* Decision support systems (DSS) for agriculture 4.0
Our approach involves a training phase based on high-performance hardware, followed by inference carried out as far as possible on edge computers, also in real time, using machine learning accelerators and other specialized AI devices.
The DSS will first be applied and tested at an experimental farm and, once an appropriate level of development has been reached, tested at several pilot farms, with a view to disseminating new developments.

* Smart agricultural machinery and low-cost precision agriculture
Our aim is to use recent developments in microelectronics to implement affordable devices for precision farming, which allow full accessibility and use of data at a fraction of the cost of official listings. In particular, we are studying the feasibility of low-cost implementation of RTK (Real Time Kinematic) solutions for accurate geolocation in the field and the use of wireless sensors of various kinds in the field or on board agricultural vehicles. We are also interested in the development of libraries for data transfer that allow a completely open workflow, without using proprietary software.
Low-cost solutions for precision farming will be validated at an experimental farm and then tested at several pilot farms, in view of detailed design.

At CRS4 we have 30 years of experience in the field of computational mathematics and numerical simulation focused on the study of engineering, industrial and environmental science applications. We specialize in computer analysis of large-scale problems (HPC, High Performance Computing) for which we use both optimized mathematical algorithms and high-performance hardware platforms (parallel machines, accelerators, etc.). We are convinced that the optimal implementation of complex machine learning algorithms today requires the adoption of large-scale numerical algebra technologies alongside the use of commonly used traditional libraries. More recently, we have focused our research on edge computing and the use of extremely low-cost, low-power single-board computers suitable for on-site analysis of data collected from remote sensors, in the field or on agricultural machinery. This latest-generation hardware makes real-time implementation of artificial intelligence algorithms and decision support tools feasible, which up to now has only been possible on remote machines and in the cloud.

We collaborate with a small number of end-users from relevant agricultural sectors, who participate in identifying priority agriculture 4.0 technologies and make on-farm experimentation possible by providing agricultural machinery, their day-to-day experience as operators in the sector and the necessary information on field practices. As our mission is highly implementation-oriented, we also consider it essential to cooperate with research institutes, public agencies and professional organisations operating in the agricultural sector.