Search papers, labs, and topics across Lattice.
EmbodiedSmith addresses the data scaling bottleneck for embodied AI by introducing a recursive self-improvement (RSI) flywheel that jointly refines scene and task generation in simulation. This agentic loop allows scene generation to anticipate task needs and task generation to guide scene edits, leading to iterative improvements in both and enabling the creation of more diverse and complex embodied data. The framework supports advanced embodiments and physical interactions, demonstrating significant improvements in task generation success and downstream policy generalization.
A simulation flywheel recursively bootstraps embodied data, generating richer scenes and tasks for robot pretraining and evaluation.
Scaling robotic foundation models requires diverse training data and reliable evaluation environments. Simulation offers a scalable solution, yet existing generation pipelines remain constrained by predefined assets and skills, a disconnect between scene generation and task generation, and limited support for complex embodiments and physics. We introduce EmbodiedSmith, a framework for scalable embodied data generation through recursive self-improvement (RSI). EmbodiedSmith unifies asset, scene, and task generation in a pipeline that supports autonomous creation and language-driven customization. Its core is an agentic refinement loop: scene generation anticipates downstream task requirements, while task generation guides targeted scene edits, allowing scenes and tasks to iteratively improve one another. This joint refinement improves task generation success, including for long-horizon tasks. The framework further supports mobile manipulators, humanoids, and dexterous hands, as well as interactions involving deformable objects and fluids, broadening the range of behaviors and physical phenomena represented in generated data. Together, these capabilities provide a flexible simulation engine for both robot pretraining and evaluation. Extensive experiments validate the quality, diversity, and generation efficiency of the resulting data, while downstream policy experiments demonstrate that increased data diversity improves generalization.
