[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"topic-LLM":3},{"name":4,"kind":5,"tokens":6,"total":7,"page":8,"page_size":9,"items":10},"LLM","tag",[4],4,1,100,[11,67,109,157],{"id":12,"title":13,"url":14,"summary":15,"summary_zh":16,"content":17,"source_name":18,"source_url":16,"published_at":19,"category":20,"cover_url":16,"hotness":21,"is_selected":22,"score":23,"score_detail":24,"sources":31,"tags":33,"primary_topic":38,"secondary_topics":44,"classification":45,"search_phrases":62,"slug":65,"view_count":29,"doi":16,"paper":16,"created_at":66},1123,"Closed-Loop LLM Co-Pilots for Digital Agriculture (arXiv:2608.09949)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.09949","Serge Kernbach 发表论文。本研究评估大语言模型(LLM)在复杂生物系统中的应用,从数据分析演进到AI引导的自主实验。框架由49通道植物传感器网络数据驱动,涵盖多光谱、电化学和介电模态。在最少时间模式下,系统将生产周期缩短35%;在能源优化模式下,通过利用光脉冲的生理惯性,降低能源消耗18%。最后,代理自主开发了黑暗诱导叶绿素积累策略,实现67.9%的节能。本框架将LLM转变为数字农业的自主副驾驶,改善成本-价值比,降低计算和专家劳动力约束。",null,"[View PDF](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2608.09949)[HTML (experimental)](https:\u002F\u002Farxiv.org\u002Fhtml\u002F2608.09949v1)\n\n> Abstract:This study evaluates the application of Large Language Models (LLMs) in complex biological systems, evolving from data analysis to autonomous, AI-guided experimentation. The framework is driven by data from a 49-channel phytosensor network, encompassing multispectral, electrochemical, and dielectric modalities. To enhance accessibility, the system provides real-time natural-language interpretation for both specialists and non-experts. However, its core advantage lies in the transition from human-in-the-loop analysis to autonomous control. Processing biophysical data, the LLM evaluates plant physiology and triggers hardware actuators to optimize microclimates, execute phenotyping protocols, or induce controlled stress scenarios. This closed-loop architecture establishes a direct AI-biology interface, enabling data-driven exploration of complex biosystems and ecologies. The framework was validated across three case studies, based on a vertical farm and a single-plant setup and deciphered complex micro- and macro-fluctuations in plant physiology. Agents in a production-scale deployment executed multi-parameter optimization, balancing biomass accumulation, chlorophyll content, and energy consumption. The LLM processed biosensing telemetry to modulate full-spectrum, 450 nm, and 660 nm lighting at 2-hour intervals. Compared to periodic control, the system in minimal-time mode reduced the production cycle by 35%. In the energy-optimization mode, it reduced energy consumption by 18% with only a marginal increase in cultivation time, exploiting physiological inertia via light pulses. Finally, the agents autonomously developed an unforeseen strategy of dark-induced chlorophyll accumulation, resulting in a 67.9% energy saving. This framework transforms LLMs into autonomous co-pilots for digital agriculture, improving the cost-to-value ratio and lowering computational and expert-labor constraints.\n\nSubjects:Artificial Intelligence (cs.AI); Biological Physics (physics.bio-ph)\nCite as:[arXiv:2608.09949](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.09949) [cs.AI]\n(or [arXiv:2608.09949v1](https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.09949v1) [cs.AI] for this version)\n[https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.09949](https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2608.09949)\n\narXiv-issued DOI via DataCite\n\n## Submission history\n\nFrom: Serge Kernbach [[view email](https:\u002F\u002Farxiv.org\u002Fshow-email\u002Fcaefcffe\u002F2608.09949)] \n\n**[v1]** Thu, 9 Jul 2026 11:05:42 UTC (7,699 KB)","arXiv (cs.AI)","2026-07-09T03:05:00Z","论文",10,false,74,{"impact":25,"substance":26,"depth":27,"authority":28,"freshness":29,"relevant":8,"comment":30},20,22,18,12,2,"论文提出LLM闭环控制农业系统，显著节能并缩短生产周期，方法新颖，数据详实，但发布于近两个月前，时效性较低。",[32],{"name":18,"url":14},[34,4,35,36,37],"数字农业","闭环控制","垂直农场","节能",{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},"I1","smart-agriculture","智慧农业总论","\u002Ftopic\u002Fdigital-agriculture\u002Fsmart-agriculture","I",[],{"status":46,"locked":22,"version":47,"method":48,"evidence":49},"classified","2026-10-22.2","rules",{"candidates":50,"agricultural_evidence":52},[51,57],{"code":39,"score":28,"supported":52,"title":53,"abstract":55,"tags":56},true,[54],"digital agriculture",[34],[34],{"code":58,"score":8,"supported":22,"title":59,"abstract":60,"tags":61},"I2",[],[],[4],[63,64],"垂直农场 数字农业 闭环控制 节能","垂直农场 数字农业","垂直农场数字农业闭环控制节能-1123","2026-08-30T00:11:11.661645Z",{"id":68,"title":69,"url":70,"summary":71,"summary_zh":16,"content":72,"source_name":73,"source_url":16,"published_at":74,"category":20,"cover_url":16,"hotness":21,"is_selected":22,"score":75,"score_detail":76,"sources":78,"tags":80,"primary_topic":84,"secondary_topics":90,"classification":91,"search_phrases":104,"slug":107,"view_count":29,"doi":16,"paper":16,"created_at":108},1122,"Agentic AI-Driven Cultivation Advisory and Symptom-Level Diagnostic Support in a Controlled Indoor Farming System (AgriEngineering 2026, 8(9), 350)","https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350","Jutarut Chaoraingern、Akarat Pattaraanuvong、Kantapon Paraksa、Kantiporn Khunthong、Tirawat Nontiwantok、Arjin Numsomran(泰国国王科技大学)发表论文。本研究提出一个集成平台,围绕代理AI咨询管道构建,完全在商用硬件上本地运行。管道通过基于条件的路由、意图分类、多步推理和LLM验证门,将基于RAG的Mistral 7B语言模型与LLaVA 7B视觉语言模型耦合,从对话界面提供上下文感知的文本和图像症状级指导。