Open X-Embodiment
A large-scale robot demonstration dataset that aggregates millions of motion trajectories from more than 22 different robot platforms. It is designed for training generalist robot control models across diverse embodiments and tasks.
Core Facts
Timeline (last 90 days)
OpenVLA uses an open-source vision-language model with about 7 billion parameters as its backbone and is fine-tuned on the Open X-Embodiment dataset, with downloadable weights provided.
Previously influential robot datasets, such as Google's RT series and the Open X-Embodiment project, typically range in scale from hundreds of thousands to millions of trajectories.
Open X-Embodiment aggregates over 1 million manipulation trajectories from 22 different robot embodiments.
All Facts (3)
Open X-Embodiment aggregates over 1 million manipulation trajectories from 22 different robot embodiments.
65%UnverifiedOpenVLA uses an open-source vision-language model with about 7 billion parameters as its backbone and is fine-tuned on the Open X-Embodiment dataset, with downloadable weights provided.
50%UnverifiedPreviously influential robot datasets, such as Google's RT series and the Open X-Embodiment project, typically range in scale from hundreds of thousands to millions of trajectories.
50%