FOR 5351:  KI-FOR Automation and Artificial Intelligence for Monitoring and Decision Making of Horticultural Crops (AID4Crops)

• Subject Area: Computer Science, Systems and Electrical Engineering ; Agriculture, Forestry and Veterinary Medicine ; Geosciences
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 459376902

Achieving sustainable agriculture is a central ambition of the United Nations' Agenda 2030. Sustainable agriculture systems are an essential instrument to achieve the goal of zero hunger (SDG 2). Despite this, there is still considerable uncertainty on how to best achieve sustainable agricultural systems. Precision agriculture has emerged as an integral strategy to enabling sustainable agricultural systems. It harnesses modern sensing and monitoring technology to inform data-driven optimization of agricultural management. However, the development of precision agriculture has brought with it significant challenges ranging from what features of a cropping system should be sensed, how frequently they should be sensed through to how to incorporate this information into the decision management process of farmers.An area of research that has been largely neglected is how to jointly optimize the measurements of plants that enable better management decisions. To date, research on applying Artificial Intelligence (AI) to the horticultural sector has treated the tasks of sensing and management as decoupled processes. The reasons for this are two-fold. First, decision-making in horticulture has remained relatively crude as it usually makes use of rather coarse inputs such as temperature, lighting and CO2 and does not incorporate finer-grained information such as the plant status (phenotyping). Second, AI-based sensing algorithms have only recently attained sufficient performance to provide plant status information (e.g. fruit counts and ripeness estimation). These two issues are further confounded as to date there has been no joint feedback from researchers of AI-based sensing algorithms regarding "what can be sensed" and researchers of automated decision management regarding "what should be sensed".The AID4Crops research unit will provide a step change in precision agriculture. It will develop novel AI algorithms that can optimally sense and forecast the status of the plant and then couple this with decision analysis. This will lead to an unprecedented two-way coupling of monitoring and decision-making in precision agriculture and horticulture.

• DFG Programme: Research Units

Projects

• Coordination Funds (Applicant McCool, Christopher )
• Efficient Sensing for Long-Term Crop Monitoring (Applicant Bennewitz, Maren )
• IP1: Incorporating Short-Term Spatial-Temporal Information for Robotic Sensing (Applicant McCool, Christopher )
• IP2: Exploiting Repeated Data Acquisitions for Improved Long-term Monitoring Capabilities (Applicant Stachniss, Cyrill )
• IP4: Forecasting Phenotypes based on Management Decisions (Applicant Gall, Jürgen )
• IP5: Uncertainty meets explainability -- Combining Uncertainty Quantification and Explainable Machine Learning for Crop Monitoring (Applicant Roscher, Ribana )

• Spokesperson: Professor Dr. Christopher McCool