在Cos生菜测试中,该咨询管道在50个查询中达到82.00%加权准确率,涵盖健康、症状、浇水、害虫、根系健康和生长阶段类别。","## 1. Introduction\n\nThe global demand for food continues to rise alongside population growth and increasing pressure on arable land, making food security one of the most pressing challenges of the twenty-first century [[1](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B1-agriengineering-08-00350)]. Indoor farming has emerged as a promising alternative to conventional open-field cultivation, enabling year-round crop production in controlled-environment enclosures that are decoupled from seasonal variability, extreme weather events, and geographic constraints [[2](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B2-agriengineering-08-00350),[3](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B3-agriengineering-08-00350)]. By precisely regulating light, irrigation, humidity, and temperature, indoor systems can achieve substantially higher yield consistency compared with traditional agriculture [[4](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B4-agriengineering-08-00350),[5](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B5-agriengineering-08-00350)]. Despite these advantages, the majority of small-scale and urban indoor farming operations continue to depend on manual observation for plant monitoring and cultivation decision-making, resulting in delayed detection of stress conditions, inconsistent crop outcomes, and inefficient use of water and energy resources [[6](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B6-agriengineering-08-00350),[7](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B7-agriengineering-08-00350)].\n\nThe integration of Internet of Things (IoT) technologies into agricultural systems has enabled continuous, automated monitoring of key environmental parameters, including soil moisture, temperature, humidity, and photosynthetically active light intensity [[8](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B8-agriengineering-08-00350),[9](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B9-agriengineering-08-00350)]. IoT-based platforms can trigger automated actuation such as irrigation pulses and grow-light scheduling, substantially reducing the burden of manual intervention and supporting resource-efficient cultivation [[10](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B10-agriengineering-08-00350),[11](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B11-agriengineering-08-00350)]. Recent deployments in smart greenhouse environments have demonstrated stable closed-loop control of temperature, humidity, and fertigation with measurable improvements in crop yield and resource use [[12](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B12-agriengineering-08-00350),[13](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B13-agriengineering-08-00350)]. However, translating raw sensor streams into actionable cultivation guidance remains a significant challenge, particularly for non-expert users who lack the agronomic knowledge to interpret environmental readings or identify early signs of plant disease [[14](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B14-agriengineering-08-00350)].\n\nRecent advances in large language models (LLMs) and retrieval-augmented generation (RAG) have opened new avenues for intelligent advisory systems across a broad range of domains, including agriculture [[15](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B15-agriengineering-08-00350),[16](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B16-agriengineering-08-00350)]. RAG-based agricultural chatbots ground LLM responses in domain-specific knowledge bases such as cultivation guides and pest management manuals, substantially reducing hallucination and improving factual reliability compared with vanilla LLMs [[17](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B17-agriengineering-08-00350),[18](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B18-agriengineering-08-00350)]. Systems such as CottonBot [[19](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B19-agriengineering-08-00350)] have demonstrated that RAG-augmented LLMs integrated with IoT soil-moisture sensors and weather APIs can deliver expert-level irrigation recommendations responsive to real-time field conditions. Similarly, PEZEGO [[20](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B20-agriengineering-08-00350)] illustrated how LLM reasoning coupled with IoT telemetry can support pest management decisions in precision agriculture. However, most existing systems rely either on static knowledge bases without real-time sensor integration, or on single-task advisory pipelines without multimodal disease diagnosis capabilities. The concept of agentic AI, where an LLM autonomously routes inputs through condition-based pipelines and invokes external knowledge retrieval as part of its reasoning chain [[21](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B21-agriengineering-08-00350)], offers a qualitative advance over standard RAG by enabling context-aware multi-step responses, yet end-to-end agentic deployments within unified indoor farming platforms remain largely unexplored.\n\nAccurate and timely plant disease diagnosis is equally critical to cultivation advisory systems as undetected pathogens and nutrient deficiencies can severely reduce crop quality and yield [[22](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B22-agriengineering-08-00350)]. Traditional diagnosis depends on expert visual inspection, which is both labor-intensive and inaccessible for non-specialist growers. Deep learning-based approaches have demonstrated strong performance in plant disease classification and segmentation tasks [[23](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B23-agriengineering-08-00350),[24](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B24-agriengineering-08-00350)], and transformer-based models combining in situ sensor data with plant images have further improved diagnostic accuracy under realistic field conditions [[25](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B25-agriengineering-08-00350)]. More recently, LLM-based knowledge graph construction for plant diseases [[26](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B26-agriengineering-08-00350)] and multimodal vision-language models have shown the potential to support conversational, image-grounded diagnosis accessible to end users. Nevertheless, integrating such multimodal diagnosis capability within a real-time controlled indoor farming system, where automated closed-loop control and symptom-level diagnosis are delivered through a single natural-language interface, has not been demonstrated in the prior literature.