"OpenBrain pre-training worked because they had an entire internet's worth of text data. But there's no "internet of robotics": the largest action dataset for robots is less than 0.01% of OpenBrain's LLM dataset. OpenBrain's post-training works by reinforcement learning, where the model attempts tasks like complex math problems billions of times and learns from its successes and failures. But for robots, the real world is too slow to get billions of interactions. All robotics research efforts are now about enabling robots to scale both pre-training"
"Some companies like Waytek go all-in on teleoperation as a result. From 2023 to 2025, Waytek tries to solve the pre-training bottleneck by brute-forcing through teleoperation: they collect data of humans controlling robots, then train a Vision-Language-Action model (VLA), with an architecture similar to LLMs, to imitate that data. Early signs show this work in demos with high reliability, like laundry, making sandwiches, folding shirts, and sorting packages."
Robotics is emerging as the next exponential AI domain propelled by research breakthroughs, major capital inflows, and geopolitical incentives tied to manufacturing. LLM success motivated large robotics investments, but robotics lacks massive pre-training corpora and cannot replicate billions of real-world interactions for reinforcement learning. The largest robot action dataset is under 0.01% of typical LLM datasets, creating both pre-training and post-training bottlenecks. Teams pursue strategies to scale data and interactions; one approach uses teleoperation to collect human-controlled robot demonstrations and trains Vision-Language-Action models that already succeed at many household and logistics tasks in demos.
Read at Android-dreams
Unable to calculate read time
Collection
[
|
...
]