Robotics · 2026-10-07
EmbodiedSmith: Scaling Embodied Data through Recursive Self-Improvement Flywheel in Simulation
Yikai Qin, Yifei Deng, Mingjian Liang, Wenxuan Song, Zepeng Lin, Zhiyi Jiang, Jiajun Fu, Qiao Sun, Huashuo Lei, Xicheng Gong, Jiayi Chen, Han Zhao, Shuanghao Bai, Pengxiang Ding, Pengwei Wang, Haoang Li
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
EmbodiedSmith is a simulation pipeline for scalable embodied data generation via recursive self-improvement. It unifies asset, scene, and task generation, using a loop where scene generation anticipates task requirements and task generation drives targeted scene edits. The authors report higher scene quality metrics and higher demonstration success rates than prior scene-generation methods.
Why it matters
The paper targets common breaks in simulation pipelines: decoupled scene vs task feasibility, limited scene/asset diversity and customizability, limited physics realism, and limited support for complex embodiments. EmbodiedSmith’s RSI loop explicitly couples scene generation and executable task generation/validation, aiming to raise long-horizon task success and produce simulation-ready data for both pretraining and evaluation.
Method
- Unifies asset, scene, and task generation in an RSI pipeline; scene generation anticipates downstream task requirements, while task generation revises scenes through targeted edits.
- Builds task-conditioned simulation scenes with asset retrieval/generation plus reachability-validated layout optimization, then triggers scene edit + revalidation on execution failures.
- Generates executable robot demonstrations by grounding tasks to scene entities and atomic skills as a Skill DAG, validated via simulation checks and an independent VLM review; supports humanoids and dexterous hands via embodiment adapters.
Limitation
Incomplete post-release recovery is documented as a known limitation rather than an acceptance criterion for the evaluated deformable configuration.
Abstract (from arXiv)
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.
Related papers
- Awomo-SimDataEngine: Agentic Simulation-ReadyWorld Generation
- AutoRSR: Automated Real-to-Sim-to-Real Data Generation for Scalable Robot Learning
- SkillWeaver: Agentic Exploration over Neural Interaction Skills for Scalable Robot Data Generation
- AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents
- AffordCraft: Scalable Construction of Task-Ready Simulation Assets from Single Images
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-06 · Indexed by Pub-AI: 2026-10-07 UTC
EmbodiedSmith: Scaling Embodied Data through Recursive Self-Improvement Flywheel in Simulation
Preprint posted: 2026-10-06 · Original record ↗
- arxiv
- 2610.07969
Explicit source versions (1)
- v1 · 2026-10-06
- Preprint available
Source date: 2026-10-06 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
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.