[{"data":1,"prerenderedAt":234},["ShallowReactive",2],{"content-query-MPzQKkQnk2":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":228,"_id":229,"_source":230,"_file":231,"_stem":232,"_extension":233},"/news/zh/3668","zh",false,"","化学研究全自动化！昇思助力中科大团队打造多智能体驱动的机器人AI化学家","化学实验室的未来正在发生翻天覆地的变化，而这一变化的核心驱动力之一便是大语言模型（LLM）和机器人技术的结合。","2025-04-01","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/04/03/1ed52cb3fb4c43499444d392406d49f4.png","news",{"type":14,"children":15,"toc":225},"root",[16,24,30,35,46,54,62,70,78,83,88,96,101,108,113,121,126,131,138,143,151,156,164,169,176,181,189,197,202,207,212,220],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"化学研究全自动化昇思助力中科大团队打造多智能体驱动的机器人ai化学家",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":23,"value":29},"化学实验室的未来正在发生翻天覆地的变化，而这一变化的核心驱动力之一便是大语言模型（LLM）和机器人技术的结合。近年来，在化学和材料科学领域，LLM的应用正在加速科学研究的进程。尽管LLM在化学研究中展现了巨大的潜力，现有的系统大多局限于单一任务或简单实验，难以应对复杂的多步骤、多站点实验。",{"type":17,"tag":25,"props":31,"children":32},{},[33],{"type":23,"value":34},"中国科学技术大学罗毅&陈林江&张飞&尚伟伟&傅尧&江俊团队在《J. Am. Chem. Soc.》报道了一种名为ChemAgents的多智能体系统，结合了LLM和自动化机器人技术，能够自主执行复杂的化学实验，极大地减少了人为干预。",{"type":17,"tag":25,"props":36,"children":37},{},[38,40],{"type":23,"value":39},"这项研究的核心目标是解决如何在复杂的实验场景中有效利用LLM，构建一个能够按需执行化学任务的自动化系统。",{"type":17,"tag":41,"props":42,"children":43},"strong",{},[44],{"type":23,"value":45},"通过多智能体的协作，ChemAgents不仅能够自主设计实验、执行实验，还能通过机器学习模型进行数据分析和优化，最终实现化学研究的全自动化。",{"type":17,"tag":25,"props":47,"children":48},{},[49],{"type":17,"tag":41,"props":50,"children":51},{},[52],{"type":23,"value":53},"该研究使用的大语言模型是基于昇思MindSpore的Llama-3.1-70B模型，目前已在MindSpore Transformers开源。",{"type":17,"tag":25,"props":55,"children":56},{},[57],{"type":17,"tag":41,"props":58,"children":59},{},[60],{"type":23,"value":61},"# 01",{"type":17,"tag":25,"props":63,"children":64},{},[65],{"type":17,"tag":41,"props":66,"children":67},{},[68],{"type":23,"value":69},"多智能体系统介绍：",{"type":17,"tag":25,"props":71,"children":72},{},[73],{"type":17,"tag":41,"props":74,"children":75},{},[76],{"type":23,"value":77},"1、多智能体系统：ChemAgents的架构",{"type":17,"tag":25,"props":79,"children":80},{},[81],{"type":23,"value":82},"ChemAgents的核心是一个基于Llama-3.1-70B大语言模型的多智能体系统，其中含有一个“任务管理”智能体，并对“文献阅读”“实验设计”“机器操控”和“计算执行”等另外四个智能体进行管理。这些智能体能够接受文字指令进行相应的操作，随后输出可解释的结果。四个智能体分别通过文献数据挖掘与知识整合、实验流程生成与逻辑验证、实验指令到机器人代码的转化与执行和模型预测与迭代优化等步骤，实现自动化化学研究。此外，ChemAgents背后还拥有四个基础资源库：科学文献数据库、实验方案文库、机器学习模型(已预训练)文库以及化学实验室自动工作站。",{"type":17,"tag":25,"props":84,"children":85},{},[86],{"type":23,"value":87},"这种多智能体驱动的AI机器化学家可以解决以下三个复杂度逐渐提高的问题：(1)合成或表征所需化合物；(2)拓展或筛选所需功能材料的实验参数；(3)通过挖掘文献、计算建模和闭环优化，发现功能材料。",{"type":17,"tag":25,"props":89,"children":90},{},[91],{"type":17,"tag":92,"props":93,"children":95},"img",{"alt":7,"src":94},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/04/03/54fc680b75324f9a9162bcdfdfcc546d.png",[],{"type":17,"tag":25,"props":97,"children":98},{},[99],{"type":23,"value":100},"图1 ChemAgents的多智能体系统架构（展示了ChemAgents的四个角色特定智能体如何通过任务管理器进行协作，完成复杂的化学研究任务）",{"type":17,"tag":25,"props":102,"children":103},{},[104],{"type":17,"tag":92,"props":105,"children":107},{"alt":7,"src":106},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/04/03/dc2aa3d791a342ff89610615ed3abb39.png",[],{"type":17,"tag":25,"props":109,"children":110},{},[111],{"type":23,"value":112},"图2 各智能体的详细架构（展示了文献阅读器、实验设计器、机器人操作员和计算执行器的内部架构，详细说明了每个智能体如何利用外部工具和资源完成任务）",{"type":17,"tag":25,"props":114,"children":115},{},[116],{"type":17,"tag":41,"props":117,"children":118},{},[119],{"type":23,"value":120},"2、自动化实验室：机器人与实验站的完美结合",{"type":17,"tag":25,"props":122,"children":123},{},[124],{"type":23,"value":125},"ChemAgents的自动化实验室配备了多种实验设备，包括液体分配器、固体分配器、磁力搅拌器、干燥器、红外光谱仪、X射线衍射仪等。