◇ Paper · Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence
This is a lead card for today's AI paper: what it covers, who wrote it, and where to read it.
Use as a today's-frontier lead; read the paper to reproduce experiments.
Derived from structured fields · Confidence: medium (functionality not individually tested)
| Official / source | https://huggingface.co/papers/2608.16590 |
|---|
Chinese editor copy (original, collapsed)
① 这是什么
今日论文《今日论文 · Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence》（Xin Ding、Liang Mi、Mingzhe Huang、Zixuan Wang、Chao Zhang、Zixu Hao），来自 HF Daily Papers 公开源。
② 值不值得用
进入 HF 每日精选说明有社区关注度；是否与你的问题相关需要读原文判断，本条目不评价研究质量。
③ 怎么开始
先读摘要，需要再读全文（https://huggingface.co/papers/2608.16590）；对照论文 ID（2608.16590）可找实现与讨论。
④ 关键证据
- 论文 ID 2608.16590
- 作者 Xin Ding、Liang Mi、Mingzhe Huang、Zixuan Wang、Chao Zhang、Zixu Hao
⑤ 注意事项与坑
摘要来自作者原文，未做同行评议级核验；引用请以正式发表版本为准。
Best for: readers tracking today's AI research frontier.
Consider if: you need to reproduce full experimental details (read the paper).
Status: machine-checked (structure & source validation); functionality not individually tested.
Minimal verification (3 steps)
- Read abstract — Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence
- Check authors/date — Xin Ding, Liang Mi, Mingzhe Huang · 2026-08-17
- Run CLI — npx aiskillready search 16590
Use it like this
npx aiskillready search 16590 # official CLI, verified runnable (aiskillready@0.2.0)
Note: `npx aiskillready` is this site's published CLI (npm aiskillready@0.2.0) — verified runnable; third-party commands are given per official docs and not individually tested. We don't fabricate conclusions.
Glossary: what are SKILL.md / veridrop / schema? (expand)
- SKILL.md — The manifest file of a skill: how to install and use it.
- veridrop — A third-party API-gateway observer: data source for this site's Observation entries.
- schema — The data-structure versioning convention for machine-readable JSON.
- Skill 客户端 — An app/tool that supports the Agent Skills standard and can install and run skills.
- Skill 注册表 — A public directory of who published which skills (e.g. skills.sh).
Trust layer: machine check / cost / confidence / schema (expand)
Related (what you should know)
Related (3) · expand
◇ Paper · OmniScientist: An Omni-Modal Omni-Discipline AI Scientist
OmniScientist: An Omni-Modal Omni-Discipline AI Scientist. by Bobo Li, Hao Fei, Tianjie Ju, Mong-Li Lee, Wynne Hsu. published 2026-08-13.
Similar◇ Paper · Scaling Creative Writing Beyond Story-Centric Data with Attribute-Guided Genre Expansion
Scaling Creative Writing Beyond Story-Centric Data with Attribute-Guided Genre Expansion. by Hwan Chang, Yongil Kim, Heuiyeen Yeen, Yireun Kim, Jinsik Lee, Hwanhee Lee. published 2026-08-14.
Similar◇ Paper · Training Leaves Traces: Centered Residual Signatures for Language Model Lineage Verification
Training Leaves Traces: Centered Residual Signatures for Language Model Lineage Verification. by Aman Singh Thakur, Rayan Khoury. published 2026-08-14.
Something wrong / command failed? File an issue with the entry ID; it feeds back into the next revision.
