Search papers, labs, and topics across Lattice.
1
0
3
Visuomotor policies can learn to ignore distracting visual variations simply by preprocessing raw RGB images into task-aware, semantic-geometric representations *before* feeding them to the policy.
