Trend AnalysisOther Sciences
AI-Powered Agriculture: Precision Farming for Food Security
Precision agriculture powered by AI, IoT sensors, drones, and federated learning promises to optimize crop yields while reducing resource waste. Recent work documents where these technologies deliver measurable improvements and where adoption barriers—cost, connectivity, expertise—persist.
By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.
Global food production must increase by an estimated 60% by 2050 to feed a projected population of 9.7 billion—while simultaneously reducing agriculture's environmental footprint (water use, chemical inputs, greenhouse gas emissions). Precision agriculture, powered by AI and IoT technologies, aims to resolve this tension by applying inputs (water, fertilizer, pesticide) with spatial and temporal precision rather than uniform application across entire fields. The approach reduces waste, lowers costs, and can improve yields—but its adoption remains uneven.
The Research Landscape
The Full Stack of Autonomous Farming
Mim and Hasan (2025), with 6 citations, provide a comprehensive overview of AI applications across the agricultural lifecycle: smart farming (sensor-based monitoring), vertical and urban farming (controlled environments), and fully autonomous farming systems (robotic planting, weeding, and harvesting).
Key applications documented:
- Crop monitoring: Satellite and drone imagery analyzed by deep learning models to detect crop stress, nutrient deficiency, and disease before visible symptoms appear.
- Yield prediction: ML models combining weather data, soil measurements, and historical yields to predict harvest volumes with 85-95% accuracy, enabling better market planning.
- Autonomous operations: Robotic systems for precision weeding (targeting individual weeds rather than broadcasting herbicide), automated harvesting (using computer vision to assess fruit ripeness), and drone-based spraying (applying pesticide only to affected areas).
The paper notes a significant adoption gap: while large-scale commercial farms in the US, EU, and Australia increasingly deploy these technologies, smallholder farmers in the Global South—who produce a substantial share of global food—lack access due to cost, connectivity, and technical expertise barriers.
UAV-Aided Smart Agriculture
Akbari and Kennedy (2024), with 27 citations, address a specific technical challenge: how to efficiently orchestrate UAV (drone) services for agricultural monitoring while maintaining data freshness. Their solution uses asynchronous federated learning—a privacy-preserving approach where individual farms train local models on their own data and share only model updates through UAV relays.
The innovation is the Age of Information (AoI) metric, which measures how fresh the data is at each decision point. In agriculture, stale data leads to delayed responses to crop stress or pest outbreaks—so maintaining data freshness is directly tied to crop protection effectiveness. The federated learning approach achieves this while keeping individual farm data private—important for farmers who may not want to share proprietary information about their operations.
Plant Disease Detection
Balamurugan et al. (2024), with 6 citations, demonstrate deep learning for plant leaf disease identification. Using CNN, RNN, and LSTM architectures trained on image datasets of diseased and healthy leaves, they achieve classification accuracies exceeding 95% for common diseases across rice, tomato, and corn crops.
The practical impact: early disease detection enables targeted treatment (spraying only affected areas) rather than prophylactic treatment (spraying entire fields), reducing both chemical use and cost. The challenge is deployment—the model runs well on high-end hardware but needs optimization for the low-cost smartphones that most smallholder farmers use.
IoT-Driven Precision
Sachdeva and Anand (2025) review how IoT sensors—measuring soil moisture, temperature, humidity, nutrient levels, and pest activity—enable data-driven crop management. The review documents measurable benefits: significant reductions in water use through sensor-guided irrigation, lower fertilizer application through variable-rate technology, and yield improvements through optimized planting and harvesting timing.
Critical Analysis: Claims and Evidence
<
| Claim | Evidence | Verdict |
|---|
| AI crop monitoring detects stress before visible symptoms | Mim et al.'s review of satellite and drone applications | ✅ Supported |
| Federated learning enables privacy-preserving agricultural AI | Akbari et al.'s UAV-based system | ✅ Supported — 27 citations, technically demonstrated |
| Deep learning achieves >95% accuracy for plant disease identification | Balamurugan et al.'s multi-crop experiments | ✅ Supported — on curated datasets; field deployment accuracy may be lower |
| IoT-guided precision reduces water and fertilizer use by 20-30% | Sachdeva & Anand's review | ✅ Supported — across multiple documented deployments |
Open Questions
Smallholder access: How can precision agriculture technologies be made accessible to the 500+ million smallholder farms in the Global South that produce 80% of food in developing countries?Connectivity: Many agricultural regions lack reliable internet connectivity required for IoT and cloud-based AI. Can edge computing and low-bandwidth solutions bridge this gap?Data ownership: When farms generate data through IoT sensors, who owns it? Platform companies that provide the sensors? The farmer? This has implications for data-driven agricultural markets.Climate adaptation: Can precision agriculture help farmers adapt to climate change—adjusting crop selection, planting timing, and irrigation in response to changing conditions?What This Means for Your Research
For agricultural engineers, the evidence supports precision agriculture as a measurable efficiency improvement. The challenge is making it affordable and accessible beyond large commercial operations.
Explore related work through ORAA ResearchBrain.
Global food production must increase by an estimated 60% by 2050 to feed a projected population of 9.7 billion—while simultaneously reducing agriculture's environmental footprint (water use, chemical inputs, greenhouse gas emissions). Precision agriculture, powered by AI and IoT technologies, aims to resolve this tension by applying inputs (water, fertilizer, pesticide) with spatial and temporal precision rather than uniform application across entire fields. The approach reduces waste, lowers costs, and can improve yields—but its adoption remains uneven.
