Masters Thesis Proposal

Understanding human engagement using Transformers. 

 

Everyday interaction with technologies can become more and more detached, as technologies do not have the social awareness to adapt to our needs and interaction types. Therefore we aim to increase these levels of social awareness for robots by understanding human reactions during specific scenarios. One of these crucial reactions is engagement – it is important for robots to understand when users are engaged in the interaction as well as the moment when they disengage and move on. 

The aim of the project is to detect levels of (dis)engagement using machine learning techniques. As a starting point, we have available a dataset of brief language-based interactions between people and a social robot in a real-world environment.

 

Useful skills

• Programming experience (python, C++)

• Interest in Human Robot Interaction

• Knowledge of machine learning algorithms (LSTMs, Transformers)