A host of novel and potentially low-cost in situ observational platforms, such as drones, internet of things, and micro-electronic light-based technology systems are rapidly maturing and becoming viable alternatives to costlier traditional solutions. We develop novel integrated sensing systems to analyze (detect and quantify) chemical and biological parameters that could support appropriate interventions at the various food production stages; ranging from soil enrichment to mitigation of food safety risks.
The blooming space sector has put into orbit a plethora of earth observation satellites, which are producing massive amounts of Big Data. At the SpectraLab, we combine these data with in situ and ground truth data, to develop products and services so that we can observe and understand our planet. We achieve this by developing and applying eXplainable Artificial Intelligence algorithms to make sense of the heterogeneous Earth Observation data. The explainability of our models helps our analysts understand how they work, allows us to overcome biases and false positives, gives confidence in the outputs by providing clarity in the inference process, increases their transferability in new areas, and encourages AI adoption and acceptance.
As far as environmental monitoring concerns our team strives to deliver state of the art integrated solutions and ecosystem services for sustainable management of natural resources, environmental protection and food security. The ultimate goal is the generation of analysis ready data for several environmental indicators to support the sustaining development at local regional and international level with supporting also European policies such as Common Agricultural Policy (CAP) and Water Framework Directive (WFD). In order to achieve the above and considering that the new Era of Earth Observation satellites provide more and large amounts of free and open data our team offers a powerful digital infrastructure for automated data downloading, storing, handling and processing. Our digital pipeline exploits readily available geospatial data from several sensors (e.g., hyperspectral, multispectral, radar) and in synergy with both advance modelling techniques and physical process-based models adopts a hybrid approach, delivering high-resolution ecosystem services, including the temporal dimension according to the end user needs.
The systematic attempt of modernization and digitalization of soil science with the adoption of eco-friendly and cost effective methods of soil analysis have brought soil spectroscopy to the forefront. Its non destructive nature and its proven reliability for the estimation of soil properties, attracted the attention of the research community, and the advances on quantitative soil spectroscopy for the prediction of soil properties is continuing to grow. Recent advances in the photonics industry with the development of miniaturised spectroradiometers with high signal-to-noise ratio that can be mounted to different platforms, unveil new possibilities of developing new services based on the synergy of different sensors. Having recognized this potential from an early stage, our lab serves the soil spectroscopy community as a regional champion laboratory of FAO and also holds the IEEE-P4005 “Standards and protocols for soil spectroscopy” secretariat position.