Search papers, labs, and topics across Lattice.
M trajectories from 100 isomorphic robots, covering 200 tasks and 87 atomic skills. (2) InternData-A1 (Sim):
3
0
7
Pre-training on universal 3D poses lets robots learn new tasks from just 100 demonstrations, sidestepping the usual VLA efficiency bottleneck.
Iteratively prompting a graph neural network at test time to amplify out-of-distribution signals dramatically improves OOD detection accuracy.
Achieve 90%+ accuracy in reusing enterprise workflows by breaking down platform-specific DSLs into standardized, modular components that can be intelligently reassembled.
