Research Progress on the Early Identification of Diseases and Pests of Specific High-value Crops by Micro-drones Equipped with Multispectral/Hyperspectral Sensors and Edge AI Technology

Xie Mohan

Detroit Institute of Green Industry, Hubei university of technology

Abstract：

In response to the demand for early warning of pests and diseases of high-value crops in precision agriculture, this paper reviews the research progress of integrating multispectral/hyperspectral sensors with edge AI on micro unmanned aerial vehicle platforms. Key technologies such as spectral acquisition, radiometric/geometric correction and denoising, multi-source data fusion, edge-side lightweight models and compressed deployment, edge hardware implementation, and edge-cloud collaborative architecture under payload constraints are summarized. The paper further summarizes the main bottlenecks of existing work in terms of robustness and cross-regional generalization under complex field noise, trade-offs between diagnostic accuracy and real-time energy consumption, and collaborative scheduling and continuous update mechanisms, and propose future directions for standardized data and evaluation, sensor-algorithm collaborative design, interpretable and transferable learning, and low-cost large-scale applications.

Key Words：

micro drones; multispectral/hyperspectral; edge AI; early identification of pests and diseases; high-value crops