\n\nTo address these gaps, this paper presents a system for agentic AI-driven cultivation advisory and symptom-level diagnostic support deployed in a controlled indoor farming system. The proposed system, built around Cos lettuce cultivation as the validation crop, integrates an ESP32-based sensing and control layer for real-time monitoring of soil moisture and grow light status together with automated closed-loop control of irrigation and LED grow lighting. An agentic LLM pipeline, grounded through an RAG pipeline constructed from domain-specific cultivation knowledge, employs condition-based routing and multi-step reasoning to deliver context-aware cultivation recommendations. A multimodal vision-language model provides symptom-level diagnostic support from user-submitted image and text inputs, enabling non-expert users to receive symptom interpretation and management guidance through a web-based conversational interface.\n\nWhile the individual building blocks employed in this work are established open-source techniques and are not claimed as contributions in themselves, the contribution lies in the architecture that governs how inputs are routed between the text and vision paths, how responses are grounded in retrieved documentation and validated before delivery, and how the advisory layer is isolated from the control layer. [Table 1](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#table_body_display_agriengineering-08-00350-t001) illustrates that existing systems address these capabilities in isolation, and the present work targets the identified gap.\n\n**Table 1.** Design scope of representative IoT-, LLM-, and vision-based agricultural systems in relation to the proposed platform.\n\nTwo points of scope should be stated at the outset. First, this article reports a systems-engineering contribution. The research question is whether a platform integrating threshold-based actuation, on-device vision monitoring, and an agentic LLM advisory pipeline can be constructed and operated stably on commodity hardware without cloud dependency. Whether such a system improves agronomic outcomes relative to conventional management is a separate question, requiring a replicated and randomized cultivation trial with crop-performance measurement, and is not addressed here. Second, the term diagnosis is used throughout in the sense of symptom-level diagnostic support: the identification of visible symptoms, the enumeration of plausible differential causes, and the provision of corresponding management guidance. It does not denote pathogen confirmation, which requires microbiological or molecular verification and lies outside the scope of this study.\n\nThe main contributions of this work are as follows:\n\n*   **Agentic advisory pipeline.** An LLM pipeline combining condition-based routing, intent classification, RAG-grounded multi-step reasoning, and an LLM validation gate. It is specified as an explicit dual-path dispatch algorithm rather than a monolithic prompt.\n\n*   **Multimodal diagnostic support.** A LLaVA 7B module that interprets plant symptoms from images and text. It identifies visible symptoms, lists plausible causes, and returns management guidance for users without agronomic training.\n\n*   **Browning detection.** A vision routine pairing YOLO-based plant localization with color-ratio analysis to flag abnormal leaf coloration. Each alert logs the incident image, growth stage, and sensor state for follow-up.\n\n*   **Isolated control layer.** A threshold-driven architecture for irrigation and lighting actuation, separated from the advisory layer so that no language model output can reach an actuator ([Section 3.1.6](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec3dot1dot6-agriengineering-08-00350)). Continuous unattended operation was demonstrated across a Cos lettuce deployment period.\n\nThe remainder of this paper is organized as follows. [Section 2](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec2-agriengineering-08-00350) reviews related work and positions the proposed system against the existing literature. [Section 3](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec3-agriengineering-08-00350) presents the materials and methods, covering the system architecture and design ([Section 3.1](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec3dot1-agriengineering-08-00350)) and the materials and experimental methods ([Section 3.2](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec3dot2-agriengineering-08-00350)). [Section 4](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec4-agriengineering-08-00350) presents the results. [Section 5](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec5-agriengineering-08-00350) discusses their implications, limitations, and directions for future research. [Section 6](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#sec6-agriengineering-08-00350) concludes the paper.\n\n## 2. Related Work\n\nThis section reviews prior work across three themes directly relevant to the proposed system, closed-loop control in indoor farming, LLM- and agentic-AI-based advisories, and vision-based plant monitoring and diagnosis, and then synthesizes the recurring limitations that motivate the present integration.