实验室中的机器人能够根据实验设计器的指令，自动执行实验步骤，并实时收集数据。",{"type":17,"tag":25,"props":127,"children":128},{},[129],{"type":23,"value":130},"例如，在金属氧化物合成实验中，机器人成功合成了六种不同的金属氧化物（如ZrO2、ZnO等），并通过X射线衍射（PXRD）进行了表征，结果与标准数据高度吻合。",{"type":17,"tag":25,"props":132,"children":133},{},[134],{"type":17,"tag":92,"props":135,"children":137},{"alt":7,"src":136},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/04/03/83c70a73ce95490c92113fda0e5d8de0.png",[],{"type":17,"tag":25,"props":139,"children":140},{},[141],{"type":23,"value":142},"图3 六项实验任务的结果（展示了ChemAgents在六项不同实验任务中的表现，包括金属氧化物的合成与表征、光催化降解有机污染物、以及金属-有机高熵催化剂的发现与优化）",{"type":17,"tag":25,"props":144,"children":145},{},[146],{"type":17,"tag":41,"props":147,"children":148},{},[149],{"type":23,"value":150},"3、从实验到优化：机器学习的强大助力",{"type":17,"tag":25,"props":152,"children":153},{},[154],{"type":23,"value":155},"ChemAgents不仅能够执行实验，还能通过机器学习模型进行数据驱动的优化。在金属-有机高熵催化剂（MO-HEC）的发现任务中，系统通过文献阅读器确定了五种常见的金属元素（Co、Ni、Fe、Mn、Cu），并通过实验设计器和机器人操作员合成了100种不同比例的MO-HEC。随后，计算执行器利用贝叶斯优化和预训练的机器学习模型，成功找到了性能最优的催化剂，其过电位（overpotential）显著低于随机合成的样品。",{"type":17,"tag":25,"props":157,"children":158},{},[159],{"type":17,"tag":41,"props":160,"children":161},{},[162],{"type":23,"value":163},"4、复杂任务的自主执行：从合成到光催化反应",{"type":17,"tag":25,"props":165,"children":166},{},[167],{"type":23,"value":168},"ChemAgents还展示了其在复杂任务中的自主执行能力。在光催化有机反应任务中，系统成功执行了芳基卤化物的光催化脱卤反应，并通过气相色谱-质谱（GC-MS）分析了反应产物。实验结果显示，反应24小时后，起始材料几乎完全转化为目标产物，验证了系统在有机化学研究中的潜力。",{"type":17,"tag":25,"props":170,"children":171},{},[172],{"type":17,"tag":92,"props":173,"children":175},{"alt":7,"src":174},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2025/04/03/aacf49c9e1fd4ebb9c35e31e0dacf135.png",[],{"type":17,"tag":25,"props":177,"children":178},{},[179],{"type":23,"value":180},"图4 自动化实验室的实验流程（展示了ChemAgents在自动化实验室中执行光催化有机反应的实验流程，包括样品制备、液体转移、光催化反应和产物分析）",{"type":17,"tag":25,"props":182,"children":183},{},[184],{"type":17,"tag":41,"props":185,"children":186},{},[187],{"type":23,"value":188},"# 02",{"type":17,"tag":25,"props":190,"children":191},{},[192],{"type":17,"tag":41,"props":193,"children":194},{},[195],{"type":23,"value":196},"总结与展望",{"type":17,"tag":25,"props":198,"children":199},{},[200],{"type":23,"value":201},"这项研究标志着化学研究向全自动化迈出了重要一步。通过结合大语言模型、多智能体系统和自动化机器人技术，ChemAgents能够自主执行从实验设计到数据分析的全流程任务，极大地提高了化学研究的效率和精度。",{"type":17,"tag":25,"props":203,"children":204},{},[205],{"type":23,"value":206},"ChemAgents的模块化架构使其能够灵活适应不同的研究需求，不仅适用于化学领域，还可以扩展到生物学、材料科学等其他学科。未来，随着更多自动化实验室的建设和LLM技术的进一步发展，ChemAgents有望成为科学研究的“标配”，推动科学发现的加速和民主化。",{"type":17,"tag":25,"props":208,"children":209},{},[210],{"type":23,"value":211},"ChemAgents所使用的基于MindSpore的Llama-3.1-70B模型，目前已在MindSpore Transformers开源，欢迎开发者们使用。",{"type":17,"tag":25,"props":213,"children":214},{},[215],{"type":17,"tag":41,"props":216,"children":217},{},[218],{"type":23,"value":219},"文章链接",{"type":17,"tag":25,"props":221,"children":222},{},[223],{"type":23,"value":224},"Tao Song, Man Luo, Xiaolong Zhang, Linjiang Chen,* Yan Huang, Jiaqi Cao, Qing Zhu, Daobin Liu, Baicheng Zhang, Gang Zou, Guoqing Zhang, Fei Zhang,* Weiwei Shang,* Yao Fu,* Jun Jiang,* and Yi Luo*. A Multiagent-Driven Robotic AI Chemist Enabling Autonomous Chemical Research On Demand. J. Am. Chem. Soc. 2025, DOI: 10.1021/jacs.4c17738",{"title":7,"searchDepth":226,"depth":226,"links":227},4,[],"markdown","content:news:zh:3668.md","content","news/zh/3668.md","news/zh/3668","md",1776506086949]