Diploma Thesis: Air Path Control with Multi - agent Reinforcement Learning
A combination of multiple autonomous agents trained in a common environment with Multi-agent reinforcement learning (MARL) shall be applied to improve the control strategy of an air path control system for diesel engines. The following tasks shall be performed during this thesis:
- Literature review
- Environment Setup for MARL (supported by AVL)
- Applying MARL algorithms to create the model
- Model evaluation with different algorithms
- Integration of the machine learning model with the other SW modules
- Validation with test cases provided by AVL
FIELDS OF STUDY
- Telematics / Informatics
- Electrical / Electronic Engineering
- Strong programming skills in Python
- Being familiar with Python libraries and ML frameworks, such as Numpy, Pandas, Tensorflow, Keras, ...
- Control theory basics
Remuneration: The successful completion of the thesis is remunerated with a one-time fee of EUR 2,600 before tax.
According to the Austrian Employment of Foreign Nationals Act it is unfortunately not possible to assign graduate work to third- country citizens (Non-EU citizens) who study at a university abroad.