Volkswagen wants to use “Computer Vision” to improve production efficiency

The Volkswagen Group’s plants are to become more efficient. For this reason, productivity in vehicle factories must increase by 30 percent in the decade between 2016 and 2025, according to the Wolfsburg-based automaker’s internal guidelines. Digital processes, more computers and new software play a major role in this.

More and more assembly processes at the Zwickau plant are automated.

The next new field of application at the level of the entire group is optical systems: image recognition and image processing, called “Computer Vision”. “The use of the technology offers great potential to make our production even more efficient,” says Gerd Walker, Head of Production in the VW Group.

In a first step, Volkswagen calculates across all brands with annual savings in the double-digit million range. By contrast, the Group’s own investments by 2024 will reach a single-digit million amount.

So far, individual group brands have often only pursued their optical projects for themselves. That should change now, the “Computer Vision” becomes a central project of the entire group. The brands should continue to pursue and implement their own ideas. The new projects should then also be passed on to other subsidiary brands in order to achieve maximum savings at group level.

Automated processes also play a decisive role in the installation of the front axle. Here the robot automatically places the heavy components on the chassis frame.

The sports car subsidiary Porsche provides an example of this. The Stuttgart-based manufacturer has developed an app that simplifies the acceptance of a car. Several labels are attached to each vehicle manufactured, such as the detailed vehicle information and the instructions for the airbags.

Many of these stickers contain country-specific information and are written in the customer’s language. A German Porsche employee has to deal with language problems alone.

The driverless transport system in assembly line 4.

A new solution from the field of “Computer Vision” ensures that the labels are attached correctly. At the Porsche plant in Leipzig, an employee on the line has recently been scanning the vehicle number to identify the vehicle. He then takes photos of every label placed in the car.

The app checks the images in real time for the correct content and language. It gives the employee feedback whether everything is really OK. This saves several minutes per vehicle produced. Next year, the new app will also be passed on to sister brands in the group.

Cameras detect the smallest cracks

Even Audi has developed a process in the program, which may soon be used throughout the Group. The Ingolstadt Audi factory has had an optical process for quality inspection in the press shop for two years. The large sheet metal parts of a car, such as fenders or the bonnet, are pressed there.

In the press shop, cameras detect the smallest cracks and deviations from the norm on the components. Applying such a procedure across the Group is actually obvious. “The smallest damage to the sheet metal looks just like it does in Wolfsburg,” says Frank Göller, who heads the digital production department at Volkswagen.

Here the robot installs the cockpit fully automatically.

However, the cameras used in production also need support: Artificial Intelligence (AI). The AI ​​is used to evaluate the large amount of information that the cameras provide. AI processes are similar to human ability to capture, process, and analyze images.

However, there are also limits: the self-learning machines do not ultimately achieve higher levels of complexity. If it gets too complicated, people will be unbeatable for now. VW tries to find such uses that are best suited for the artificial optical systems, so they do not overwhelm them.

The reason why the VW Group now wants to increasingly use optical processes and the “computer vision” is due to technical progress. “The technology has developed rapidly,” emphasizes Jakob Engelmann, digitalization project manager at Volkswagen.

Computer capacity and speed have improved significantly. That makes it worthwhile to use the technology much more frequently in Volkswagen’s vehicle production. The cameras have been improved in parallel – and also become cheaper.

Lars Thielemann, Head of Planning at the Zwickau plant, stands in front of the cockpit installation in the assembly hall…

Together with Amazon and Siemens, Volkswagen is also building an industrial cloud for its own factories, which will thereby be linked together. The new “Computer Vision” optical systems are also part of this cloud: The cloud makes it easier for VW engineers to use the new cameras in vehicle production.

The cloud also ensures a certain degree of centralization. The group can more easily pass on projects such as those from “Computer Vision” to the plants involved.

The number of cameras used in production will increase at Volkswagen in the near future. Digital boss Göller speaks of a “multiple” of today’s number. At the same time, this creates problems with data law. Personal data of the employees cannot be used by the company without further ado. Furthermore, the cameras must not become a control instrument; the anonymity of the employees must be preserved.

This has very simple practical consequences. “Captured images with recognizable faces and clear identifications such as tattoos are pixelated right from the start. In this way, we generally protect personal rights, ”says Frank Göller. “It’s about increasing quality and improving processes, but not building a monitoring apparatus,” he emphasizes. Tattoos are sometimes very individual and allow individual employees to be identified.

Contact the author:

Inline Feedbacks
View all comments