The Research Landscape
The Full Stack of Autonomous Farming
Mim and Hasan (2025), with 6 citations, provide a comprehensive overview of AI applications across the agricultural lifecycle: smart farming (sensor-based monitoring), vertical and urban farming (controlled environments), and fully autonomous farming systems (robotic planting, weeding, and harvesting).
Key applications documented:
- Crop monitoring: Satellite and drone imagery analyzed by deep learning models to detect crop stress, nutrient deficiency, and disease before visible symptoms appear.
- Yield prediction: ML models combining weather data, soil measurements, and historical yields to predict harvest volumes with 85-95% accuracy, enabling better market planning.
- Autonomous operations: Robotic systems for precision weeding (targeting individual weeds rather than broadcasting herbicide), automated harvesting (using computer vision to assess fruit ripeness), and drone-based spraying (applying pesticide only to affected areas).
The paper notes a significant adoption gap: while large-scale commercial farms in the US, EU, and Australia increasingly deploy these technologies, smallholder farmers in the Global South—who produce a substantial share of global food—lack access due to cost, connectivity, and technical expertise barriers.
UAV-Aided Smart Agriculture
Akbari and Kennedy (2024), with 27 citations, address a specific technical challenge: how to efficiently orchestrate UAV (drone) services for agricultural monitoring while maintaining data freshness. Their solution uses asynchronous federated learning—a privacy-preserving approach where individual farms train local models on their own data and share only model updates through UAV relays.
The innovation is the Age of Information (AoI) metric, which measures how fresh the data is at each decision point. In agriculture, stale data leads to delayed responses to crop stress or pest outbreaks—so maintaining data freshness is directly tied to crop protection effectiveness. The federated learning approach achieves this while keeping individual farm data private—important for farmers who may not want to share proprietary information about their operations.
Plant Disease Detection
Balamurugan et al. (2024), with 6 citations, demonstrate deep learning for plant leaf disease identification. Using CNN, RNN, and LSTM architectures trained on image datasets of diseased and healthy leaves, they achieve classification accuracies exceeding 95% for common diseases across rice, tomato, and corn crops.
The practical impact: early disease detection enables targeted treatment (spraying only affected areas) rather than prophylactic treatment (spraying entire fields), reducing both chemical use and cost. The challenge is deployment—the model runs well on high-end hardware but needs optimization for the low-cost smartphones that most smallholder farmers use.
IoT-Driven Precision
Sachdeva and Anand (2025) review how IoT sensors—measuring soil moisture, temperature, humidity, nutrient levels, and pest activity—enable data-driven crop management. The review documents measurable benefits: significant reductions in water use through sensor-guided irrigation, lower fertilizer application through variable-rate technology, and yield improvements through optimized planting and harvesting timing.
Critical Analysis: Claims and Evidence
<
| Claim | Evidence | Verdict |
|---|
| AI crop monitoring detects stress before visible symptoms | Mim et al.'s review of satellite and drone applications | ✅ Supported |
| Federated learning enables privacy-preserving agricultural AI | Akbari et al.'s UAV-based system | ✅ Supported — 27 citations, technically demonstrated |
| Deep learning achieves >95% accuracy for plant disease identification | Balamurugan et al.'s multi-crop experiments | ✅ Supported — on curated datasets; field deployment accuracy may be lower |
| IoT-guided precision reduces water and fertilizer use by 20-30% | Sachdeva & Anand's review | ✅ Supported — across multiple documented deployments |
Open Questions
Smallholder access: How can precision agriculture technologies be made accessible to the 500+ million smallholder farms in the Global South that produce 80% of food in developing countries?Connectivity: Many agricultural regions lack reliable internet connectivity required for IoT and cloud-based AI. Can edge computing and low-bandwidth solutions bridge this gap?Data ownership: When farms generate data through IoT sensors, who owns it? Platform companies that provide the sensors? The farmer? This has implications for data-driven agricultural markets.Climate adaptation: Can precision agriculture help farmers adapt to climate change—adjusting crop selection, planting timing, and irrigation in response to changing conditions?What This Means for Your Research
For agricultural engineers, the evidence supports precision agriculture as a measurable efficiency improvement. The challenge is making it affordable and accessible beyond large commercial operations.
Explore related work through ORAA ResearchBrain.
References (4)
[1] Mim, M.I., Sultana, F., & Hasan, M.R. (2025). AI-Powered Autonomous Farming: The Future of Sustainable Agriculture. European Journal of Theoretical and Applied Sciences, 3(1). ).02.
[2] Akbari, M., Syed, A., & Kennedy, W.S. (2024). AoI-Aware Energy-Efficient SFC in UAV-Aided Smart Agriculture Using Asynchronous Federated Learning. IEEE Open Journal of the Communications Society.
[3] Balamurugan, M., Srividhya, N., & Indhumathi, G. (2024). Deep Learning Innovations for Improved Plant Leaf Disease Detection in Smart Agriculture. Proc. ICDSIS 2024, IEEE.
[4] Sachdeva, L. & Anand, D. (2025). Impact of Smart Technologies on Precision Agriculture for Sustainable Crop Management. Agrotechniques for Food and Technology Sciences.