\n\n### 2.1. IoT-Enabled Monitoring and Closed-Loop Control\n\nControlled-environment agriculture has been advanced substantially by IoT instrumentation that continuously senses soil moisture, temperature, humidity, and light and drives automated actuation [[8](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B8-agriengineering-08-00350),[9](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B9-agriengineering-08-00350),[27](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B27-agriengineering-08-00350)]. Smart indoor and greenhouse platforms have demonstrated stable closed-loop regulation of temperature, humidity, and fertigation with measurable gains in yield and resource efficiency [[4](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B4-agriengineering-08-00350),[10](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B10-agriengineering-08-00350),[12](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B12-agriengineering-08-00350),[28](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B28-agriengineering-08-00350)], and AI-assisted extensions have added fault detection [[13](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B13-agriengineering-08-00350)], nutrient optimization [[5](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B5-agriengineering-08-00350)], and low-cost sensing on commodity hardware [[29](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B29-agriengineering-08-00350),[30](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B30-agriengineering-08-00350)]. These systems establish that reliable autonomous control is achievable on affordable hardware. While their primary focus is on sensing and actuation, the translation of raw telemetry into natural-language cultivation guidance remains an open challenge, particularly for non-expert operators [[6](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B6-agriengineering-08-00350),[14](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B14-agriengineering-08-00350)]. The proposed system builds on this closed-loop foundation and adds a conversational advisory layer to address this gap.\n\n### 2.2. LLM- and RAG-Based Agricultural Advisory\n\nLarge language models grounded through retrieval-augmented generation (RAG) have emerged as a means of delivering domain-specific advice while mitigating hallucination [[15](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B15-agriengineering-08-00350),[16](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B16-agriengineering-08-00350)]. RAG-based advisory frameworks anchor model responses in curated agronomic corpora, improving factual reliability over vanilla LLMs [[17](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B17-agriengineering-08-00350),[18](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B18-agriengineering-08-00350),[22](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B22-agriengineering-08-00350)]. Beyond static retrieval, a smaller body of work applies agentic designs, in which the LLM autonomously routes inputs, selects tools, and reasons over multiple steps. CottonBot [[19](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B19-agriengineering-08-00350)] couples an RAG-augmented LLM with agentic tool invocation and IoT soil-moisture and weather sources to produce irrigation recommendations for outdoor cotton, and PEZEGO [[20](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B20-agriengineering-08-00350)] pairs LLM reasoning with IoT telemetry for pest management. At a larger scale, Pardo-Pina et al. [[21](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B21-agriengineering-08-00350)] propose a multi-agent service architecture for smart greenhouses with RAG-grounded LLM agents performing telemetry analytics and decision support. These systems demonstrate the value of integrating RAG and agentic reasoning for agricultural advisory systems. However, none incorporates image-grounded disease diagnosis alongside text-based advisories within a single pipeline. The present work extends this direction by combining dual-modality routing (text and image), intent classification, and output validation within a single pipeline, and by adding multimodal disease diagnosis as a complementary capability alongside cultivation advisory systems, deployed entirely on-device without external API dependency.\n\n### 2.3. Vision-Based Plant Monitoring and Disease Diagnosis\n\nDeep learning has become the standard for plant monitoring and disease recognition. Convolutional detectors and segmentation models achieve strong performance in disease classification [[23](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B23-agriengineering-08-00350),[24](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B24-agriengineering-08-00350),[31](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B31-agriengineering-08-00350)], single-stage YOLO detectors have been applied to plant and crop localization across growth stages [[32](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B32-agriengineering-08-00350),[33](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B33-agriengineering-08-00350)], and transformer-based models that fuse in situ sensor data with plant images further improve diagnostic accuracy under field conditions [[25](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B25-agriengineering-08-00350)]; broader reviews document the rapid uptake of deep learning for plant-stress imaging [[34](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B34-agriengineering-08-00350)]. More recent work explores LLM-driven knowledge representation for plant disease, such as knowledge-graph construction from the disease literature [[26](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#B26-agriengineering-08-00350)]. These approaches are primarily designed for recognition tasks and operate independently of cultivation control systems. In the proposed system, YOLO11 detection serves a dual role, localizing the plant for growth-stage classification and for region-constrained browning analysis, while a vision-language model makes diagnosis accessible through the same conversational interface that delivers cultivation advice.