NVIDIA Advances Robotics with New AI Training Models
5
2
BTCCSquare news:
NVIDIA has unveiled groundbreaking advancements in robotics training, leveraging generative AI and world foundation models (WFMs) to address the challenge of efficient robot training without extensive data collection. The Cosmos WFMs process millions of hours of real-world data to predict future world states, enabling robots and autonomous vehicles to anticipate events and accelerate high-fidelity training data creation.
DreamGen, a synthetic data generation pipeline, reduces the need for costly human teleoperation data by using WFMs to produce diverse, realistic training datasets with minimal human intervention. This innovation promises scalable robot learning and faster adaptation to new environments.
By:
Previous
BlackRock’s Rieder Predicts September Rate Cut as Inflation Moderates
Next
US Targets Chinese Tech in Undersea Internet Infrastructure Amid Security Concerns
