— The explanatory text is provided here in English only. However, the thesis may also be written in German. —
We are planning to have a thesis that investigates how autonomous systems can anticipate future states with increasing uncertainty when making decisions over time. The focus lies on how such systems can reason about the likelihood that a situation will become more difficult, how quickly external assistance could realistically be provided, and what the consequences are of intervening too late. Based on these factors, the project aims to explore how a system can determine when it should continue acting autonomously and when it should request help or transfer control.
Why is this topic relevant?
The trade-off between autonomy and safety arises from three interacting factors: the anticipated development of uncertainty, the latency of external assistance, and the cost of delayed intervention. As uncertainty increases, the probability of future system failure or misalignment also rises; however, seeking help too early can unnecessarily interrupt autonomous operation or disturb the human. At the same time, the usefulness of human or external support depends on how quickly it can be provided, since delayed responses may come too late to prevent errors. Finally, the consequences of waiting too long – such as safety-critical failures, degraded performance, or loss of trust – define the cost side of the decision. In human–robot interaction and autonomous driving, this creates a threshold problem in which systems must decide when expected future risk, weighted by potential cost, justifies interrupting autonomy in favour of external input or human control.
Focus of your work
This is an exploratory project intended to first develop and test the core idea and assess its feasibility in a structured research setting. A central objective is to formalise a model capturing this decision process, which can be applied across different experimental contexts, depending on the specific research question. The proposed model should then be tested in a small evaluation study.
Depending on your interests, different questions can be investigated. Your own interests and suggestions are expressly welcome and can also be developed further in the course of your work!
What you should bring with you
- General foundation in areas relevant to decision-making under uncertainty, such as computer science, robotics, cognitive science, data science, or a related discipline or a willingness to familiarise yourself with the topic
- An interest in topics relating to human-centred AI and AI systems designed to support people in their daily lives
- A particularly independent and self-motivated approach to work, as it involves a great deal of developing your own ideas – what we are, of course, happy to support with
What you will learn (at least) during your thesis
- Modelling decision-making under uncertainty and reasoning about future system states
- Understanding and formalising trade-offs between autonomy, risk, and intervention costs
- Conducting a small-scale empirical evaluation of the proposed approach
- Interpreting results in terms of system behaviour, robustness, and practical feasibility
What we offer
- Application of theoretical concepts with direct connection to real-world scenarios
- Expertise in various fields, such as psychology, cognitive science, computer science, and robotics
- A reliable contact person who will support you with questions or problems
- The satisfaction of having contributed to ongoing research, meaning that your thesis will have added value beyond its immediate scope
- Communication and thesis in German or English (depending on your preference)
Sounds like your kind of role? Let’s talk. Please let us know:
- 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?
- 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?