Train in simulation. Survive the real world.
Simulation is where embodied AI learns; reality is where it's judged. This group works the learning side of Physical AI — reinforcement and imitation learning, domain randomization, and the emerging embodied foundation models — with a relentless focus on policies that transfer.
How this group works.
The methods and commitments that define the lab's approach to the problem.
RL + imitation
Learning control from reward and from demonstration — and knowing when each is the right tool.
Crossing the gap
Domain randomization, system identification, and calibrated twins that make a policy trained in sim work on metal.
Generalist policies
Vision-language-action models that bring broad priors to control — and the open question of how far they generalize.
Scale that transfers
The collection, curation, and replay pipelines that turn fleet experience into training signal.
Open problems we're pursuing.
The questions the lab is taking on now — each a gap between what works in a demo and what works in the world.
What actually transfers
Which simulation fidelity matters for transfer, and which is wasted compute — measured, not assumed.
Imitation at scale
Learning from large, messy demonstration sets without a human in the reward loop.
Generalist policies for flight
Adapting vision-language-action models to the dynamics and safety envelope of real airframes.
Lead this group, or join it.
Step into the principal-investigator role through the Physical AI Investigator Program — an independent appointment with PI authority — or join the group as a research intern.
Physical AI Investigator Program
An independent investigator appointment and PI authority to lead a research line in this group — built to position you for early-career funding.
Research in this group can spin out into a company — explore Launch →