C. Karen Liu

Pretraining through play can revolutionize how robots learn dexterous assembly, achieving 60% success in tight insertions with minimal contact clearance.

Cloak enables VLA models to seamlessly adapt to new robotic embodiments without any additional training data, revolutionizing the way we think about robotic adaptability.

Zero-shot sim-to-real transfer for articulated tool manipulation is now achievable with just a few clicks, revolutionizing how robots interact with complex objects.

LadderMan enables humanoid robots to climb ladders and manipulate objects with unprecedented robustness and adaptability in real-world scenarios.

AnyLift outperforms existing methods by accurately reconstructing 3D human motion and interactions from challenging Internet videos, including gymnastics and in-the-wild scenarios.

Humanoid robots can now navigate complex environments zero-shot, learning directly from human demonstrations without needing any robot-specific training.

By jointly modeling hand-object motion, WHOLE overcomes occlusion and out-of-sight challenges in egocentric videos, achieving state-of-the-art results in hand and object pose estimation.

A single RL policy trained on procedurally generated tools in simulation can achieve zero-shot dexterous manipulation of diverse real-world tools, rivaling task-specific policies.

Humanoid robots can now perform vision-based parkour, chaining together dynamic skills like climbing, vaulting, and rolling, adapting to real-time obstacle changes.