TWINFARMS

The concept of digital twins, derived from engineering, has enabled improvements in many areas. With the increasing number of sensors in farms, the amount of data collected, combined with already acquired knowledge, could support the agroecological transition in the face of climate change.

An agricultural digital twin can help farmers adjust their tactical choices based on the state of their crops or animals as measured by sensors or benefiting from external data sources (e.g., satellite images) when their technical expertise is insufficient to deal with unpredictable climatic sequences. In a strategic approach, a digital twin can help farmers project into the future to anticipate desirable changes in their farming systems.

To develop this type of decision-support tool, TWINFARMS faces several challenges:

• an increasing amount of heterogeneous, incomplete data/knowledge with restricted access, hindering practical application;
• a combination of different modeling paradigms (physical, biological, economic models, and machine learning) that remains to be built in order to obtain effective and robust predictions;
• taking into account the unique individual characteristics of each farm and the user's needs is necessary for developing a relevant digital twin that provides actionable and useful information.

To address these challenges, TWINFARMS will build nine demonstrators where digital twins will be assessed for their added value in assisting tactical or strategic choices. Their scales will range from plots to farms and their environment, from daily recommendations to multi-annual projections.

Beyond specific difficulties, building digital twins for these demonstrators will raise questions that will be collectively addressed to bring out generic components useful for future developments and ensure the transfer of models from one farm to another. Thus, the project aims for model adaptability between farms despite different configurations or temporal scales.

To facilitate future developments of agricultural digital twins, dissemination is crucial and central to this project through scientific and technical articles, as well as webinars and training sessions. Furthermore, a reflection will be carried out on the economic models that can be developed to monetize the services offered by digital twins in the agricultural field and on the perspectives of this approach on a larger scale.

TWINFARMS relies on a large consortium of research and development partners from research and educational institutes, technical institutes, a robotics company, an agricultural cooperative, and a farmer association with expertise in mathematics, computer science, agronomy, and animal science to tackle the challenges of consolidating innovative digital infrastructures and data sharing.