SONIC – 5G-based AI inference services for non-destructive laser ultrasonic testing

Non-destructive testing (NDT) is a key component in ensuring quality, safety, and sustainability in industrial manufacturing and maintenance processes. Precise and reliable testing methods are indispensable, especially for safety-critical components such as welds or load-bearing components. At the same time, many industries are facing the challenge of digitizing their testing processes, making them more flexible and implementing them more efficiently. The current state of the art is often characterized by locally bound test setups, high investment costs, and a strong dependence on manual data evaluation.

The SONIC project is developing a modern, digital solution that significantly simplifies this process. Modern NDT imaging techniques such as laser ultrasound generate large amounts of high-resolution sensor data. Their timely and reliable evaluation requires a powerful technical infrastructure that supports both mobile data acquisition and data-intensive processing. The SONIC project addresses these requirements through the targeted combination of three key technologies: 5G communication, cloud-based high-performance computing, and artificial intelligence.

5G for fast data transmission

5G technology enables low-latency, reliable, and secure transmission of large amounts of sensor data directly from mobile testing systems. Laser ultrasonic measurement data is captured in real time and transmitted to central computing resources via 5G networks. This allows inspections to be carried out regardless of location, without the need for extensive local computing infrastructure. 5G thus forms the basis for flexible, mobile NDT scenarios and seamless integration into existing industrial processes.

Cloud computing for high computing power

The data-intensive processing and reconstruction of NDT measurement data takes place in a cloud-based HPC environment. The use of scalable cloud resources provides the high computing power required for complex reconstruction and analysis processes on demand. This enables short processing times, high scalability, and economical use of powerful computing infrastructure without industrial users having to invest in costly HPC systems themselves.

Artificial intelligence for precise results

Centrally hosted generative AI models take over the inference, analysis, and interpretation of the transmitted sensor data. The AI reconstructs high-resolution three-dimensional models from the measurement data and supports the automated detection and evaluation of material defects. The use of modern AI architectures achieves reproducible, precise, and objective test results that complement human expertise and significantly accelerate testing processes.

Through the integrative interaction of these three technologies, SONIC creates a novel, service-oriented infrastructure for non-destructive testing. The solution enables continuous testing of industrial components throughout the entire product life cycle. At the same time, it supports new business models such as AI-based testing services and remote diagnostics.

SONIC thus makes a significant contribution to the digitalization, sustainability, and competitiveness of industry. Technological barriers are being broken down, testing processes are being made more efficient, and access to powerful AI and computing infrastructure is being simplified through modern communication architectures. The project is positioned as an innovative flagship project with high technical relevance and sustainable economic benefits.