Search papers, labs, and topics across Lattice.
3
0
3
Treating 3D Gaussian reconstruction as an active structural prior rather than a passive auxiliary target drives a 22% performance leap in category-level 6D pose estimation.
PANY outperforms existing model-free methods by over 20% in pose accuracy, even in challenging conditions with limited reference overlap.
Forget fine-tuning: DM0 shows that pretraining a VLA model from scratch on diverse embodied and non-embodied data leads to SOTA performance in physical AI tasks.
