Škoda uses AI-based ‘Magic Eye’ camera to quickly identify maintenance needs on its assembly line

Škoda Auto is using AI-based image recognition to ensure timely identification of maintenance needs on the assembly line. At the main plant in Mladá Boleslav, Škoda has installed a system that continuously monitors assembly line equipment and uses Artificial Intelligence (AI) to detect irregularities in the processes and identify any required maintenance work.

Cameras on the assembly line’s overhead conveyor
The images of equipment and parts subject to wear, such as girders, bolts or cabling, are captured by cameras on the overhead conveyor of the assembly line. As soon as the AI-based system – connected to the camera – detects irregularities in the process, or a need for maintenance actions, it flags them in real time.

Continuous checks enable early detection of required maintenance measures
Magic Eye instantly compares its high-precision photographs against thousands of stored images. This enables it to detect departures from the optimal baseline conditions and to identify sources of error. The use of blue light ensures that the AI tool reliably differentiates between cracks and scratches and makes the correct diagnoses. In addition, the system continuously expands its knowledge base. If it finds a worn bolt, for example, it marks the spot as error-free as soon as the part has been replaced and checked again. To evaluate detected deviations, the system uses information on irregularities it has identified in the past.

Current use on the assembly line for the Enyaq iV and Octavia models
Škoda is using Magic Eye at its main plant in Mladá Boleslav, on the assembly line for the Enyaq iV and Octavia models. To enable further optimisation of the system and accelerate wider integration at the Mladá Boleslav and Kvasiny sites, Škoda has simulated a section of the assembly line. This “implementation arena” can be used to experiment with different camera settings, configure system parameters and simulate damage to the assembly line.