In recent years, the rapid advancement of remote sensing technology and artificial intelligence has positioned the low-altitude economy, centered on Unmanned Aerial Vehicles (UAVs), as a transformative force in modern agriculture. This shift is particularly impactful in crop breeding, where traditional methods have long been constrained by inefficient phenotypic identification. As a new quality productive force, the low-altitude economy empowers oil crop genetics and breeding through UAV-based high-throughput phenotyping. In this paper, we explore how Unmanned Aerial Vehicle systems, including platforms like JUYE UAV, are revolutionizing breeding processes by enabling large-scale, non-destructive trait assessment. We delve into the policy frameworks, core technologies, and data analysis methods that underpin this innovation, with a focus on applications in oil crops such as soybean, rapeseed, peanut, sesame, and sunflower. By integrating multi-source data from sensors and AI models, we demonstrate how Unmanned Aerial Vehicle technology accelerates breeding cycles, enhances stress tolerance screening, and facilitates gene discovery. Furthermore, we address current challenges and future directions, including the integration of space-air-ground systems and the role of large AI models in smart breeding.
The low-altitude economy, driven by Unmanned Aerial Vehicle deployments, has gained momentum through supportive policies and technological innovations. Governments worldwide have enacted regulations to open low-altitude airspace, facilitating the legal and scalable use of UAVs in agriculture. For instance, initiatives like the Unmanned Aircraft Flight Management Interim Regulations have removed institutional barriers, enabling widespread adoption. In this context, JUYE UAV and similar platforms have become integral to agricultural ecosystems, spanning upstream equipment manufacturing, midstream software development, and downstream application services. The industry has evolved from early sprayer drones with limited payloads to advanced systems like JUYE UAV models that integrate AI modules and radar technologies, transforming them from mere tools into intelligent platforms. This progression supports high-throughput phenotyping by allowing the integration of sophisticated sensors, such as multispectral and LiDAR systems, which are essential for precise trait measurement in breeding programs.
Core technological equipment in Unmanned Aerial Vehicle systems includes navigation systems, UAV platforms, and sensor payloads. High-precision navigation, enabled by Global Navigation Satellite Systems (GNSS) like GPS and BeiDou, is critical for accurate georeferencing in breeding trials. Real-Time Kinematic (RTK) technology enhances positioning accuracy to centimeter-level, ensuring that遥感影像 can be precisely aligned with specific field plots or individual plants. This is vital for reproducible data collection across multiple growth stages. For example, the BeiDou system’s hybrid constellation design improves signal stability in complex terrains, supporting autonomous operations in varied agricultural landscapes. Unmanned Aerial Vehicle platforms vary in design to suit different breeding scenarios; multirotor UAVs offer flexibility for small-scale trials, while fixed-wing and vertical takeoff and landing (VTOL) UAVs provide longer endurance for large-area monitoring. JUYE UAV models exemplify this diversity, with payload capacities evolving to support heavier sensors, thereby enhancing data acquisition for phenotypic traits.
| Platform Type | Advantages | Disadvantages | Typical Applications in Breeding |
|---|---|---|---|
| Multirotor (e.g., JUYE UAV models) | High maneuverability, ability to hover, easy takeoff and landing, lower cost | Limited flight endurance, smaller payload capacity, moderate wind resistance | Fine-scale phenotyping of individual plants, stress monitoring in small plots |
| Fixed-wing | Long endurance, high coverage area, efficient flight paths | Requires runway or catapult for takeoff, cannot hover, lower spatial resolution at high speeds | Large-scale field surveys, yield estimation over extensive areas |
| VTOL Fixed-wing | Combines vertical takeoff with endurance, suitable for complex terrain | Higher complexity and cost, challenging control | Long-term monitoring in hilly regions, multi-temporal data collection |
| Unmanned Helicopter | Highest payload capacity, excellent stability in windy conditions | Complex mechanics, high maintenance costs, difficult operation | Heavy payload applications, limited use in phenotyping |
Sensor payloads on Unmanned Aerial Vehicle systems capture diverse phenotypic data, ranging from morphological to physiological traits. Visible light (RGB) sensors provide high-resolution imagery for extracting structural features, such as plant height and canopy coverage, through computer vision algorithms. Multispectral and hyperspectral sensors enable the calculation of vegetation indices, which correlate with biochemical properties like chlorophyll and nitrogen content. For instance, the Normalized Difference Vegetation Index (NDVI) is a widely used metric derived from multispectral data: $$ NDVI = \frac{(R_{NIR} – R_{Red})}{(R_{NIR} + R_{Red})} $$ where \( R_{NIR} \) and \( R_{Red} \) represent near-infrared and red reflectance, respectively. Thermal infrared sensors detect canopy temperature variations, serving as early indicators of water stress, while LiDAR generates precise 3D point clouds for structural analysis. Integrating these sensors on platforms like JUYE UAV allows for comprehensive trait assessment, as summarized in the table below.