\n\n### 2.4. Synthesis and Research Gap\n\n[Table 1](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#table_body_display_agriengineering-08-00350-t001) summarizes the preceding review along five capability dimensions that a non-expert-accessible indoor farming platform would ideally satisfy: autonomous closed-loop control, RAG-grounded advisory, agentic routing, multimodal image diagnosis, and cloud-free on-device operation. Existing systems each address a subset. Control-oriented IoT platforms provide autonomous actuation but no advisory intelligence; RAG and agentic advisory systems provide language-based guidance but omit multimodal diagnosis and depend on cloud infrastructure; and vision-based diagnosis models provide image understanding in isolation from control and advisory delivery. No prior system spans all five dimensions within a single deployment. The present work targets exactly this gap, integrating threshold-based control, agentic RAG-grounded advisories, and multimodal symptom interpretation on commodity hardware without cloud dependency, which is the integration this paper claims as its contribution.\n\nA recurring constraint in this literature is that detection models deployed on low-power IoT nodes are typically restricted to compact architectures such as Tiny-YOLO in order to remain within the compute budget of the edge device, which limits achievable detection accuracy. The present work addresses this constraint through the placement of the inference workload rather than through further model compression: the detection model executes locally on a Raspberry Pi 5 acting as a supervisory node above the ESP32 sensing layer, at a five-second inference interval, so that a mid-capacity detector can be used without cloud offloading while the microcontroller retains responsibility only for deterministic threshold actuation.\n\n## 3. Materials and Methods\n\n### 3.1. System Architecture and Design\n\n#### 3.1.1. Overall System Architecture\n\nThe proposed system integrates three distinct architectural tiers into a unified indoor farming platform: (i)a physical sensing and control layer responsible for environmental sensing and automated actuator control, (ii)a backend AI intelligence layer running on the Raspberry Pi 5 comprising a FastAPI server, a YOLO11m plant-monitoring module, and a structured local data logging system, and (iii)a frontend layer hosted on a laptop comprising the Flowise-based agentic chatbot pipeline, Ollama LLM inference engine, and a React 19.2.0\u002FVite 7.2.5 web application accessible to the end user via a local network browser. [Figure 1](https:\u002F\u002Fwww.mdpi.com\u002F2624-7402\u002F8\u002F9\u002F350#fig_body_display_agriengineering-08-00350-f001) illustrates the data-flow relationships among these three layers.\n\n**Figure 1.** Three-layer s","AgriEngineering (MDPI)","2026-08-23T01:00:00Z",64,{"impact":28,"substance":25,"depth":27,"authority":28,"freshness":29,"relevant":8,"comment":77},"论文提出基于Agentic AI的室内农业栽培咨询与症状级诊断系统，方法新颖，数据详实，但发表于一周前，时效性较低。",[79],{"name":73,"url":70},[4,81,82,35,83],"代理AI","室内农业","诊断",{"code":85,"slug":86,"name":87,"path":88,"parent_code":89},"B1","vegetables","蔬菜","\u002Ftopic\u002Fhorticulture\u002Fvegetables","B",[],{"status":46,"locked":22,"version":47,"method":48,"evidence":92},{"candidates":93,"agricultural_evidence":52},[94,100],{"code":85,"score":95,"supported":52,"title":96,"abstract":97,"tags":99},3,[],[98],"生菜",[],{"code":58,"score":8,"supported":22,"title":101,"abstract":102,"tags":103},[],[],[4],[105,106],"室内农业 闭环控制 代理AI 诊断","室内农业 闭环控制","室内农业闭环控制代理AI诊断-1122","2026-08-30T00:11:11.616495Z",{"id":110,"title":111,"url":112,"summary":113,"summary_zh":16,"content":114,"source_name":115,"source_url":16,"published_at":74,"category":20,"cover_url":16,"hotness":21,"is_selected":22,"score":23,"score_detail":116,"sources":118,"tags":120,"primary_topic":125,"secondary_topics":131,"classification":132,"search_phrases":152,"slug":155,"view_count":29,"doi":16,"paper":16,"created_at":156},1121,"Intelligent Agents for Smart Agriculture: Architectures, Applications, and Future Challenges (Agriculture 2026, 16(17), 1808)","https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808","Wenzheng Tao、Qiwei Sang(江苏大学计算机科学与通信工程学院)、Cong Chen(农业农村部南京农业机械化研究所)、Qirong Mao(江苏大学)发表综述论文。本综述聚焦由大语言模型及相关多模态基础模型驱动的现代农业智能体,考察这一新兴领域如何重塑智能农业系统的组织模式。综述阐明农业智能体的概念边界,综合其架构基础、关键能力、应用场景、部署挑战和未来研究方向。综述文献表明,农业智能体正从孤立的感知、预测和响应生成,向农业任务链中农业知识、动态数据、外部工具和决策流程的目标导向协调转变,开始支持更集成的知识服务、作物监测与诊断、决策支持和农场级协作管理形式。","## Journals\n\n[Active Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)[Find a Journal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournalfinder)[Journal Proposal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals\u002Fproposal)[Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n## [Topics](https:\u002F\u002Fwww.mdpi.com\u002Ftopics)\n\n## Information\n\n[For Authors](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)[For Reviewers](https:\u002F\u002Fwww.mdpi.com\u002Freviewers)[For Editors](https:\u002F\u002Fwww.mdpi.com\u002Feditors)[For Librarians](https:\u002F\u002Fwww.mdpi.com\u002Flibrarians)[For Publishers](https:\u002F\u002Fwww.mdpi.com\u002Fpublishing_services)[For Societies](https:\u002F\u002Fwww.mdpi.com\u002Fsocieties)[For Conference Organizers](https:\u002F\u002Fwww.mdpi.com\u002Fconference_organizers)\n\n[Open Access Policy](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess)[Institutional Open Access Program](https:\u002F\u002Fwww.mdpi.com\u002Fioap)[Special Issues Guidelines](https:\u002F\u002Fwww.mdpi.com\u002Fspecial_issues_guidelines)[Editorial Process](https:\u002F\u002Fwww.mdpi.com\u002Feditorial_process)[Research and Publication Ethics](https:\u002F\u002Fwww.mdpi.com\u002Fethics)[Article Processing Charges](https:\u002F\u002Fwww.mdpi.com\u002Fapc)[Awards](https:\u002F\u002Fwww.mdpi.com\u002Fawards)[Testimonials](https:\u002F\u002Fwww.mdpi.com\u002Ftestimonials)\n\n## [Author