---
title: "A Self-Evolving Agentic System for Automated Generation and Execution of Biological Protocols"
canonical_url: "https://www.modelscope.cn/papers/305155"
md_url: "https://www.modelscope.cn/papers/305155.md"
arxiv_id: 2606.31763
published: 2026-06-30
last_updated: 2026-06-30
authors:
  - "Yankai Jiang"
  - "Weiting Tang"
  - "Haoran Sun"
  - "Zhenyu Tang"
  - "Yuejie Hou"
  - "Yingnan Han"
  - "Rubo Wang"
  - "Yueyuxiao Yang"
  - "Cheng Liang"
  - "Lilong Wang"
  - "Wenjie Lou"
  - "Xiaosong Wang"
  - "Lei Bai"
  - "Meng Yang"
model_name: ProtoPilot
model_developer: "上海人工智能实验室、Genoria AI"
domain:
  - "人工智能"
  - "智能体系统"
  - "生物实验自动化"
  - "合成生物学"
  - "科学计算"
type:
  - "人工智能"
  - "智能体系统"
  - "生物实验自动化"
  - "合成生物学"
  - "科学计算"
  - "Artificial Intelligence"
arxiv_url: "https://arxiv.org/abs/2606.31763"
pdf_url: "https://arxiv.org/pdf/2606.31763.pdf"
---

# A Self-Evolving Agentic System for Automated Generation and Execution of Biological Protocols

> Autonomous wet-lab experimentation requires more than plausible protocol text: biological intent, quantitative procedures, device constraints and experimental feedback must remain aligned from protocol and SOP design to code and physical execution. We…

「A Self-Evolving Agentic System for Automated Generation and Execution of Biological Protocols」是 ModelScope 魔搭社区收录的论文，arXiv 2606.31763，作者为 Yankai Jiang, Weiting Tang, Haoran Sun et al.，发表于 2026-06-30，属于 人工智能、智能体系统、生物实验自动化 领域。

- **ArXiv**: 2606.31763
- **Published**: 2026-06-30
- **Authors**: Yankai Jiang, Weiting Tang, Haoran Sun, Zhenyu Tang, Yuejie Hou, Yingnan Han, Rubo Wang, Yueyuxiao Yang, Cheng Liang, Lilong Wang, Wenjie Lou, Xiaosong Wang, Lei Bai, Meng Yang
- **Model**: ProtoPilot
- **Developer**: 上海人工智能实验室、Genoria AI
- **Domain**: 人工智能, 智能体系统, 生物实验自动化, 合成生物学, 科学计算
- **ArXiv URL**: https://arxiv.org/abs/2606.31763
- **PDF**: https://arxiv.org/pdf/2606.31763.pdf

Source: https://www.modelscope.cn/papers/305155

---

> ProtoPilot: 面向生物实验自动化生成的自进化多智能体系统

## 摘要

ProtoPilot是一个自进化的多智能体系统，旨在实现从自然语言生物意图到物理湿实验执行的端到端自动化。该系统通过分层智能体协作、运行时技能学习和闭环反馈机制，将实验目标转化为标准操作规程（SOP）、仪器可执行代码及物理工作流，并利用执行结果指导迭代优化。论文同时提出了专家基准BioLab Bench，包含294个合成生物学与分子生物学任务，用于评估协议生成、代码翻译、跨平台机器人执行及湿实验验证能力。

## Abstract

Autonomous wet-lab experimentation requires more than plausible protocol text: biological intent, quantitative procedures, device constraints and experimental feedback must remain aligned from protocol and SOP design to code and physical execution. We developed ProtoPilot, a self-evolving multi-agent system, together with an expert-grounded benchmark and evaluation framework for testing this conversion as an experimental automation problem. The framework spans 294 synthetic-biology and molecular-biology tasks derived from 98 gold-standard protocols, wet-lab expert rubrics, device-level validity gates and real experimental tests. ProtoPilot incorporates layer-wise verifiability, multi-agent orchestration and a runtime-updated skill library to generate protocols, expand SOPs, synthesize SDK-compliant code and revise workflows from wet-lab feedback. It achieved a Top@3 expert-preference rate of 90.2%, an overall protocol-to-code gate pass rate of 89.5% and an Opentrons pass rate of 88.24%, compared with 32.35% for OpenTrons-AI. Wet-lab validation produced interpretable readouts, Sanger-confirmed products and feedback-corrected PCA-assembled DNA targets, establishing a verifiable route to autonomous experimentation. Together, these results show that the evaluation framework captures execution-relevant requirements for autonomous wet-lab automation, and that ProtoPilot can meet them by converting protocol and code generation into validated execution and feedback-guided revision.