Search papers, labs, and topics across Lattice.
3
0
5
5
Dexterous teleoperation can now achieve high-fidelity hand-state estimation with just a few calibration poses, bypassing per-user model retraining.
A single generative model recovers temporally consistent 3D hand motion from video by jointly learning 2D-3D pose dynamics, bypassing unreliable per-frame estimates.
LLMs can now generate more accurate and complex CAD models by pointing to existing geometric entities, rather than relying on discretized command sequences prone to topological errors.
