Robert Babuška

This PhD project at Delft University of Technology focuses on safe planning and decision-making for long-horizon robot autonomy under uncertainty. The work sits in the Cognitive Robotics Department and is embedded in the Reliable Robot Control Lab , within the EU-funded OPERA consortium. You will develop methods that help robots operate among people over extended time spans while balancing deliberate planning with fast, reactive behavior. The central research challenge is to make robots act safely and reliably when their knowledge of the environment and human behavior is incomplete, and when predictions may fail. The position emphasizes methods with clear formulations, well-understood failure modes, and validation in both simulation and on real robots. The research lies at the intersection of model-based planning and control and learning-based models. Depending on your background, you may work with optimal control, model predictive control, motion planning, probabilistic modeling and estimation, or learning-based approaches. You will also analyze guarantees and failure modes, and contribute to OPERA tasks on GPAI-powered planning and decision-making, long-term reasoning in human-centered environments, and adaptive human-robot and robot-environment interaction. As part of the project, you will collaborate with partners across Europe on simulation, perception, and the OPERA open-source toolbox. The role is located in Delft and involves participation in an international research environment with close links to broader general-purpose AI for robotics activities.