Robots trained on 1M hours of human video reach 90% task success

Robots trained on 1M hours of human video reach 90% task success

Dyna Robotics has unveiled a new robot foundation model trained on more than 1 million hours of human video, a scale the company says could help overcome one of the biggest challenges in teaching robots to handle physical tasks. The Redwood City, California-based company said its DYNA-2 World-Action Model was trained entirely on human egocentric video rather than robot action data. The dataset represents roughly 170 years of continuous waking ex…