Intelligent welding process monitoring based on Vision AI, that automates QA & QC for laser cutting, laser welding and additive manufacturing processes. This is an off-axis Industry 4.0 monitoring solution, reassuring and alerting in real-time the end user for the quality of thermal laser processes. The system’s capabilities include tensile shear strength classification, electrical resistance classification, penetration status, burn-through/pits detection.
Blowholes
Insufficient Penetration
Vision AI System Components
Under its hood, PerCV.ai-powered Computer Vision AI and ML algorithms are processing the visual data captured by two vision sensors (cameras) attached to the laser head:
- An Optical camera inspecting the process & extracting prominent features DURING and AFTER the welding: ➩ on-line to reveal morphological features of the plasma distribution➩ off-line monitoring the part being cooled down, for post-process seam inspection and calculation of the geometrical features
- A Thermal camera (IR) inspecting the thermal distribution in real-time, performs noise cancellation, segments the thermal signal and extracts prominent features, even for highly-reflective materials such as aluminum.
PerCV.ai platform for on-device Vision Intelligence is bringing together all the building blocks necessary for powering the computer vision-based Laser Process Monitoring System, including:
- Weld segmentation algorithm for on-line process
- Detection / segmentation the stitching signal in off-line process
- Estimation of geometrical and intensity features
- Geometrical: area and perimeter, width/length, Aspect ratio, Extent, Eccentricity
- Intensity: Average, Maximum and Minimum intensity
- Robust statistical params: Median and Majority intensity
Vision sensors are connected via USB to an edge device Nvidia Jetson Xavier AGX with high-end GPU and multi-core CPU.
The edge device connects (via COM or digital) to the laser head and also provides Gigabit Ethernet connectivity to external devices/systems.
The PerCV.ai AI Monitoring User Interface is providing analytics for monitored laser process. Furthermore, all produced metadata can be easily transmitted to your existing software stack via the industry standard API and MQTT protocols.
-
Quality OK
Optical Camera
Thermal Camera
-
Quality FAIL
Optical Camera
Thermal Camera
Vision AI Benefits
-
Cost Efficiency
Minimise material scrap, labour costs, process completion time through real-time defect detection
-
Scalability
Compared to other QA methods, such as human inspection or electromagnetic transduction
-
Easy Integration
Off-axis solution supporting variety of industries (automotive, consumer electronics, aerospace) & laser processes
-
Quality Consistency
Reducing the costly weld rework caused by weld defects and ensuring consistent quality
Would you like to know more?
Get in touch with our experts to learn more about the Laser Process Monitoring System and explore how AI and Computer Vision can help automate Quality Control and Quality Assurance!
Success Story
Through our collaboration with COMAU, a world-class leader in Industrial Automation and Robotics, within the ZELD-e consortium, we have successfully integrated the system with the company’s robotic arms deployed in electromobility and developed a turnkey solution for the Quality Assessment of e-car batteries welding processes.
Custom Implementation?
Do you need a custom Vision AI Industry 4.0 solution, tailored to your specific use case? Talk to our team, we will support you throughout the design and the deployment phases!