Services](https:\u002F\u002Fwww.mdpi.com\u002Fauthors\u002Fservices)\n\n## Initiatives\n\n[Sciforum](https:\u002F\u002Fsciforum.net\u002F)[MDPI Books](https:\u002F\u002Fwww.mdpi.com\u002Fbooks)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002F)[Scilit](https:\u002F\u002Fwww.scilit.com\u002F)[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002F)[Encyclopedia](https:\u002F\u002Fencyclopedia.pub\u002F)[JAMS](https:\u002F\u002Fjams.pub\u002F)[Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n## About\n\n[Overview](https:\u002F\u002Fwww.mdpi.com\u002Fabout)[Contact](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontact)[Careers](https:\u002F\u002Fcareers.mdpi.com\u002F)[News](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fannouncements)[Press](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fpress)[Blog](http:\u002F\u002Fblog.mdpi.com\u002F)\n\n[Sign In \u002F Sign Up](https:\u002F\u002Fwww.mdpi.com\u002Fuser\u002Flogin)\n\n## Notice\n\n You can make submissions to other journals [here](https:\u002F\u002Fsusy.mdpi.com\u002Fuser\u002Fmanuscripts\u002Fupload). \n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n## Notice\n\nYou are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.\n\n[Continue](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)[Cancel](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\nAll articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to [https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess).\n\nFeature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.\n\nFeature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.\n\nEditor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.\n\nOriginal Submission Date Received: .\n\n[![Image 4: MDPI Open Access Journals](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Fmdpi-pub-logo-black-small1.svg?da3a8dcae975a41c?1788001719)](https:\u002F\u002Fwww.mdpi.com\u002F)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808#)[_zoom\\_out\\_map_](https:\u002F\u002Fwww.mdpi.com\u002Ftoggle_desktop_layout_cookie \"Toggle desktop layout\")[_search_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808#)[_menu_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808# \"MDPI main page\")\n\n[![Image 5: MDPI Open Access Journals](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Fmdpi-pub-logo-black-small1.svg?da3a8dcae975a41c?1788001719)](https:\u002F\u002Fwww.mdpi.com\u002F)\n\n*   [Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)\n    *   \n        *   [Active Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)\n        *   [Find a Journal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournalfinder)\n        *   [Journal Proposal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals\u002Fproposal)\n        *   [Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n*   [Topics](https:\u002F\u002Fwww.mdpi.com\u002Ftopics)\n*   [Information](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)\n    *   \n        *   [For Authors](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)\n        *   [For Reviewers](https:\u002F\u002Fwww.mdpi.com\u002Freviewers)\n        *   [For Editors](https:\u002F\u002Fwww.mdpi.com\u002Feditors)\n        *   [For Librarians](https:\u002F\u002Fwww.mdpi.com\u002Flibrarians)\n        *   [For Publishers](https:\u002F\u002Fwww.mdpi.com\u002Fpublishing_services)\n        *   [For Societies](https:\u002F\u002Fwww.mdpi.com\u002Fsocieties)\n        *   [For Conference Organizers](https:\u002F\u002Fwww.mdpi.com\u002Fconference_organizers)\n\n        *   [Open Access Policy](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess)\n        *   [Institutional Open Access Program](https:\u002F\u002Fwww.mdpi.com\u002Fioap)\n        *   [Special Issues Guidelines](https:\u002F\u002Fwww.mdpi.com\u002Fspecial_issues_guidelines)\n        *   [Editorial Process](https:\u002F\u002Fwww.mdpi.com\u002Feditorial_process)\n        *   [Research and Publication Ethics](https:\u002F\u002Fwww.mdpi.com\u002Fethics)\n        *   [Article Processing Charges](https:\u002F\u002Fwww.mdpi.com\u002Fapc)\n        *   [Awards](https:\u002F\u002Fwww.mdpi.com\u002Fawards)\n        *   [Testimonials](https:\u002F\u002Fwww.mdpi.com\u002Ftestimonials)\n\n*   [Author Services](https:\u002F\u002Fwww.mdpi.com\u002Fauthors\u002Fservices)\n*   [Initiatives](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Finitiatives)\n    *   \n        *   [Sciforum](https:\u002F\u002Fsciforum.net\u002F)\n        *   [MDPI Books](https:\u002F\u002Fwww.mdpi.com\u002Fbooks)\n        *   [Preprints.org](https:\u002F\u002Fwww.preprints.org\u002F)\n        *   [Scilit](https:\u002F\u002Fwww.scilit.com\u002F)\n        *   [SciProfiles](https:\u002F\u002Fsciprofiles.com\u002F)\n        *   [Encyclopedia](https:\u002F\u002Fencyclopedia.pub\u002F)\n        *   [JAMS](https:\u002F\u002Fjams.pub\u002F)\n        *   [Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n*   [About](https:\u002F\u002Fwww.mdpi.com\u002Fabout)\n    *   \n        *   [Overview](https:\u002F\u002Fwww.mdpi.com\u002Fabout)\n        *   [Contact](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontact)\n        *   [Careers](https:\u002F\u002Fcareers.mdpi.com\u002F)\n        *   [News](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fannouncements)\n        *   [Press](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fpress)\n        *   [Blog](http:\u002F\u002Fblog.mdpi.com\u002F)\n\n[Sign In \u002F Sign Up](https:\u002F\u002Fwww.mdpi.com\u002Fuser\u002Flogin)[Submit](https:\u002F\u002Fsusy.mdpi.com\u002Fuser\u002Fmanuscripts\u002Fupload)\n\nSearch for Articles:\n\nTitle \u002F Keyword\n\nAuthor \u002F Affiliation \u002F Email\n\nJournal\n\n[Agronomy](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\nArticle Type\n\n[All Article Types](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n[Advanced Search](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808#)\n\nSection\n\n[All Sections](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\nSpecial Issue\n\n[All Special Issues](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\nVolume\n\nIssue\n\nNumber\n\nPage\n\nLogical Operator Operator\n\nSearch Text\n\nSearch Type\n\n_add\\_circle\\_outline_\n\n_remove\\_circle\\_outline_\n\n[_share_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808# \"Share\")[_announcement_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808# \"Help\")\n\n## Need Help?