Search papers, labs, and topics across Lattice.
12
0
9
4
Geometry-aware reconstruction reduces hallucinations in object poses, leading to more reliable robot planning.
Surpassing existing benchmarks, SpatialAvatar-0 achieves superior 4D head avatar quality with up to 60x fewer training iterations than traditional methods.
A dual-stream tokenizer reveals that traditional methods misrepresent high-frequency motion dynamics, leading to over 50% improvement in diversity alignment with real data.
DragMesh-2 achieves unprecedented robustness in dexterous manipulation of articulated objects, outperforming traditional methods even under varying contact loads.
State-of-the-art generative models struggle to maintain physical consistency and coherent interactions over time, revealing critical gaps in their world modeling capabilities.
Latent spatial memory can accelerate video generation by over 10 times while dramatically reducing memory usage, revolutionizing how we model dynamic scenes.
By explicitly enforcing action-conditioned consistency during training and distillation, MWM enables more reliable planning in imagined future spaces for embodied navigation.
Generative video models can now simulate a continuously evolving world, even when objects are out of sight, thanks to a new framework that maintains persistent global state.
RAG agents can be tricked by a "Visual Placebo Effect" where they inherit latent visual biases from foundation models, but a new memory weighting scheme can help them abstain when evidence is weak.
Coding LLMs can now generate more physically plausible and dynamically rich 4D worlds, thanks to a novel closed-loop framework that iteratively refines simulation code based on physics-aware self-reflection.
Achieve zero-shot robotic manipulation by guiding a hierarchical vision-language-action model with knowledge-guided trajectory planning, outperforming existing methods and eliminating the need for real-world data collection.
