Search papers, labs, and topics across Lattice.
4
0
3
Primitive steerability in VLAs allows for autonomous skill acquisition, enabling robots to learn new tasks without human demonstrations.
Learning-based warm-starts can cut SCP runtime by nearly half while maintaining trajectory robustness, revolutionizing on-orbit robotic servicing.
High-quality dense rewards can elevate robotic manipulation success rates from 50% to near perfection, transforming how robots learn from their environments.
Training VLA policies without human demos is now feasible, with LEGS achieving better performance than traditional methods at a fraction of the cost.
