Initiative funded by the Spanish Ministry of Industry, Trade and Tourism within the support programme for clusters of innovative companies to contribute to improving the competitiveness of Spanish industry, and with the support of the European Union through the Recovery, Transformation and Resilience Plan.

Reference: AEI-010500-2023-17

Project type: Industrial Research

Objective: the project aims to provide Spanish SMEs in the industrial sector with a tool to automate the process of capturing and generating 3D data to create digital twins, enabling them to accelerate the digitisation and modernisation of their infrastructures towards the new Industry 4.0 paradigm. Second phase.

The second mission aims to continue the efforts initiated in the first part of the project, addressing the needs and opportunities identified during the execution of the first phase. Specifically, it seeks to enhance the tool's capabilities in object detection and automatic geometry reconstruction by incorporating new state-of-the-art techniques and algorithms. In the past year, the field of artificial intelligence has seen unprecedented advancements, with many new models and approaches that have the potential for application in the project's tasks.

By adopting the missions outlined, this new phase of the project assumes the inclusion of new state-of-the-art techniques and the enabling of new functionalities that are highly valuable for the tool and the research line started in the first phase. This improvement is proposed through three fundamental pillars.

1. Expansion of the AI model's capacity through the provision of synthetic data. It is a known fact that the performance of artificial intelligence models is only as good as the data used in their training. Therefore, collecting a sufficiently large database becomes a fundamental necessity to obtain satisfactory results. This mission presents several associated challenges since synthetic data must emulate the characteristics of data obtained in a real measurement process, which has a significant impact on the results.
2. Enabling advanced detection capabilities through intensified model training processes. The increased abundance of data achieved with the accomplishment of the previous point will allow for a much more comprehensive approach to training artificial intelligence-based detection models. Models can be trained with a significantly larger number of cases, starting from scratch or additionally incorporating other available databases. The training process is a key step that involves significant computational effort to carry out successfully. At this point, the capacity and knowledge of system architectures are crucial to optimizing these processes efficiently and within reasonable timeframes.
3. Incorporation of new techniques for geometric reconstruction and BIM (Building Information Modeling). This is one of the most remarkable aspects of the project, as it represents the final step in delivering digitized industrial asset files, resulting in significant cost and time savings. It will be addressed by incorporating new modeling techniques in addition to those already implemented during the first phase. During the execution of the first phase of the project, it was concluded that no reconstruction technique is superior to the others by itself, but rather, depending on the specific case, some techniques are more promising than others. Therefore, it is important to provide the tool with greater versatility in the range of options to fulfill this function, which is highly relevant for an industrial market characterized by its heterogeneous nature.

In collaboration with PLAIN CONCEPTS, Clúster onTech Innovation and United ITS