\n\n### Support\n\nFind support for a specific problem in the support section of our website.\n\n[Get Support](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontactform)\n\n### Feedback\n\nPlease let us know what you think of our products and services.\n\n[Give Feedback](https:\u002F\u002Fwww.mdpi.com\u002Ffeedback\u002Fsend)\n\n### Information\n\nVisit our dedicated information section to learn more about MDPI.\n\n[Get Information](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n# Error 404 - File not found\n\nThe webpage you are looking for could not be found. The URL may have been incorrectly typed, or the page may have been moved into another part of the mdpi.com site.\n\n[Return to the main page](https:\u002F\u002Fwww.mdpi.com\u002F)\n\n[Contact](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontact)\n\n### Further Information\n\n[Article Processing Charges](https:\u002F\u002Fwww.mdpi.com\u002Fapc)[Pay an Invoice](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fpayment)[Open Access Policy](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess)[Contact MDPI](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontact)[Jobs at MDPI](https:\u002F\u002Fcareers.mdpi.com\u002F)\n\n### Guidelines\n\n[For Authors](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)[For Reviewers](https:\u002F\u002Fwww.mdpi.com\u002Freviewers)[For Editors](https:\u002F\u002Fwww.mdpi.com\u002Feditors)[For Librarians](https:\u002F\u002Fwww.mdpi.com\u002Flibrarians)[For Publishers](https:\u002F\u002Fwww.mdpi.com\u002Fpublishing_services)[For Societies](https:\u002F\u002Fwww.mdpi.com\u002Fsocieties)[For Conference Organizers](https:\u002F\u002Fwww.mdpi.com\u002Fconference_organizers)\n\n### MDPI Initiatives\n\n[Sciforum](https:\u002F\u002Fsciforum.net\u002F)[MDPI Books](https:\u002F\u002Fwww.mdpi.com\u002Fbooks)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002F)[Scilit](https:\u002F\u002Fwww.scilit.com\u002F)[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002F?utm_source=mpdi.com&utm_medium=bottom_menu&utm_campaign=initiative)[Encyclopedia](https:\u002F\u002Fencyclopedia.pub\u002F)[JAMS](https:\u002F\u002Fjams.pub\u002F)[Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n### Follow MDPI\n\n[LinkedIn](https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fmdpi)[Facebook](https:\u002F\u002Fwww.facebook.com\u002FMDPIOpenAccessPublishing)[X](https:\u002F\u002Fx.com\u002FMDPIOpenAccess)\n\n![Image 6: MDPI](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Fmdpi-pub-logo-white-small.png?71d18e5f805839ab?1788001719)\n\n © 1996-2026 MDPI (Basel, Switzerland) unless otherwise stated \n\n[Disclaimer](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n Disclaimer\u002FPublisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and\u002For the editor(s). MDPI and\u002For the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. \n\n[Legal Notice](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Flegal-notice)[Terms and Conditions](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fterms-and-conditions)[Privacy Policy](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fprivacy)[Privacy Settings](javascript:void(0);)[Accessibility](https:\u002F\u002Fwww.mdpi.com\u002Faccessibility)\n\n We use cookies on our website to ensure you get the best experience.\n\n Read more about our cookies [here](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fprivacy). \n\n[Accept](https:\u002F\u002Fwww.mdpi.com\u002Faccept_cookies)\n\n## Share Link\n\n[](mailto:?&subject=From%20MDPI%3A%20%22%22&body=https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808%3A%0A%0A \"Email\")[](https:\u002F\u002Fx.com\u002Fintent\u002Ftweet?text=&hashtags=&url=https%3A%2F%2Fwww.mdpi.com%2F2073-4395%2F16%2F17%2F1808&via=MDPIOpenAccess \"Twitter\")[](http:\u002F\u002Fwww.linkedin.com\u002FshareArticle?mini=true&url=https%3A%2F%2Fwww.mdpi.com%2F2073-4395%2F16%2F17%2F1808&title=%26source%3Dhttps%3A%2F%2Fwww.mdpi.com \"LinkedIn\")[](http:\u002F\u002Fwww.facebook.com\u002Fsharer.php?u=https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808 \"facebook\")[](javascript:void(0); \"Wechat\")[](http:\u002F\u002Fwww.reddit.com\u002Fsubmit?url=https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808 \"Reddit\")[](http:\u002F\u002Fwww.mendeley.com\u002Fimport\u002F?url=https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808 \"Mendeley\")[](http:\u002F\u002Fwww.citeulike.org\u002Fposturl?url=https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808 \"CiteULike\")\n\n[Copy](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n## Share\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808)\n\n[Back to Top Top](https:\u002F\u002Fwww.mdpi.com\u002F2073-4395\u002F16\u002F17\u002F1808#)","Agriculture (MDPI)",{"impact":25,"substance":25,"depth":27,"authority":28,"freshness":7,"relevant":8,"comment":117},"智能体在智慧农业中的应用综述，内容专业，但时效性较低。",[119],{"name":115,"url":112},[121,122,123,4,124],"多智能体","智能农业","农业智能体","决策支持",{"code":126,"slug":127,"name":128,"path":129,"parent_code":130},"F5","machinery","农机装备与机械化","\u002Ftopic\u002Finputs-equipment\u002Fmachinery","F",[],{"status":46,"locked":22,"version":47,"method":48,"evidence":133},{"candidates":134,"agricultural_evidence":52},[135,141,148],{"code":39,"score":136,"supported":52,"title":137,"abstract":139,"tags":140},8,[138],"smart