Search papers, labs, and topics across Lattice.
1
0
3
2
Current physical AI models struggle not with visual recognition, but with understanding space, physics, and action – PRISM, a new retail video dataset, dramatically improves embodied reasoning by focusing on these crucial elements.