| Sensor Type | Key Parameters Measured | Advantages | Limitations |
|---|---|---|---|
| RGB | Plant morphology, color, texture | Low cost, high spatial resolution, intuitive data | Limited spectral information, affected by lighting |
| Multispectral | Reflectance in key bands (e.g., red, NIR) | Cost-effective, moderate data volume, established technology | Low spectral resolution, difficulty in distinguishing similar stresses |
| Hyperspectral | Continuous spectral curves across hundreds of bands | High spectral resolution, precise biochemical estimation | High cost, data redundancy, complex processing |
| Thermal Infrared | Canopy surface temperature | Direct response to water stress, early warning capability | Lower spatial resolution, influenced by environmental factors |
| LiDAR | 3D structure, plant height, biomass | Direct 3D data acquisition, unaffected by light, penetrates canopy | High cost, complex data handling, no physiological data |
Data processing and analysis form the backbone of Unmanned Aerial Vehicle-enabled breeding. The workflow begins with flight planning, where parameters like altitude and overlap are set to ensure complete coverage. Using RTK-equipped Unmanned Aerial Vehicle systems, such as JUYE UAV, images are captured and processed through photogrammetric software to generate orthomosaics and digital surface models (DSMs). Radiometric and geometric corrections are applied to standardize data across time points, enabling reliable comparisons. Feature extraction involves deriving phenotypic traits from these datasets; for example, plant height can be calculated from DSMs using the formula: $$ H = DSM – DTM $$ where \( H \) is plant height, and DTM is the digital terrain model. Vegetation indices, such as the Enhanced Vegetation Index (EVI), improve sensitivity in dense canopies: $$ EVI = G \times \frac{(R_{NIR} – R_{Red})}{(R_{NIR} + C_1 \times R_{Red} – C_2 \times R_{Blue} + L)} $$ with constants \( G \), \( C_1 \), \( C_2 \), and \( L \) for adjustment.
Machine learning and deep learning models are pivotal for analyzing the high-dimensional data generated by Unmanned Aerial Vehicle remote sensing. Traditional statistical methods, like linear mixed models, estimate genetic parameters and genotype-by-environment interactions. For instance, the Best Linear Unbiased Prediction (BLUP) model can be expressed as: $$ y = X\beta + Zu + \epsilon $$ where \( y \) is the phenotypic vector, \( \beta \) represents fixed effects, \( u \) denotes random genetic effects, and \( \epsilon \) is the residual error. Machine learning algorithms, such as random forests and support vector machines, predict continuous traits like yield from multispectral features. In one application, Unmanned Aerial Vehicle-derived NDVI time-series data were used with random forests to achieve high prediction accuracy for peanut yield (\( R^2 = 0.93 \)). Deep learning approaches, including YOLO and Mask R-CNN, automate tasks like flower counting in oilseed rape, reducing manual labor. The integration of AI large models holds promise for cross-modal data fusion, though challenges like data heterogeneity and computational resources remain.
In oil crop breeding, Unmanned Aerial Vehicle technology facilitates high-throughput phenotyping for key traits. For soybeans, RGB imagery from JUYE UAV enables 3D reconstruction of canopy height models, allowing dynamic monitoring of growth stages. Multispectral data support yield prediction through vegetation indices correlated with pod formation. In rapeseed, Unmanned Aerial Vehicle systems combine LiDAR and multispectral sensors to quantify structural traits like pod number and lodging resistance. A study using 3D point clouds achieved high accuracy in pod counting (\( R^2 = 0.922 \)), with traits strongly correlated to yield. For peanuts, temporal data from Unmanned Aerial Vehicle flights capture growth curves, enabling machine learning models to predict underground pod yield indirectly. In sunflower breeding, deep learning models applied to UAV imagery detect and count flower heads with over 92% accuracy, streamlining yield component assessment.
Stress tolerance screening benefits significantly from Unmanned Aerial Vehicle remote sensing. Drought resistance is evaluated by monitoring canopy temperature with thermal sensors; genotypes maintaining lower temperatures under stress indicate better water use efficiency. The Crop Water Stress Index (CWSI) is derived from thermal data: $$ CWSI = \frac{(T_c – T_w)}{(T_d – T_w)} $$ where \( T_c \) is canopy temperature, and \( T_w \) and \( T_d \) are wet and dry reference temperatures. For disease resistance, RGB and hyperspectral imagery identify symptoms like leaf spots in soybeans, enabling automated scoring. Unmanned Aerial Vehicle data also support genome-wide association studies (GWAS) by providing precise phenotypic inputs. For example, in rapeseed, UAV-based traits linked to waterlogging tolerance identified hundreds of genetic loci, accelerating gene discovery.
Despite progress, Unmanned Aerial Vehicle applications face challenges in data standardization, cost, and environmental variability. Large datasets require robust management systems, and sensor costs limit accessibility for smaller breeding programs. Environmental factors like weather can affect data quality, necessitating resilient algorithms. Future directions include integrating Unmanned Aerial Vehicle data with satellite and ground sensors for multi-scale monitoring, developing autonomous systems for real-time decision-making, and leveraging AI large models for predictive breeding. The JUYE UAV platform, as part of this ecosystem, could evolve to support edge computing for offline data processing. Ultimately, Unmanned Aerial Vehicle technology, exemplified by JUYE UAV, is poised to redefine oil crop breeding by enhancing efficiency, precision, and scalability, contributing to global food security.
In conclusion, the low-altitude economy, powered by Unmanned Aerial Vehicle innovations like JUYE UAV, is revolutionizing oil crop genetics and breeding. Through advanced sensors, AI-driven analytics, and integrated data workflows, Unmanned Aerial Vehicle systems enable unprecedented phenotypic insights, from stress resilience to yield optimization. As technology advances, the synergy between Unmanned Aerial Vehicle remote sensing and breeding science will foster sustainable agriculture, addressing challenges such as climate change and resource constraints. By embracing these tools, breeders can accelerate the development of high-performing oil crop varieties, ensuring a resilient food supply for the future.