Monitoring of agri-environmental variables in a coffee farm through an experimental IoT network to optimize decision-making by applying deep learning models

Industry 4.0, automation, and data processing are transforming business models across various sectors, including agriculture. This work focuses on the coffee sector in Colombia, analyzing the current situation and proposing 4.0 technologies as tools to improve processes such as production and the detection of nutritional deficiencies in crops. Trends are explored, and coffee farms in the department of Quindío, Colombia, are visited. Interviews with coffee growers are also conducted to gather information about their work and needs. Additionally, an experimental IoT network model is proposed to collect data on certain agro-environmental variables, which employs the LoRaWAN protocol to send and receive data between sensor nodes and the base station. The term “Digital Coffee Grower” is also defined as an artificial intelligence model that replicates or emulates the decision-making of an expert coffee grower. The implementation of technology in the coffee-growing area is reflected upon, where empirical processes are still evident, but without undermining the experience and knowledge of local coffee growers. Preliminary results are evaluated through an MLP (multilayer perceptron) neural network model. Despite initially having few data sets, the concept of “Digital Coffee Grower” promises to substantially improve the decision-making process in coffee plantations. Finally, the importance of continuing data collection and cleaning, as well as experimenting with artificial intelligence models to generate significant advances in this field, is emphasized.