Search papers, labs, and topics across Lattice.
1
0
3
Text, combined with learned image embeddings, can compress maps by 2x while preserving localization accuracy, offering a practical solution to the growing memory demands of robotic mapping.
