Search papers, labs, and topics across Lattice.
Affiliation:
3
0
4
Simply handing VLMs a 360掳 field of view fails to improve embodied navigation unless paired with dynamic multi-step action planning and explicit geometric representation, which together yield an 11.9% jump over SOTA.
Current instruction-based video editing models are far from satisfactory, revealing critical gaps in evaluation that could reshape the field.
