Periodic Reporting for period 1 - AgriDataValue (Smart Farm and Agri-environmental Big Data Space)
Berichtszeitraum: 2023-02-01 bis 2024-07-31
Smart Farming 4.0 faces a number of barriers and challenges. One of the most important is the agricultural data availability and heterogeneity. Sensors’ data generated locally are often more precise and valuable, in comparison to global, national or regional datasets. On the other hand, the combination of local data with datasets from a broader area, allows for comparison of crop and stock raising conditions, automated identification of crop or stock diseases or production delays, well before it could be observed in the area of interest, while it offers significant support for informed decisions related to agricultural production adaptation to climate change. Moreover, the utilization of AI models for combining and upscaling heterogeneous agricultural data could generate more accurate knowledge, covering larger areas with reasonable overall cost.
The upscale of agriculture sensor data to witness the benefits of knowledge in Smart Farming is imperative. Within the AgriDataValue project, in parallel to sharing data in a FAIR way , a new, open and radically different approach is followed, allowing for utilizing the data in their original location, rather than copying the data in a new location. We base our research on mature IoT results and state-of-the art solutions such as the such as BDVA/DAIRO , GAIA-X and the IDSA and leverage on emerging technologies, such as Federated Deep Machine Learning (FDML) and human eXplainable AI (XAI) to ensure data sovereignty and knowledge creation, 5G communications and edge cloud computing to introduce local/on-device intelligence, Blockchain/Distributed Ledger Technology (DLT) to give real incentives and new business models for data and knowledge sharing.
AgriDataValue (or simply ADV) aims to strengthen the smart-farming capacities, competitiveness and fair income by introducing an innovative, open source, intelligent and multi-technology, fully distributed Agri-Environment Data Space (ADS). To achieve technological maturity, fast and massive acceptance, AgriDataValue adopts and adapts a multidimensional approach that combines state of the art big data and data-spaces’ technologies (BDVA/ IDSA/ GAIA-X) with agricultural knowledge, monetization, new business models and agri-environment policies, leverages on existing platforms, edge computing and network/ services, and introduces novel concepts, methods, tools, pilot facilities and engagement campaigns to go beyond today’s state of the art, perform breakthrough research and create sustainable innovation
The upscale of agriculture sensor data to witness the benefits of knowledge in Smart Farming is imperative. Within the AgriDataValue project, in parallel to sharing data in a FAIR way , a new, open and radically different approach is followed, allowing for utilizing the data in their original location, rather than copying the data in a new location. We base our research on mature IoT results and state-of-the art solutions such as the such as BDVA/DAIRO , GAIA-X and the IDSA and leverage on emerging technologies, such as Federated Deep Machine Learning (FDML) and human eXplainable AI (XAI) to ensure data sovereignty and knowledge creation, 5G communications and edge cloud computing to introduce local/on-device intelligence, Blockchain/Distributed Ledger Technology (DLT) to give real incentives and new business models for data and knowledge sharing.
AgriDataValue (or simply ADV) aims to strengthen the smart-farming capacities, competitiveness and fair income by introducing an innovative, open source, intelligent and multi-technology, fully distributed Agri-Environment Data Space (ADS). To achieve technological maturity, fast and massive acceptance, AgriDataValue adopts and adapts a multidimensional approach that combines state of the art big data and data-spaces’ technologies (BDVA/ IDSA/ GAIA-X) with agricultural knowledge, monetization, new business models and agri-environment policies, leverages on existing platforms, edge computing and network/ services, and introduces novel concepts, methods, tools, pilot facilities and engagement campaigns to go beyond today’s state of the art, perform breakthrough research and create sustainable innovation
Within the reporting period, we have designed and implemented a flexible federation of Agri-environment Data Space in the form of an AgriDataValue (ADV) platform. The ADV platform is a Platform of Platforms based on state-of-the-art technologies able to store, combine, upscale and share sensors’, drones’ and satellite agri-environment data, interfacing various heterogeneous external platform (i.e. Copernicus Land Monitoring Service platform), along with food value chain traceability. Interviews, questionnaires and co-creation workshops with ADV beneficiaries/pilot owners/end users/farmers have facilitated the use cases requirements extraction. At least 15 state of the art technologies (e.g. Federated ML, IDSA adapters, secure storage, blockchain, smart contracts, IoT processing, in memo databases, drones and satellite data geolocation, DevOps, Kubernetes, OAuth 2.0 etc) have already been integrated.
AgriDataValue platform has already implemented standardized adapters (based on AIM and IDSA) to collect, store and share in (near) real-time measurements from more than 7 different (pre-existing) IoT and satellite data platforms. More than 30 different data types ranging from micro-clima, air-quality, soil, leaf, and greenhouse sensors, drones’ and satellite monitoring services have been combined, while 7 semantic interoperability mechanisms from previous projects and initiatives have been analysed.
