Search papers, labs, and topics across Lattice.
3
0
3
10
Robot-centric pointmaps can drastically improve action prediction in VLA models, especially when faced with diverse camera viewpoints.
3D HAMSTER achieves superior robotic manipulation performance by directly predicting 3D trajectories, eliminating the distortions caused by 2D guidance.
Off-policy RL can now train in minutes instead of hours for sim-to-real robot control, thanks to a new algorithm that borrows scaling laws from supervised learning.
