We make robots that stay safe and useful over time.
We study how robots learn, plan, and act reliably, individually and in teams.

Our research focuses on:

• multi-robot systems under uncertainty
• data-efficient embodied learning and manipulation
• learning and inference methods for physical systems

We are interested in approaches that combine learning, probabilistic reasoning, and real-world experimentation, with an emphasis on efficiency, robustness, and physical grounding.

The embodied intelligence and collective robotics lab is based at UNSW and led by Oliver Obst. We expect to host PhD projects aligned with these research themes and welcome informal enquiries from prospective students.