Our theses topics cover the following multi-disciplinary areas, in accordance with the research vision of the iROSA group.
- Robotic Perception for manipulation and interaction;
- Robot learning for embodied agents like mobile manipulation;
- Graph Neural Network representations for Robot Learning
- Long-horizon reasoning, with a focus on combining Graph-based Reinforcement Learning and Planning.
- Uncertainty-aware robot learning for (mobile) manipulation
- Uncertainty-aware model-based reinforcement learning
- Neural representations for (safe) motion planning
- Human behavior understanding, and uncertainty quantification;
- Human-Robot Interaction;