AgriDataValue core capabilities are based on three breakthrough technologies: a) Utilise blockchain and smart contracts to enable traceability of data and trained ML models sharing, b) advanced confidentiality-preserving Federated Deep Machine Learning (FDML) to train ML models, without moving, copying or disclosing any data from their original locationand c) human-eXplainable AI (XAI) technology in the form of semantic graphs in a symbiotic relationship between humans and an intelligence-capable system, so that collaborative cooperatives, farmers’ and decision makers’ will be able to understand the reasoning behind a recommendation.
AgriDataValue pilots extend in at least 181,000ha, with a least 25 different crop types and 5 animal species from 9 EU countries. At least 20 end-user organizations and more than 2,200 farmers are already informed. Beyond the pre-existing IoT and satellite data platforms that have been integrated, new IoT equipment (SynField/ SynAir) has been installed in most AgriDataValue pilots. As a result, we are already collecting at least 160 datasets (40 sources and 20 locations x 8 measurement types each) and more will be added.
AgriDataValue platform has already implemented standardized adapters (based on AIM and IDSA) to collect, store and share in (near) real-time measurements from more than 7 different (pre-existing) IoT and satellite data platforms. More than 30 different data types ranging from micro-clima, air-quality, soil, leaf, and greenhouse sensors, drones’ and satellite monitoring services have been combined, while 7 semantic interoperability mechanisms from previous projects and initiatives have been analysed.
AgriDataValue core capabilities are based on three breakthrough technologies: a) Utilise blockchain and smart contracts to enable traceability of data and trained ML models sharing, b) advanced confidentiality-preserving Federated Deep Machine Learning (FDML) to train ML models, without moving, copying or disclosing any data from their original locationand c) human-eXplainable AI (XAI) technology in the form of semantic graphs in a symbiotic relationship between humans and an intelligence-capable system, so that collaborative cooperatives, farmers’ and decision makers’ will be able to understand the reasoning behind a recommendation.
AgriDataValue pilots extend in at least 181,000ha, with a least 25 different crop types and 5 animal species from 9 EU countries. At least 20 end-user organizations and more than 2,200 farmers are already informed. Beyond the pre-existing IoT and satellite data platforms that have been integrated, new IoT equipment (SynField/ SynAir) has been installed in most AgriDataValue pilots. As a result, we are already collecting at least 160 datasets (40 sources and 20 locations x 8 measurement types each) and more will be added.
- Interoperable/federated “Platform of Platforms”. Designed and implemented a modular framework of derated platforms that may accomodate different types of data at interoparable data silos. The interoperability is achieved utilising international Data Spaces Association (IDSA) adapters.
- Agri-environment ADS Decision Support System (DSS). Designed and implemented a DSS mechanism that is based on Federated Deep Machine Learning (FDML) mechanism. This mechanism enables AI decisions based on data that are stored remorely. The DSS has already been inmplemented and various ML models that iplement project Use cases are progressively build.
- Human-interpretable Agri-environment AI Models. Worked towards a mechanism that will provide human explainable justification of the system generated decisions. The XAI mechanism will be integrated in the next project phase.
- Novel approaches for in-situ real-time IoT/EO data processing. Worked towards combining various satellite, geolocated drones and IoT data in ML models to provide decision that are based on multimodal data.
- By-design food products/ supply chain traceability. Worked towards blockchain technologies to implement food products traceability. This technology will be integrated in project pilots in the next project phases/
- Use Cases/Apps to support agri-economy/climate monitoring. Worked towards imlementing various use cases that demonstarte the value of the platform.
- Agri-environment ADS Decision Support System (DSS). Designed and implemented a DSS mechanism that is based on Federated Deep Machine Learning (FDML) mechanism. This mechanism enables AI decisions based on data that are stored remorely. The DSS has already been inmplemented and various ML models that iplement project Use cases are progressively build.
- Human-interpretable Agri-environment AI Models. Worked towards a mechanism that will provide human explainable justification of the system generated decisions. The XAI mechanism will be integrated in the next project phase.
- Novel approaches for in-situ real-time IoT/EO data processing. Worked towards combining various satellite, geolocated drones and IoT data in ML models to provide decision that are based on multimodal data.
- By-design food products/ supply chain traceability. Worked towards blockchain technologies to implement food products traceability. This technology will be integrated in project pilots in the next project phases/
- Use Cases/Apps to support agri-economy/climate monitoring. Worked towards imlementing various use cases that demonstarte the value of the platform.