Autonomous process management based on Reinforcement Learning: Practical application in industrial environments [Industrial Doctorate nº 2019 DI 99]

[Call: Doctorats industrials DI 2019]

Student: Lluís Echeverria Rovira (company / entity: Eurecat)

Thesis director: Cèsar Fernández

The following stages are carried out for the development of this doctoral program:

1. Study of the state of the art in the field of intelligent management of industrial environments, such as wastewater treatment plants or manufacturing industry.
2. Search for algorithms, in the field of Artificial Intelligence, for the optimal management and control of processes autonomously (Dynamic Programming, Learning by reinforcement, …).
3. Search for algorithms, in the field of Artificial Intelligence, for the implementation of state approximation functions in sets of very large spaces (continuous and / or infinite) (Deep learning and its derivatives)
4. Research and development of methodologies and algorithms to develop simulation environments in which to carry out the training of autonomous control agents.
5. Research and development of advanced reward methodologies and techniques for the implementation of autonomous control agents that take into account complex metrics and such as making the most of available resources or maximizing process results.
6. Research and development of transfer learning methodologies (Transfer Learning) for the deployment of new autonomous agents in environments and conditions similar to the original.
7. Test and experimental deployment of agents in real environments for validation.

This project has received the support of the Secretaria de Universitats i Recerca del Departament d’Empresa i Coneixement de la Generalitat de Catalunya