MUNICH—BMW AG sees artificial intelligence (AI) as a way to improve productivity and efficiency at its assembly plants. The automaker is using the technology for a variety of quality-related applications, such as error proofing and inspection tasks that leverage automated image recognition and image tagging.

“Artificial intelligence offers great potential,” says Christian Patron, head of innovation, digitalization and data analytics at BMW Group Production. “It helps us maintain our high quality standards and at the same time relieves our people of repetitive tasks. The technology is fast, reliable and, most importantly, easy to use.”

AI-based applications are gradually replacing permanently installed camera portals on BMW assembly lines.

“A mobile standard camera is all that is needed to take the relevant pictures in production,” explains Patron. “The AI system can be set up quickly. Employees take pictures of the component from different angles and mark potential deviations on the images. This way, they create an image database in order to build a neural network that can later evaluate the images without human intervention.

“Employees do not have to write code; the algorithm does that virtually on its own,” claims Patron. “At the training stage, which may mean overnight, a high-performance server calculates the neural network from around 100 images, and the network immediately starts optimizing. After a test run and possibly some adjustments, the reliability reaches 100 percent. The learning process is completed and the neural network can now determine on its own whether or not a component meets the specifications.”

According to Patron, the increasing integration of smart data analytics, state-of-the-art measurement technology and AI opens up new opportunities in production management. In the final inspection area of BMW’s plant in Dingolfing, Germany, vehicle order data is compared to a live image of the model designation of the newly produced car. If live image and order data don’t match, workers carrying out the final inspection receive a notification.

“Images from the final inspection may show that weld metal has sprayed out at the same welding point in several car bodies,” says Patron. “Using AI, the control loop can be closed and system control or maintenance cycles be adjusted even faster and more efficiently. AI and analytics applications offer the potential to detect sources of error at such an early stage that errors can hardly occur any more.

“[Our] Intranet of Things platform ensures a smooth integration of new AI applications into production IT,” says Dirk Hilgenberg, BMW’s senior vice president for production systems, technical planning, tool shop and plant construction.  “Workers can choose the most suitable tools from a digital toolbox, combine them into their own solutions and install them via plug and play.”

BMW recently publicly shared parts of its digital image tagging software, which has seen successfully applied in various AI applications. In turn, software developers around the world support the development of AI software, allowing the automaker to focus primarily on the advancement of specific AI applications in production and logistics.

“The open source approach benefits both interested software developers and the BMW Group,” says Patron. “We provide elements of our innovative digital image tagging software, which has proven effective in multiple AI applications; in turn, we receive support in taking our AI software to the next level of development. This allows us to focus more strongly on advancing specific AI applications in production and logistics.”