Fruit cracking remains one of the most difficult challenges in orchard production. Losses can appear quickly, vary from one tree to another, and reduce both yield and fruit quality across several crops.
Within the CrackSense piloting activities, partners developed and tested a range of technologies, sensing systems, and predictive models focused on fruit cracking monitoring and risk assessment. To make these activities easier to understand and more accessible to different stakeholders, the project prepared a series of tech briefs presenting the methodologies, tools, and results behind the work.
Tech Brief 1: Spatial Decision Support System
Fruit cracking does not affect every orchard, tree, or fruit in the same way. Weather conditions, irrigation, cultivar traits, and fruit development all interact differently across growing regions.
The first tech brief introduces a Spatial Decision Support System designed to estimate cracking intensity at tree, plot, and regional levels. The system combines environmental data, sensing technologies, and artificial intelligence models into one operational framework.
The system developed within CrackSense aims to provide practical support for farmers, advisers, and agricultural stakeholders who need clearer information when making orchard management decisions. Instead of reacting after damage appears, users can access spatial predictions that support earlier interventions and better resource planning.
Tech Brief 2: Edge Processing in Orchards
Modern orchards generate large amounts of data from drones, weather stations, imaging systems, and field sensors. Processing all this information in real time can quickly become difficult, especially in rural areas with unstable connectivity.
The second tech brief focuses on edge enabled data processing for agricultural pilot studies. The framework was designed to collect, filter, and analyse heterogeneous data streams directly on edge devices instead of depending entirely on cloud processing.
This reduced data volumes by more than 50% on average while improving the quality of information at the source. Near real time machine learning and decision support could also be performed directly in the orchard environment. Faster processing means quicker decisions in situations where timing can influence irrigation management, stress detection, or cracking prevention measures.
Tech Brief 3: Thermal Point Clouds With TOMMY
Understanding cracking at fruit level requires more than general weather observations. Conditions around each fruit can vary within the canopy, making local monitoring difficult.
The third tech brief explores how RGB imaging, thermal imaging, and LiDAR technologies were combined through the TOMMY robotic system to generate temperature annotated 3D point clouds and high resolution orchard data. These datasets help researchers visualise fruit geometry, surface temperature, and moisture conditions associated with cracking development.
One challenge often raised in orchard monitoring is data availability at fruit level. CrackSense addressed this by developing multi sensor workflows capable of collecting spatially resolved datasets across different crops and orchard conditions.
The work also supports open data accessibility. As larger datasets become available, researchers and technology developers can continue improving predictive models and orchard monitoring methods across different agricultural systems.
Tech Brief 4: AI Models for Cracking Prediction
The fourth tech brief presents a multi-country framework that combines UAV remote sensing, physiological measurements, and weather data to estimate plant water status, predict cracking risk, and support orchard management.
The framework integrates thermal, multispectral, and LiDAR sensing with machine learning models that operate at tree level and orchard level. This supports more consistent monitoring across different climates and production systems.
Rather than treating each orchard as an isolated case study, the project focused on creating transferable methodologies that could support precision management across multiple regions and crop types.
Tech Brief 5: The Integrated CrackSense Solution
The final tech brief brings together the different components developed throughout the CrackSense project into one integrated solution.
The integrated system supports cracking risk estimation before visible symptoms appear. Data from sensors, physiological measurements, and environmental observations are processed through predictive models and connected to a Spatial Decision Support System web platform.
Stakeholders can select orchard fields, provide available information, and receive cracking risk estimations supported by additional Earth Observation data sources. Public datasets collected during the project are also being organised into accessible repositories linked to wider European agricultural data initiatives.
For growers and advisers, this creates a more complete pathway from monitoring to practical orchard management support. For researchers, it provides a foundation for further development of predictive tools and smart farming applications. The integrated approach also reflects one important idea running through all five tech briefs: fruit cracking cannot be understood through a single sensor, model, or dataset alone.