At RDT, we continue driving the development of innovative solutions for industrial digitalization. One of these initiatives is RECAP (Repair of Critical Components through Hybrid Additive-Subtractive Technologies and Intelligent Processes), a technological platform designed for the monitoring, analysis, and intelligent diagnosis of welding processes through computer vision and artificial intelligence.
Weld quality is a critical factor across numerous industrial environments. However, inspection processes are often performed after the welding operation has been completed, making early defect detection difficult and increasing the costs associated with rework and production issues. RECAP was created to transform this approach by incorporating real-time analysis capabilities, enabling more efficient, traceable, and automated processes.
The solution is structured around three main development stages.
Real-Time Monitoring and Data Acquisition
The first stage focuses on the development of a data acquisition and streaming platform capable of capturing and visualizing the welding process using Xiris industrial cameras. This infrastructure enables the collection of high-quality images of the welding arc while recording key process information for subsequent analysis.
One of the distinguishing features of this solution is its Open Source-based approach, which offers greater integration flexibility compared to proprietary alternatives and facilitates adaptation to a wide range of production environments.
Defect Analysis and Data Annotation
The second stage focuses on generating knowledge from the captured data. To achieve this, RECAP incorporates a dedicated tool for the annotation and classification of welding discontinuities and defects.
This infrastructure will enable the creation of high-quality datasets that serve as the foundation for training advanced computer vision models. The ability to efficiently navigate and analyze image sequences will facilitate the identification of patterns associated with different types of anomalies and welding defects.
Intelligent Diagnosis through Artificial Intelligence
The final stage involves the integration of Deep Learning models capable of analyzing captured images and automatically detecting potential issues during the production process itself.
Thanks to these capabilities, the system can act as an intelligent quality-control assistant, providing real-time alerts about potential defects and supporting manufacturing and inspection teams in their decision-making processes. The objective is to reduce inspection times, minimize human error, and improve overall process efficiency.
With RECAP, RDT continues exploring the potential of computer vision, advanced data analytics, and artificial intelligence to promote a more connected, efficient, and competitive industry, where systems are capable not only of recording information but also of interpreting it and transforming it into valuable knowledge for informed decision-making.
