In order to achieve product servitisation, industry has in recent years made its products capable of collecting data, which is then integrated into IoT platforms.

In addition, tools are needed to transform the available data into valuable and usable information to monetise services and improve the performance of their products.

This is the context in which the MACSI project was developed, which ended in December 2023 and in which the NAITEC Technology Centre, the Azkoyen Group, Geoactio and IED participated.

The objective of the MACSI project was to create a platform that allows automatic and continuous learning, applying machine learning (ML) and deep learning (DL) techniques to the information collected by IoT devices, in order to facilitate the use of predictions in production processes through secure solutions in the cloud or in machines or devices (“edge”).

The MACSI platform has been deployed in 3 different pilots to test its transversality and usability:

• Demand forecasting in vending machines and payment methods.
• Prediction of electricity generation in photovoltaic parks.
• Prediction of bicycle parking occupancy in urban bike-sharing services.

In addition to the technical management of the project, NAITEC was responsible for the development of software management for continuous integration and distribution models, as well as for the application of continuous machine learning (MLops) strategies for the automatic retraining of predictive models based on the new data continuously collected.

The result of the project is a new continuous learning platform that receives data from IoT systems embedded in products that provide a service. Through electronics embedded in the products, data is continuously sent via IoT communication to a data centre where it is processed.

Continuous learning allows automated retraining and implementation of new models, to which is added a CI/CD (Continuous Integration/Continuous Deployment) system that allows new pipelines to be tested and implemented automatically.

Compared to traditional machine learning, where workflow steps are executed manually, MACSI’s platform deals with rapid changes in data and the business environment by aiming for the highest level of automation. MACSI offers an efficient, adaptive and scalable solution.

MACSI project has been funded by the Government of Navarra within the Call for Strategic Projects 2021.