Towards a Phenotype Classification of Agricultural Robots

The use of robot systems in agriculture has heavily increased in recent years. Together with the Internet of Things and Big Data, these technologies can represent an opportunity to make food production more sustainable. Being able to carry out hard physical tasks hitherto covered by operators and potentially ensuring night work cycles, these technologies accompany many advantages related to reducing work-related accidents and the autonomous conduction of operations. Despite these opportunities, their large-scale diffusion is limited today by a lack of clarity and exhaustiveness in the regulatory framework that is intrinsically tied with ethical and legal issues concerning the management of robots. Existing legislation places obligations related to machine registration and human supervision in operations, but several issues concerning the use of such technologies still have to be addressed, such as legal responsibility, privacy issues, and data management. To date, a clear and agreed classification of the various types of agricultural robots is missing. The main goal of this study is to classify the many types of robots available (i.e., phenotyping) to obtain an exhaustive and efficient information framework able to facilitate the production of ad-hoc legislative supports and promote primary market segmentation practices. We used an observational survey method that involved a web search of robots and systems, which were described with qualitative variables based on specific criteria describing their usage in farms. Collected data was hence used to picture homogeneous groups of robots. The classification was performed with a double step cluster analysis based on the nominal descriptive variables and a factor analysis used to reduce classification redundancy. Five clusters of robots have been identified, opening up the possibility to set specific regulatory policies and market strategies based on recurring characteristics within the identified clusters.