Optimal Controller with the learning-based observer in existing Volvo GTT solutions and evaluate performance in suitable simulations.

 

The aim of this thesis is to use MPC and machine learning to solve problems within vehicle energy management. The work will include optimization, machine learning, and programing. This work will be carried out together with Volvo Group Trucks Technology. The thesis is recommended for one or two students with a strong background in optimization, control, machine learning and Matlab/Simulink with a solid mathematical background. Prior experience with data analysis and vehicle simulation is meritorious. 

Contact persons:
Nikolce Murgovski – Chalmers University of Technology
tel: +46 72 170 51 72
mail: nikolce.murgovski@chalmers.se

Erik Jonsson Holm – Volvo GTT- Safe and Efficient Driving
Mail: erik.jonsson.holm@volvo.com

Esteban Gelso – Volvo GTT- Safe and Efficient Driving
Mail: esteban.gelso@volvo.com

Erik Börve – Volvo GTT / Chalmers University of Technology
Mail: erik.borve@volvo.com

Data Person (?) – Volvo GTT- Safe and Efficient Driving
Mail: data.person@volvo.com
 

...

Job Category:  Technology Engineering

Organization:  Group Trucks Technology

Travel Required:  No Travel Required

Requisition ID:  14944