CHALLENGE 

In many sectors of industrial production, the smaller the components, the more complex their shapes and the more crucial the quality control.

Modern and effective quality control is therefore becoming increasingly important; this applies to the quality of molds, surfaces and processing, especially of miniature components in the automotive sector, for example, and also to quality assurance and product tracking of important data carriers such as data matrix codes. Although these have been in use since the late 1980s, they are playing an increasingly important role in the electronics industry. The 2D codes, which are only millimeters in size, can encode numerous other data in addition to simple product identification, such as product and serial numbers, technical information, etc.; in addition, error correction processes also facilitate error-minimized reading of the data. 

Especially in key future technologies such as e-mobility, components are becoming ever smaller and more complex, confronting quality inspection carried out manually in whole or in part up to now with increasing challenges. Years of experience, expertise and individual assessment by human employees are no longer keeping pace with technological requirements.

So how can automation help to make quality control for these components and data matrix codes more precise and flexible? 

SOLUTION 

In September 2023, MABRI.VISION developed a solution to this challenge with its robot-based platform ROBOT.SPECTOR: Thanks to a combination of robot-guided sensor technology and AI-based image analysis, the platform enables precise inspection of even very complex surfaces and from different angles for seamless quality control. 

Founded in 2015 by Dr. Ulrich Marx and Dr. Nicolai Brill at the technology hub in Aachen, the company develops, manufactures and supplies optical sensors and turnkey inspection systems for production. With a workforce of currently 50 employees, the company focuses on innovation and high-tech for customers in the packaging technology, automotive, metal processing and medical technology sectors - with the aim of supporting these customers in optimizing quality and efficiency in production. 

To date, the ROBOT.SPECTOR platform has been implemented in two versions, both featuring a DENSO Robotics six-axis robot at the center of a test cell, a VS-6556 and a VS-068. “We chose DENSO Robotics and the VS series because these robot models can reliably deliver the precision we require in the application,” says Dr. Ulrich Marx, co-founder and Managing Director of MABRI.VISION GmbH.

The first application with a VS-6556 at its center is used to scan and check the quality of Data Matrix codes, i.e. the ROBOT.SPECTOR acts as a Data Matrix code verifier checking the correct positioning and quality of these rather small codes (maximum edge length 2.1 mm) on circuit boards. These circuit boards are manufactured by an automotive supplier for the installation in EVs.

The inspection process in this ROBOT.SPECTOR DMC starts with the components ready for inspection to be placed in trays. These are manually inserted into the inspection cell; the robot, which can be equipped with various Data Matrix code scanners, is then checking each component (i.e. circuit board) and its codes (some of which are undercut) individually in a thorough scanning movement in order to read and inspect them. The control software displays which code has been read and the code quality, i.e. the correct positioning of the code, is being recorded for determining the code’s quality. This will ensure that the codes can be read error-free when used later in the production. On average, the inspection process takes around two seconds per code. 

While the desktop-based control software is a proprietary MABRI.VISION development, the interface to the robot is using RoboDK simulation software for the robot to be also operated with a digital twin.

The platform's second robot-based application, the ROBOT.SPECTOR Robotic Vision System, combines a DENSO VS-068 robot with state-of-the-art image processing and AI deep learning for quality control of components for chip production. 

The platform's second robot-based application, the ROBOT.SPECTOR Robotic Vision System, combines a DENSO VS-068 robot with state-of-the-art image processing and AI deep learning for the quality control of components for chip manufacturing. Achieving unprecedented inspection accuracy by detecting even hard-to-reach and hidden areas that other systems cannot access, this combination is reliably detecting even complex and barely recognizable defects. A key role is played by the AI, which identifies such defects and faults using anomaly detection - the software is trained with good data in order to make sure that it not only recognizes established or new defects, but also enabling it to classify known defects. 

Customers wanted to automate their previously manually performed quality control process - but how can human experience be translated into figures and data, e.g. threshold values? For an automated system, defects also have to be precisely defined. These tasks are relatively easy to solve for many inspection processes involving the shape, mounting and assembly of components, but when it comes to surface defects such as scratches, the programming of automated quality control using image processing reaches its limits.

This is where the latest, constantly evolving AI-based add-on component for the ROBOT.SPECTOR comes into play. The quality inspection of components for the latest generation of semiconductors is carried out in the ROBOT.SPECTOR test cell with a clean room environment. The components to be tested for surface defects and dimensional accuracy are between 50x50x10 mm and 400x300x200 mm in size. In this case, the test cell is designed to be slightly larger, as it also has a turntable to ensure that the system is sufficiently fast and at the same time cycle time-optimized: the robot can therefore feed and at the same time the component can be rotated on the turntable.

During the process, the component to be inspected is positioned on the turntable using a holding system and then scanned during an optical inspection by two camera systems attached directly to the robot; one camera in combination with a segment ring light and a second one with a UV bar light. 

Thanks to the segment ring light and using a shape-from-shading approach, any surface irregularities such as scratches or dents can be highlighted, while the UV light (using fluorescence) makes even the smallest particles or impurities visible. These tests are carried out in sequence. The duration of the inspection depends on the dimensions and complexity of the component - for larger, complex components, it’s less than two minutes with up to 40 data points being recorded.

This application, too, is controlled by the company's proprietary software, using the same interfaces as in the first application. The software is programmed so flexibly that different test protocols can be taught.

The VS-068 six-axis robot from DENSO Robotics offers high speed and high performance in a compact, slim housing. In addition, numerous options are available to enable operation in a variety of environments. Thanks to its high-speed performance and repeatability of ±0.02 mm, the VS-068 can be used in almost all areas. 

ADVANTAGES 

The ROBOT.SPECTOR is setting new standards in terms of flexibility, speed and accuracy in automated quality inspection. Speeding up quality inspections, the technology ensures a high ROI. In addition, automation using the ROBOT.SPECTOR enables consistent, automated and error-free quality control. 

“The real added value presented by the ROBOT.SPECTOR platform and the DENSO robots is less of a quantifiable automation advantage but rather its robot-based flexibility and precision it provides for the inspection process. These are the robot’s features we are relying on in the system,” says Dr. Ulrich Marx. For example, even the matrix code inspection was previously carried out manually in some cases, i.e. the sensor and the component were positioned in relation to each other manually. Later, SCARA robots were also being employed, but proved to be limited in their flexibility by only four axes. 

For the ROBOT.SPECTOR, DENSO’s six-axis robots are also expanding the range of inspection processes (even those with special requirements) the system can handle in the best possible way. Even hard-to-reach areas of components can now be inspected in the process cycle: the robots can be equipped with different inspection technologies, such as high-resolution cameras, fast code readers or 3D triangulation sensors.  

As the ROBOT.SPECTOR platform can also be combined with any other DENSO robot model and inspection axes can be designed to work in varying dimensions, the system can be suited for a wide range of production spaces without being limited to specific industries or sectors as a platform. This is why MABRI.VISION is also planning to use the system in electronics testing or metal processing in the future; further applications can be realized wherever precision is required.