Topic hub
Physical AI & Robotics
Robot training data, teleoperation and egocentric collection, dataset formats, simulation, and policy training for machines that act in the world.
All posts · 3
PHYSICAL AI & ROBOTICS01
How Much Data Do You Need to Train a Robot Policy?The scaling-law study that ran 15,000 real rollouts found demonstrations per scene stop helping fast. Diversity is the axis that pays. Sizing tables inside.
PHYSICAL AI & ROBOTICS02
LeRobot vs RLDS vs HDF5: Which Robot Data Format to UseThe same robot dataset measured 387 GB in RLDS, 779 GB in HDF5, and 16 GB in LeRobot. What each format stores, what breaks at scale, and how to convert.
PHYSICAL AI & ROBOTICS03
Teleoperation vs Simulation vs Human Video for Robot DataOne published result: an added hour of human hand video beat an added hour of robot teleop data. Cost per usable episode and how to split a budget.