Agricultural drones are revolutionizing modern farming by integrating with IoT, big data, and AI technologies, becoming pivotal in advancing precision agriculture. This study employs CiteSpace to analyze 1,492 agricultural UAV-related publications from China’s CNKI database (2013–2023), revealing evolutionary patterns and emerging trends. The linear growth of publications follows:
$$ N(t) = 35.74t – 708.48 \quad (R^2 = 0.96) $$
where \(N(t)\) represents annual publications and \(t\) denotes years since 2013. Research progressed through three phases: initial exploration (2013–2015), rapid development (2016–2021), and technological maturation (2022–2023).
Core research clusters emerged through keyword co-occurrence analysis (Q=0.812, S=0.948), forming four primary domains:
| Research Domain | Key Technologies | Applications |
|---|---|---|
| Equipment & Technology | Autonomous navigation, Battery optimization | Path planning, Payload enhancement |
| Information Monitoring | Multispectral sensors, Deep learning | Crop health assessment, Soil moisture mapping |
| Crop Management | Variable-rate systems, Precision spraying | Fertilizer/pesticide application, Pollination |
| Plant Protection | Droplet deposition modeling, Thermal fogging | Pest/disease control, Chemical deposition |
Burst detection identified frontier technologies (2019–2023):
| Keyword | Burst Strength | Duration |
|---|---|---|
| Multispectral | 3.82 | 2021–2023 |
| Smart Agriculture | 3.45 | 2021–2023 |
| Pest Monitoring | 2.91 | 2020–2023 |
Recent advances demonstrate agricultural UAV capabilities in precision tasks. Nitrogen application models leverage hyperspectral data for rice cultivation:
$$ N_{opt} = 0.87 \cdot NDVI^{1.2} + 0.12 \cdot LAI^{0.8} \quad (RMSE=0.62) $$
where \(N_{opt}\) is optimal nitrogen dosage, NDVI is normalized difference vegetation index, and LAI is leaf area index. Deep learning architectures achieve 93.6% accuracy in disease identification using convolutional networks:
$$ \mathcal{L} = -\sum_{c=1}^My_{c}\log(p_{c}) + \lambda\|\mathbf{w}\|^2 $$
with \(\mathcal{L}\) as cross-entropy loss, \(y_c\) true labels, \(p_c\) predictions, and \(\lambda\) regularization parameter.
Current challenges include battery limitations restricting operational endurance:
$$ T_{flight} = \frac{C_b \cdot V \cdot \eta}{P_{hover}} \propto \frac{1}{m_{payload}^{0.78}} $$
where \(T_{flight}\) decreases nonlinearly with payload mass \(m_{payload}\). Future trajectories emphasize AI integration and swarm coordination:
$$ \max Z = \sum_{i=1}^k \left[ \frac{A_{cover}(i)}{t_{total}(i)} \right] \quad \text{subject to} \quad \sum \epsilon_{energy}(i) \leq \epsilon_{total} $$
optimizing coverage area \(A_{cover}\) per time unit \(t_{total}\) across \(k\) agricultural drones under energy constraints \(\epsilon\). Hyperspectral-terrestrial fusion models enhance monitoring precision:
$$ \rho_{fusion} = \alpha \cdot \rho_{UAV} + (1-\alpha) \cdot \rho_{ground} \quad \alpha \in [0.6,0.8] $$
Data sources include the CNKI database covering agricultural UAV research. Institutional analysis reveals concentrated expertise at agricultural universities, though collaboration networks remain sparse (density=0.0106).
Future agricultural UAV development will prioritize: 1) AI-edge computing integration for real-time decision-making, 2) swarming coordination algorithms for large-scale operations, and 3) advanced sensor fusion for multidimensional crop phenotyping. These advancements will establish agricultural drones as indispensable cyber-physical systems in smart farming ecosystems, enhancing productivity while minimizing environmental footprints through precision resource management.