Sequential decisions in digital agriculture - A cybernet- ICS perspective

As digital technology starts to disrupt the agricultural sector and it becomes feasible to generate appropriate data, there is a great opportunity for developing and deploying novel decision and system-theory tools to assist with the management of agricultural processes as well as whole food-production systems. Agricultural Cybernetics takes a system view of agricultural production for the analysis and design of management strategies to control, optimise, and make crop behaviours of agricultural production systems robust while exploiting the intrinsic feedback information-exchanging mechanisms that these systems exhibit. Feedback is a powerful tool, but it is also subject to fundamental limitations, which when poorly understood can lead to underwhelming results as well as wasting time and resources. Cybernetics provides proven tools and processes to answer fundamental system-related questions that are directly applicable to agriculture: Can we achieve desired crop system behaviours given particular means to act on the system? Is it possible to extract the required information from the data collected? Conversely, what is the adequate infrastructure for data generation given the information requirements related to management? Is the management strategy robust to different sources of uncertainty associated with data, mathematical models and knowledge used in the analytics? The key to answering these questions lies in understanding concepts related to the characterisation of uncertainty, information, decision problems, and how system-feedback structures process information. This paper points to key system concepts applicable to in-season management of crops whereby a sequence of decisions have to be made as part of the management strategy. This relates, for example, to activities such as irrigation, nutrition, and pest-and-disease management. It discusses the potential use of cybernetics tools for analysing and solving these management problems. The material presented has a bearing on the choice and design of infrastructure for data generation, the adoption of data analytics, the formulation of decision problems, and the potential effectiveness and robustness of management strategies, and their return on investment