Flexible cobot test station reduces human error in inspections

Manually checking the quality of components or products in industry is labour-intensive and error-prone. To overcome this, the Fraunhofer Institute for Mechatronic Systems Design IEM has unveiled a cobot solution with AI-based imaging that increases efficiency and reliability in a collaboration with ATM manufacturer Diebold Nixdorf and digitalisation company verlinked GmbH. The system is designed to free employees from having to perform visual inspections and can be incorporated into many kinds of testing scenarios.

Collaborative robots (cobots) are considered a key technology in industry. Generally equipped with AI and sensor technology, these robots work alongside people in production facilities, unlocking flexible and intelligent automation concepts. In today’s world of batch size 1, fragile supply chains, and new and ever-changing regulations, this can provide a crucial advantage over competitors. In the project called CogeP (Cobot-supported test stations for Intelligent Technical Systems), employees use the robot to check the quality of ATM control panels quickly and without any errors.

AI-supported quality checks

To check the control panels, the sensor-supported robot arm moves a camera over the components, workpieces or products to be analysed from various angles. The AI-supported image analysis feature analyses the quality, and then the robot moves the camera to the next product. If the analysis software finds any defects — a screw sticking out or a wobbly plug connection, for example — the system notifies the employees responsible, who can then zero in on the issue and correct it.

“Manual visual inspections in production are a monotonous, labour-intensive task for employees, and they’re error-prone on top of that,” said Dr.-Ing. Eugen Djakow, group manager for Automation Technology and Robotics at Fraunhofer IEM. “The testing robot handles these kinds of tasks quickly and reliably, which has the added advantage of making people’s work in production more interesting and less monotonous.”

IoT platform unlocks range of testing scenarios

The solution also combines the testing robot with an IoT platform. It functions as a real-time data centre, assigning testing tasks, storing the results and collecting the corresponding data across different robots and assignments. This means testing processes can be further optimised: the data collected can also be used to adjust the testing for a new product version, without having to put as much time and effort into programming. New product versions only require minor tweaks to the testing procedure.

“Cobot-supported test stations can collect all of the testing and operational data in the central IoT platform and access the information in real time,” Djakow said. “That means companies can reconfigure their testing routines to accommodate new requirements, without costly modifications or investments.”

Benefits for SMEs: flexibility in incorporating cobot workstations into production

The cobot solution can be used not only for testing processes but also for assembly, picking and packing, and general plant support. Fraunhofer IEM also harnesses its experience in industrial process technology and expertise in areas such as machine learning for automated pattern analysis to support SMEs in incorporating cobot workstations into existing production processes, at low cost and without major investments in time and effort. This allows small and medium-sized enterprises, just like their larger counterparts, to modify their production operations on short notice or adapt to small unit volumes, right down to single items.

Image caption: The AI-powered image analysis software is integrated directly into the robot arm’s camera module. If the software detects a defect, the system will notify employees, who can then rectify the fault. © Fraunhofer IEM/Janosch Gruschczyk