— The explanatory text is provided here in English only. However, the thesis may also be written in German. —
Context & Motivation
The goal of the thesis will be to develop a Human-Aware agent in the Cooperative Cuisine environment Inspired by the multiplayer video game Overcooked, the environment tasks players with controlling chefs who must work together to fulfil various orders within a kitchen setting. The environment demands real-time, cooperative human-agent interaction and offers extensive customization options, including layout design, agent count, time constraints, and information transparency.
A Human-Aware agent can focus on various aspects:
- Coordination: Dynamic task assignment and flexibility
- Cognition: Maintaining mental models of other agents and the environment
- Social Intelligence: Preference learning and socially-aware navigation
- Communication: Effective information exchange between partners
- Giving explanation of own behaviour
- Intent Recognition: Recognizing the next actions/goals of other agents
Given the breadth of these capabilities, the thesis will focus on the implementation and evaluation of one or two specific aspects.
Methodology
The architecture of the Human-Aware agent must remain transparent and explainable. Thus, an approach utilizing ACT-R (Adaptive Control of Thought—Rational) or Bayesian Modelling (with rational boundedly agents) is preferred.
Additional Resources
Cooperative Cuisine environment: https://scs.techfak.uni-bielefeld.de/cooperative-cuisine
Schröder F, Heinrich F and Kopp S (2025) Towards fluid human-agent collaboration: From dynamic collaboration patterns to models of theory of mind reasoning. Front. Robot. AI 12:1532693. https://www.frontiersin.org/journals/robotics-and-ai/articles/10.3389/frobt.2025.1532693
Wu, S.A., Wang, R.E., Evans, J.A., Tenenbaum, J.B., Parkes, D.C. and Kleiman-Weiner, M. (2021), Too Many Cooks: Bayesian Inference for Coordinating Multi-Agent Collaboration. Top. Cogn. Sci., 13: 414-432. https://doi.org/10.1111/tops.12525
Zhi-Xuan, T., Mann, J. L., Silver, T., Tenenbaum, J. B., & Mansinghka, V. K. (2020). Online Bayesian goal inference for boundedly-rational planning agents. Advances in Neural Information Processing Systems, 2020-December.
Sound like your kind of role? Let’s talk.
If you are interested, please contact Timon (timon.dohnke@uni-luebeck.de) with the following information:
- Who are you, what are you studying and what stage are you at (which semester)?
- What is your background? Have you already taken any of our modules (e.g. HMI, Intelligent Agents, CCS)? (recommended)
- When would you like to start and what are your temporal requirements?
- What interests you about this topic? Which aspects are particularly relevant to you?
- Do you have any preferences regarding what you would like to learn during your thesis or how you would like our collaboration to be structured?