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.
