Application of Unmanned Aerial Systems for Precision Weed Detection and Management.

Unmanned aerial systems (UAS) that include unmanned aerial vehicles (UAVs), sensor suits and data management framework are gaining significant popularity in recent times as a potential solution for enhancing the efficiency of agronomic management practices. Timely and efficient assessment of weed issues is a critical first step in implementing effective weed management interventions. The current methods of human field assessments are inefficient and inaccurate; advancements in the UAS technology can provide a convenient solution in this regard. The objectives of this project are to characterize spectral and spatial features relevant for assessing weed infestations using UAS-based sensors under varying plant growth, soil, and environmental conditions and develop advanced algorithms and approaches for weed species detection and differentiation. Replicated field experiments are being conducted at the Texas A&M field research facility near Snook, TX in this regard. RGB, multi-spectral as well as hyperspectral images are being obtained on targeted vegetation to understand the spectral characteristics. Weed detection using indices developed based on RGB ratios was consistent for certain species but not for many others. Hyperspectral data obtained using a spectroradiometer was useful in distinguishing some other species. Preliminary results suggest that robust detection and differentiation of crop and weed species will require a dynamic integration of plant reflectance, spatial analysis as well as knowledge integration. More research is underway to understand and utilize the UAS platforms for precision weed management applications.