Search papers, labs, and topics across Lattice.
3
0
2
A simple reasoning recipe unlocks massive zero-shot gains in robot foundation models, surpassing the benefits of larger models or more data.
A single JEPA latent space unifies state representation, inverse dynamics, and policy prediction, outperforming prior reconstruction-based world action models in state alignment and achieving competitive closed-loop control.
Train high-speed, event-camera-guided quadrotors without expensive event rendering by decoupling representation learning from policy learning.
