Robot Assisted Weld Inspection Quality via XR Training
RobWIQ-XRT develops an immersive Virtual Reality (VR) training module for weld-defect inspection. Its mission is to modernize vocational training by replacing traditional classroom methods with a dynamic, AI-powered simulation. Trainees will use a VR headset to operate a virtual robotic crawler, learning to inspect life-size pipeline welds in a realistic, safe environment.
The project brings together a consortium with distinct, complementary expertise:
- WELDING GREEK INSTITUTE (WGI): The project coordinator and educational lead, providing curriculum design, weld-inspection expertise, and validation trials.
- Bolt Virtual: The XR development lead, responsible for creating the Unity/VIROO application, user interface, and overall virtual experience.
- Plaixus Limited: The AI and robotics technical lead, tasked with developing the CNN defect-detection models and the adaptive AI broker.
Technology
RobWIQ-XRT solution is a single-player scenario built on the VIROO® platform, where students and junior inspectors learn to calibrate, operate, and interpret data from a robotic vision system. The core innovation lies in its adaptive PC-side AI pipeline. This system uses a “compute-budget broker” that monitors performance and swaps between different AI network models in real-time.
This ensures a fluid user experience by balancing the frame rate with the accuracy of weld-defect detection. The platform will log extensive user data, including head movements, robot position, and decision-making latency. This data provides instructors with unprecedented insight into a trainee’s learning process, allowing for personalized feedback and adaptive difficulty. The project will leverage existing assets, including WGI’s ISO 9712 curriculum, Plaixus’s lightweight CNN models, and robotic assets from previous MASTER XR projects.
