Generate, curate, evaluate.
Scale diverse robot data through simulation and bridge the sim-to-real gap.
- Generative simulation
- Data quality & diversity
- Sim-to-real transfer
Closing the loop between embodied data, robot foundation models, and real-world deployment.
Scale diverse robot data through simulation and bridge the sim-to-real gap.
Pretrain generalist robot foundation models with heterogeneous data from ego videos, UMI, real robots, and simulation.
Build reliable embodied agents for deployment on real robots.
An end-to-end, GPU-accelerated, and modular platform for building generalized Embodied Intelligence
Developed a scalable, agent-driven robotic data generation pipeline for Sim2Real policy training and generalization.