agriculture",[],[],{"code":126,"score":142,"supported":52,"title":143,"abstract":144,"tags":147},6,[],[145,146],"农业机械","机械化",[],{"code":58,"score":95,"supported":22,"title":149,"abstract":150,"tags":151},[],[],[4,123,124,121],[153,154],"农业智能体 决策支持 多智能体 智能农业","农业智能体 决策支持","农业智能体决策支持多智能体智能农业-1121","2026-08-30T00:11:11.568241Z",{"id":158,"title":159,"url":160,"summary":161,"summary_zh":16,"content":162,"source_name":163,"source_url":16,"published_at":164,"category":20,"cover_url":16,"hotness":21,"is_selected":22,"score":165,"score_detail":166,"sources":168,"tags":170,"primary_topic":177,"secondary_topics":178,"classification":179,"search_phrases":191,"slug":194,"view_count":7,"doi":16,"paper":16,"created_at":195},543,"AgriAgent: A Heterogeneous Multi-Agent Orchestration Framework for Autonomous, Knowledge-Grounded Precision Agriculture","https:\u002F\u002Fwww.epj-conferences.org\u002Farticles\u002Fepjconf\u002Fabs\u002F2026\u002F39\u002Fepjconf_aist4ge26_06004\u002Fepjconf_aist4ge26_06004.html","面向精准农业受限于多模遥测难以实时自治的认知瓶颈,提出 AgriAgent 分层多智能体(MAS)框架,将大语言模型从被动分析工具重塑为目标导向的协调器,采用包含感知、认知、执行三层的三层编排(TLO)拓扑。构建基于 APSIM 生物物理模型并经 GAN 扩展的高保真 10,000 状态合成库,引入黑天鹅环境胁迫。基准测试显示推理成功率 98.2%,决策时延降低 97%,将传统人在回路模式转为智能体驱动,为生物-数字自治农场提供蓝图。","Open Access\n\n| Issue |  | EPJ Web Conf. **Volume**383, 2026 The 1 st International Conference on Artificial Intelligence and Smart Technologies for Green Energy (AIST4GE’26) |\n| --- |\n|  |\n|  |\n| Article Number |  | 06004 |\n| Number of page(s) |  | 9 |\n| Section |  | Artificial Intelligence for Precision Agriculture and Hydraulic Systems |\n| DOI |  | [https:\u002F\u002Fdoi.org\u002F10.1051\u002Fepjconf\u002F202638306004](https:\u002F\u002Fdoi.org\u002F10.1051\u002Fepjconf\u002F202638306004) |\n| Published online |  | 12 August 2026 |\n\nEPJ Web of Conferences **383**, 06004 (2026)\n\n[https:\u002F\u002Fdoi.org\u002F10.1051\u002Fepjconf\u002F202638306004](https:\u002F\u002Fdoi.org\u002F10.1051\u002Fepjconf\u002F202638306004)\n\n*, Mohammed Berrahal**, Mimoune Yandouzi, Mounir Grari, Aissa Kerkour Elmiad and Mohammed Badaoui\n\nIIA Research Lab, ESTO, Mohamed First University, Oujda, Morocco\n\n* e-mail: [n.boussala@ump.ac.ma](mailto:n.boussala@ump.ac.ma)\n\n** e-mail: [m.berrahal@ump.ac.ma](mailto:m.berrahal@ump.ac.ma)\n\nPublished online: 12 August 2026\n\nAbstract\n\nThe global agricultural sector is currently navigating an unprecedented convergence of challenges, characterized by escalating caloric demands, systemic climate instability, and the urgent necessity for resource-neutral production. While the advent of Precision Agriculture (PA) introduced data-driven monitoring, the industry remains hindered by a \"cognitive bottleneck\" the inability to transform fragmented, multi-modal telemetry into real-time, autonomous interventions with-out intensive human oversight. This paper introduces AgriAgent, a pioneering, hierarchical Multi-Agent System (MAS) framework that redefines the role of Large Language Models (LLMs) from passive analytical tools to active, goal-oriented orchestrators. Unlike traditional monolithic architectures, AgriAgent em-ploys a Tri-Layer Orchestration (TLO) topology, comprising specialized agents for Perception, Cognition, and Actuation. To validate the framework's reasoning plasticity, we developed a high-fidelity synthetic library of _N_ = 10,000 biological states, grounded in AP-SIM biophysical modeling and expanded via Generative Adversarial Networks (GANs) to include \"Black Swan\" environmental stressors. Our results demonstrate that by shifting from a reactive \"human-in-the-loop\" model to a proactive \"agent-driven\" paradigm, the system achieves a state of Ag-ronomic Equilibrium. Quantitative benchmarking shows a 98.2% reasoning suc-cess rate, a 97% reduction in decision latency, and a significant optimization of computational overhead. This research provides the definitive blueprint for the transition toward \"Bio-Digital\" autonomy, where the farm functions as a self-correcting, intelligent ecosystem capable of sustaining both global yield and eco-logical integrity in an increasingly volatile climate.\n\n_© The Authors, published by EDP Sciences, 2026_\n\n[![Image 1: Licence Creative Commons](https:\u002F\u002Fi.creativecommons.org\u002Fl\u002Fby\u002F4.0\u002F88x31.png)](https:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby\u002F4.0\u002F)This is an Open Access article distributed under the terms of the [Creative Commons Attribution License 4.0](https:\u002F\u002Fcreativecommons.org\u002Flicenses\u002Fby\u002F4.0\u002F), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.\n\nCurrent usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.\n\nData correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.\n\nInitial download of the metrics may take a while.\n\n*   [Reader's services](https:\u002F\u002Fwww.epj-conferences.org\u002Fcomponent\u002Fservices\u002F)\n*   [Email-alert](https:\u002F\u002Fwww.epj-conferences.org\u002Fcomponent\u002Fservices\u002F?label=toc_email)","EPJ Web of Conferences (AIST4GE'26)","2026-08-11T22:00:00Z",80,{"impact":27,"substance":26,"depth":25,"authority":28,"freshness":136,"relevant":8,"comment":167},"提出多智能体框架提升精准农业自主性，数据详实，但属学术前沿，产业影响尚待验证。",[169],{"name":163,"url":160},[171,172,173,4,174,175,176],"农业人工智能","精准农业","多智能体系统","自主决策","智能农场","生物-数字",{"code":39,"slug":40,"name":41,"path":42,"parent_code":43},[],{"status":46,"locked":22,"version":47,"method":48,"evidence":180},{"candidates":181,"agricultural_evidence":52},[182,187],{"code":39,"score":28,"supported":52,"title":183,"abstract":185,"tags":186},[184],"precision agriculture",[172],[172],{"code":58,"score":29,"supported":22,"title":188,"abstract":189,"tags":190},[],[],[4,171],[192,193],"农业人工智能 多智能体系统 生物-数字 智能农场","农业人工智能 多智能体系统","农业人工智能多智能体系统生物-数字智能农场-543","2026-08-15T00:02:55.829229Z"]