Precision Irrigation thanks to Artificial Intelligence

A new soil moisture monitoring system was patented at the University of Bologna. It consists of a sensor grid for measuring soil moisture in both two and three dimensions, and a specialised machine learning algorithm based on a neural network.

This innovative technology was tested in a golden kiwi field in Brisighella, near Ravenna. It achieved a significant reduction of 41% of water used for irrigation between June and October.

The patent is a result of a collaboration between researchers at the Department of Computer Science and Engineering and the Department of Agricultural and Food Sciences. It allows for the development of an irrigation system designed to maintain an optimal degree of moisture in agricultural soils.

Monitoring soil moisture is pivotal for optimising irrigation and crop performance. Once installed within the targeted area, the 2D and 3D sensors detect soil’s characteristics and, with the aid of a specific machine learning algorithm, they generate an extremely detailed soil profile.

This system enables the monitoring of soil moisture levels within specific periods. The system’s predictions reach a level of detail measured in centimetres, eliminating the need for physical soil modelling, and are updated as the soil behaviour changes.

The technology provides precise data and information to guide both quantitative and qualitative choices of irrigation. The system can provide accurate estimates of moisture level, even for for areas beyond the sensor's immediate vicinity.