Search papers, labs, and topics across Lattice.
2
0
0
0
A novel recursive memory mechanism learns to retain only action-relevant history, achieving state-of-the-art VLA performance on long-horizon tasks with a 64-token budget.
High-level robot planners routinely fail because they treat low-level skills as opaque black boxes; turning training-time inductive biases into explicit compositional interfaces bridges the gap to robust out-of-distribution execution.