Development of an integrated system of solution for decision support of crop health diagnosis : case of a machine learning enabled unmanned aerial vehicle

Abstract

The agricultural sector developed a need to utilise technology to make informed decisions about crops. Remote sensing technologies, which typically utilises satellite, airborne, or ground-based sensors, has been increasingly used in precision agriculture lately. However, Unmanned Aerial Vehicles (UAVs) or drones have become a more cost-effective and versatile solution, providing higher resolution imagery and greater flexibility in flight time, frequency, and crop visibility. The project opportunity stems from the growing usage of UAVs in agriculture. The problem statement addresses the need for a comprehensive framework for selecting, designing, and implementing a crop monitoring UAV system, which has not yet been identified. This project developed an integrated system of solution for a machine learning enabled drone that combines different attributes into a unique solution. The literature review highlighted several aspects to consider for a drone remote sensing system and illustrated how such a system fits into precision agriculture applications. Required equipment and technologies identified for a system include a machine learning enabled UAV, control systems, sensors, and data processing tools. A case study research approach is deemed appropriate as it allows for the review of literature and available solution options before designing a solution. Attributes were identified and modelled to create a unique decision support framework for a crop monitoring solution system following their relevance and combinatorial characteristics. The integrated system is divided into three solution paths, each with critical user decisions and recommended selection processes. Possible solutions are categorised by farm and aircraft specifications to facilitate simpler selection. The research objectives were addressed through the identification of these attributes and through designing the main decision systems along with the categorisation of potential solution options. A case study research approach is deployed throughout the project to allow for the integration of literature and available solution options to the holistic system and each smaller decision sub-system. The methodology was iterated within each main decision path to define and analyse a unique case for each decision system and create a solution based on the information available for the specific decision system. Despite this research being skewed towards qualitative investigations, some quantifications from the research findings include that from the 31 UAV models considered for analysis, they can be categorised into six categories relating to UAVs characteristics and two categories related to the farm characteristics. The categories are designed to group together those aircrafts with similar characteristics or specifications, to allow for an easy reference and selection by the user. The presented solution addresses the complexity of the system and identified literature gaps through an encompassing and integrated system of solution. Future work includes creating a comprehensive database that includes all possible solution options and developing a functioning decision support system based on the developed solution system.