DynaRobotics podcast: scaling robot pretraining toward 1 million hours of data
chris_j_paxton · x · 2026-08-19
RoboPapers hosted the DynaRobotics team (Jason Ma, tianyurobot and others) for a deep podcast on scaling robot-learning pretraining to 1 million hours of data.
- Excitement around world-action models promises an eventual "GPT moment" for robotics via massive egocentric human data, yet concrete scaling evidence has been scarce.
- DynaRobotics presents a systematic scaling analysis approaching the 1M-hour mark.
Key finding: as pretraining data grows, they observe predictable, statistically significant gains in accuracy metrics. Listeners called it "4-5 papers squeezed into one podcast."
More from Embodied
- EmbodiedEvo tops RoboDojo as agents inherit reusable skills and lessons — Aiden_Tech_Ai · 2026-10-16
- Sunday Robotics is hiring a Senior Infra PM to scale consumer robot deployments — saranormous · 2026-10-16
- Bond's Submarine Lotus Esprit Spotted at Event, Fueling Steering-Wheel-Free Robotaxi Speculation — PTrubey · 2026-10-16
- Maker wires Grok and Claude into ROS2 robot with RealSense stereo cameras — chrismatthieu · 2026-10-16
- NEURA Robotics launches Neuron AI, a unified Perception-Language-Action model for robots — Distinct-Question-16 · 2026-10-16
- Dev demos open wearable AI passport running multilingual flashcards on tiny memory — ezshine · 2026-10-16