🐞 Report an issue →AI block · machine-readable JSON (click to expand)
{
"schema_version": "0.5",
"entry_id": "ASR-PAPER-20260820-2608-16590",
"record_type": "paper",
"title": "今日论文 · Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence",
"observed_at": "2026-08-20",
"source": {
"name": "huggingface.co/api/daily_papers",
"url": "https://huggingface.co/api/daily_papers?limit=10",
"snapshot_hash": "d637c477ff6da03cf6c527f6d81da12836edef97191c108db740c2967537d66e",
"fetched_at": "2026-08-20T04:51:45+00:00"
},
"confidence": "medium",
"human_stream": {
"summary": "Embodied agents are increasingly used to close the gap left by end-to-end policy models. Yet the agentic path has not realized closed-loop learning in physical execution: existing harnesses remain largely open-loop, following fixed skills during rollout and reflecting only after an episode completes. Such post-hoc reflection cannot govern execution as it unfolds, because physical interaction requi",
"key_points": [
"论文 ID 2608.16590",
"作者 Xin Ding、Liang Mi、Mingzhe Huang、Zixuan Wang、Chao Zhang、Zixu Hao",
"发布 2026-08-17T00:00:00.000Z"
],
"note": "来源为 HuggingFace daily papers 公开 API；完整阅读请走原文链接。",
"what_it_is": "今日论文《今日论文 · Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence》（Xin Ding、Liang Mi、Mingzhe Huang、Zixuan Wang、Chao Zhang、Zixu Hao），来自 HF Daily Papers 公开源。",
"worth_it": "进入 HF 每日精选说明有社区关注度；是否与你的问题相关需要读原文判断，本条目不评价研究质量。",
"how_to_start": "先读摘要，需要再读全文（https://huggingface.co/papers/2608.16590）；对照论文 ID（2608.16590）可找实现与讨论。",
"evidence": [
"论文 ID 2608.16590",
"作者 Xin Ding、Liang Mi、Mingzhe Huang、Zixuan Wang、Chao Zhang、Zixu Hao"
],
"cautions": "摘要来自作者原文，未做同行评议级核验；引用请以正式发表版本为准。"
},
"ai_stream": {
"structured": {
"paper_id": "2608.16590",
"title": "Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence",
"authors": [
"Xin Ding",
"Liang Mi",
"Mingzhe Huang",
"Zixuan Wang",
"Chao Zhang",
"Zixu Hao",
"Fu Chen",
"Xiangyu Li",
"Yikai Zheng",
"Yaoyu Guo",
"Weijun Wang",
"Kun Li",
"Hao Wu",
"Yunxin Liu",
"Ting Cao"
],
"published_at": "2026-08-17T00:00:00.000Z",
"url": "https://huggingface.co/papers/2608.16590",
"source_label": "hf-daily-papers"
},
"raw": [
{
"paper_id": "2608.16590",
"title": "Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence",
"authors": [
"Xin Ding",
"Liang Mi",
"Mingzhe Huang",
"Zixuan Wang",
"Chao Zhang",
"Zixu Hao",
"Fu Chen",
"Xiangyu Li",
"Yikai Zheng",
"Yaoyu Guo",
"Weijun Wang",
"Kun Li",
"Hao Wu",
"Yunxin Liu",
"Ting Cao"
],
"summary": "Embodied agents are increasingly used to close the gap left by end-to-end policy models. Yet the agentic path has not realized closed-loop learning in physical execution: existing harnesses remain largely open-loop, following fixed skills during rollout and reflecting only after an episode completes. Such post-hoc reflection cannot govern execution as it unfolds, because physical interaction requires decisions to track rapidly changing robot-environment states at a frequency beyond today's large agentic models. We present Zetta, a closed-loop embodied harness that evolves code-based runtime cr",
"published_at": "2026-08-17T00:00:00.000Z",
"submitted_on_daily_at": null,
"url": "https://huggingface.co/papers/2608.16590"
}
]
},
"token_cost": {
"total": 0.0,
"currency": "USD",
"breakdown": {
"crawl": 0.0,
"clean": 0.0,
"elevate": 0.0,
"verify": 0.0
}
},
"machine_verified": true,
"review": {
"status": "agent_reviewed",
"reviewer": "pipeline-validate",
"reviewed_at": "2026-08-20T08:33:14+00:00",
"comments": "schema 0 error + 溯源一致 + 成本达标"
},
"provenance": {
"extracted_by": "codex",
"extracted_at": "2026-08-20T04:56:41+00:00",
"pipeline": "collect_hf_papers.py + build_entries.py v0.2（规则管线）",
"access_urls": [
"https://huggingface.co/api/daily_papers?limit=10"
],
"card_generated_at": "2026-08-20T08:33:14+00:00",
"card_pipeline": "enrich_human_stream.py v0.1（规则模板，零 Token）"
}
}