Search papers, labs, and topics across Lattice.
1
0
3
Pre-training on universal 3D poses lets robots learn new tasks from just 100 demonstrations, sidestepping the usual VLA efficiency bottleneck.
