AI-SPRINT aims to democratise access to AI technologies, by simplifying and accelerating the development of artificial intelligence applications through edge computing. The tools developed during the project were validated by three industrial case studies.
Three case studies, in personalised healthcare, agriculture 4.0 and maintenance and inspection, showed the potential and effectiveness of solutions that can be developed with the AI-SPRINT Studio platform made available on the marketplace AI-on-Demand.
In the area of personalised healthcare, a system was developed during the project to connect wearable devices (such as smartwatches) directly with medical specialists, who, thanks to an analysis of the patient's parameters via AI, can be alerted at the very moment that cardiac abnormalities occur. This experimentation will lead to the launch of a start-up.
As far as agriculture 4.0 is concerned, experimentation has led to the development of a monitoring system on agricultural machinery that uses video cameras to enable the effective distribution of plant protection products in real time with a consequent reduction in chemical pollution.
In maintenance and inspection activities, on the other hand, artificial intelligence has been used to carry out immediate reconnaissance of technical problems in wind power plants, thus reducing the need to send out teams of technicians and lowering maintenance costs.
Over the course of three years and in an attempt to democratise access to cutting-edge AI technologies and promote a deeper understanding of their components, the AI-SPRINT project launched a series of free massive open online courses (MOOCs). MOOCs enable and will enable the acquisition of the knowledge and skills needed to use artificial intelligence effectively in various contexts.
“Artificial intelligence is becoming pervasive and the global market for AI software platforms is expected to grow significantly,” explains Danilo Ardagna, professor of IT infrastructures at the Politecnico di Milano. “Many of the advantages of this evolution will come from using IT resources at the periphery of the network, i.e. where the source data are produced. Many companies are considering the use of edge computing for online data collection, processing and analysis to reduce application latency and the volume of data transferred. A growing number of cases of use can benefit from artificial intelligence applications that range from edge to cloud infrastructures, exploiting the resources available in the computing continuum.”
The AI-SPRINT Studio design environment, developed during the project, allows programmers to balance application performance (such as latency and throughput) and energy efficiency, with the accuracy of AI models, within a secure and private environment. The suite also offers tools for profiling AI applications and training models, enabling developers to deploy deep networks in different computing environments and train high-quality models even without in-depth knowledge of machine learning.
The final phase of the project was characterised by a review held at the headquarters of the European Health and Digital Executive Agency (HaDEA), during which the AI-SPRINT team presented the results achieved.
The Politecnico di Milano was supported by the Fondazione Politecnico for administrative activities and Cefriel for some of the research and development of the implemented solutions. AI-SPRINT involved a total of 11 partners including the Barcelona Supercomputing Center, the Technische Universitaet Dresden, the Universitat Politecnica De Valencia, Gregoire, Beck Et Al Services, Cloud & Heat Technologies, 7BULLS, Transformation Technology For Analysis, Trust-IT Services and IDC Italia.