Results of the BERREKIN Advanced Manufacturing Project

2024/05/22 11:40:00 GMT+2
The BERREKIN project concludes with the development of a bilingual digital assistant in Basque and Spanish based on Artificial Intelligence that resolves doubts about Fagor 8065 CNC programming and IKASMAK 5.1 machine start-up.

In the BERREKIN project, coordinated by Vicomtech, an assistant has been developed to help operators by answering specific questions about CNC code, conversational cycles and simulation of machining operations, as well as guiding the student in front of the machine.

IMH Campus has created a corpus of questions and answers and machine guidance procedures in both Spanish and Basque based on FAGOR manuals and numerical control.

The assistant has been validated on a simulator and machine with IMH Campus students from different courses: intermediate and advanced manufacturing training cycles, and with students in the second year of Dual Engineering in Innovation.

 

BERREKIN project development

A consortium coordinated by Vicomtech and made up of different technology centres and universities has researched and developed a digital assistant with the aim of facilitating the implementation of natural language interaction interfaces between operators and industrial production machines. Based on existing documentation and technical information and using the most advanced Artificial Intelligence technologies, BERREKIN has developed a prototype of a digital assistant in both Basque and Spanish to support operators in industrial manufacturing environments. This prototype can be used both for training purposes and for on-site operational assistance related to Computer Numerical Control (CNC) programming for the manufacture of parts on milling machines. In addition, the wizard displays full information on the screen and allows both text and voice interaction.
The assistant has the ability to assist operators by providing answers to questions about specific aspects of the manufacturing process, such as the meaning of certain programming commands. In addition, it can guide them step-by-step through different procedures, e.g. by simulating a programme. The conversational assistant has been fed and trained with information from the milling machine manuals and prior knowledge provided by technical experts.
The use case implemented to evaluate the usability of the wizard consisted of manufacturing a part using various parameters and commands of the numerical control of the milling machine. The manufacturing process of a part consists of several phases, ranging from the understanding of its drawing to the definition of the machining procedure and the implementation, simulation and execution of the CNC program to build the part. The assistant developed supports the operators throughout the entire process and provides them with the information and knowledge necessary to resolve any possible doubts in each of the phases, being a key component that contributes to enhancing a people-centred approach in the industrial environment.
One of the innovative features of BERREKIN is the possibility to improve the digital assistant. In case the answers are not adequate, the assistant provides a "dislike" option and collects the data so that industry professionals can improve the previously established answers.
In the development of the BERREKIN project, each partner has played different roles. Vicomtech has been the project coordinator and together with UPV/EHU and Tekniker have researched and developed incremental learning techniques for dialogue systems and Question Answering (QA) systems. These techniques have been developed together with interfaces that allow involving the "Operator in the Loop" to adapt, personalise and provide knowledge to conversational assistants in industrial manufacturing environments. In addition, Ikor and Vicomtech have also investigated the application of the "Operator-in-the-loop" paradigm for such conversational human-machine interactions in distributed computing or edge computing environments.
Another important aspect of the project has been the compilation of resources for Basque, on which IMH Campus and UZEI have worked mainly, and which have allowed the implementation of a prototype of a conversational assistant with the "Operator in the Circuit" also in that language.
To conclude the project, IMH Campus students were in charge of its validation.
  • This project has been funded by SPRI within the framework of the Elkartek Programme, which supports collaborative research by RVCTI agents in the strategic areas of RIS3 Euskadi.

More information on the BERREKIN project

Advanced manufacturing projects of the IMH Campus