Audi Makes Robots Quality Controllers

Audi is claiming to have developed AI capable of performing quality control on its cars.

It says the robots it has developed are able to “recognize and mark the finest cracks in sheet metal parts” quickly and “reliably”. This confidence stands in marked contrast to the automaker’s recent moves away from autonomous vehicle development.

Audi says its individual components are currently inspected by human employees with the assistance of a number of miniature cameras installed in its factory presses. It says it plans to replace these employees with a machine learning process, although when asked to comment, a company spokesperson insisted the process would “be introduced in addition to the visual inspection by employees” and would only “support the employees in their task” by superseding Audi’s existing image recognition software.

The procedure purportedly draws from deep learning and is “based on a complex artificial neural network” which can “detect the finest cracks in sheet metal with the utmost precision”. Audi said its machine learning team developed this robotics tech by sending it “sample images” with tiny error marks in them of no more than a pixel in size, and the robots are now capable of detecting these errors.

The automaker claims the neural network upon which the robots are based “now learns independently from the examples and detects cracks even in new, previously unknown images”. Audi head of equipment Jörg Spindler said the rationale for the move was that it would assist the automaker in its efforts to overhaul “its production plants into modern smart factories”.

A company statement from Audi warned that other elements of the automaker’s “visual quality inspections” process were also at risk of being outsourced to robots, citing roles currently taking place within its assembly and support paint shops as potential examples.


Leave a comment

Your email address will not be published. Required fields are marked *