BMW Boosting AI While Factory Lines are Paused from Pandemic

BMW Boosting AI While Factory Lines are Paused from Pandemic

August 20, 2020

Source: AI Trends Staff

With the Covid-19 pandemic forcing its factories to close, BMW AG is taking the opportunity to accelerate deployment of AI in its factories.

Matthias Schindler, head of AI innovation for BMW’s production systems group, has installed AI-powered quality control systems in many of the company’s 31 factories over the past three years, according to an account in the WSJPro. He had typically installed and tested new AI systems during planned work stoppages during holidays, but the pandemic-related shutdowns enabled work to happen in the factories without running into production.

The additional quality control checks are especially important to BMW given that cars are becoming more customizable, with different interior finishes, technical features, engine types and energy options.

“We have diesel cars, we have petrol cars, we have hybrid cars, and, in the very near future, we will have fully electric cars on the same line,” Schindler stated. “They look different from the product perspective. That brings large challenges.”

Schindler sent teams to three plants in Bavaria to install AI-based software that scans photos of car parts for defects. Teams working in groups of four or five installed high-powered computers and sensors into factories to test the software, then tune the program’s algorithms, which they are developing themselves with help from the consultants. Quality control checks are seen as investments that provide immediate return, so are not on pause due to tighter budgets from weaker sales.

The image recognition system analyzes still photos taken by cameras placed at intervals throughout the assembly line. While parts are moving on assembly lines, the checks detect abnormalities such as loose or missing screws. “Typically around 8,000 to 12,000 parts will be assembled into a car, depending on the derivative,” Schindler stated. “The customer wants all these thousands of parts to be exactly as he has ordered.”

BMW did have a system to check cars on the assembly line with cameras, but the line had to stop while the images were checked by third-party software against target images. The old system tended to produce more false positives, slowed the line and required more manual checks. BMW’s flagship factory in Munich can produce nearly 1,000 cars per day, so the line pauses are expensive.

Press Shop Makes Body Parts Out of Blank Pieces of Metal

The press shop at the BMW Group’s plant in Munich turns more than 30,000 blank pieces of metal a day into vehicle body parts. Since a system incorporating AI was installed in 2019, each blank has been given a laser code at the start of production, so the body part can be identified throughout, according to an account in The code is picked up by the system, called iQ Press, which records parameters such as the thickness of the metal, oil layer, temperature and speed of the presses. The parameters are then related to the quality of the parts produced.

The data is uploaded to the cloud in real-time and is immediately available to the production team, so they can gain a more clear picture. Among the advantages is eliminating the need for each body part to be checked; the system picks out only irregularities that need to be further investigated. The system attempts to continuously optimize as it runs as well.

Support for the advance of AI throughout BMW was also supported by Robert Engelhorn, Director of BMW Group Plant Munich. He stated in an interview, “At Plant Munich, it takes about 30 hours to manufacture a vehicle. During that time, each car we make generates massive amounts of data. With the help of AI and smart data analytics, we can use this data to manage and analyze our production intelligently.”

He added, “AI is helping us to streamline our manufacturing even further and ensure premium quality for every customer. It also saves our employees from having to do monotonous, repetitive tasks. Our team in production are highly experienced specialists, so they are the best judges of whether an AI application can boost quality and efficiency at any given stage of production.”

AI at BMW Can Paint, Too

BMW also recently started a new pilot aimed at using AI in the painting process, according to an entry on the BMWBlog. The system will first be installed in Munich and if results warrant, worldwide.

Every new car will be inspected in the paint shop to gather data to be used towards dust particle analysis, which will be stored in a database ideally of best practices. The team will be developing AI algorithms that compare live data from dust particle sensors to the reference data, in the hope of identifying any potential defect.

“Data-based solutions help us secure and further extend our stringent quality requirements to the benefit of our customers,” stated Albin Dirndorfer, Senior Vice President Painted Body, Finish and Surface at the BMW Group. “Smart data analytics and AI serve as key decision-making aids for our team when it comes to developing process improvements. We have filed for several patents relating to this innovative dust particle analysis technology.”

As an example of how the system can be used, the data gathered by the system can determine if weather conditions are not optimal for painting. During prolonged dry periods, for instance, the algorithms could potentially detect whether the number of dust particles has increased so that filters need to be replaced. If that cannot happen in time and the car gets painted, it may not meet the quality standard.

The paint system’s algorithm monitors over 160 features related to the car body, thus is able to predict the quality of paint application very accurately. Additional measuring points with more precise sensor data are required for car body cleaning stations, in order to capture needed data. BMW’s AI experts are confident the new paint system will be suitable to be deployed in more of its factories throughout the world.

Read the source articles in and BMWBlog.