Aerospace components must meet exacting industry standards. Investment-cast parts, including turbine blades and structural components, must be inspected for even microscopic imperfections before they can be approved for use.
To help automate this demanding and time-consuming manual task, Access e.V., a German research institute specializing in materials and process development, created a robotic system for documenting fluorescent penetrant inspections (FPI).
Developed within the publicly funded GATE and FAST projects, the system combines a Rainbow Robotics RB5-850 collaborative robot, automated image capture, AI-assisted image analysis and RoboDK.
Access e.V. presented the system as a technical demonstrator at the ILA Berlin Air Show 2026. The cell shows how robotics, offline programming and simulation software can improve the speed, flexibility and ergonomics of aerospace quality inspection.
The challenge: Complex components in a confined cell
During fluorescent penetrant inspection, a penetrant is applied to the component. A substance called a developer then draws the penetrant out of any defects. The component is examined under UV light from multiple angles so that potential imperfections can be identified and documented.
Automating this documentation stage is challenging. Investment-cast parts often have complex, organic and asymmetrical geometries. Every new component requires a different sequence of movements to ensure that all relevant surfaces are presented clearly to the inspection camera.
Access e.V.’s solution places the RB5-850 cobot inside a compact, darkened inspection cell. The robot grips the component and moves it through a controlled sequence of orientations in front of a stationary camera. High-resolution UV images are captured from every required angle.
The confined workspace creates a persistent collision risk. The robot arm, gripped component, camera and cell infrastructure must all be considered when trajectories are calculated. Programming and testing every movement on the physical robot would take several hours. It could also place the equipment and valuable aerospace components at risk.
“Manually programming the robot would take several hours, as the operator would have to adjust coordinates, check the camera focus, readjust the positioning, and so on. With RoboDK, we can check and adjust the positioning virtually and in real time before applying it to the actual robot” – David Geibel, Project Engineer at Access e.V.
RoboDK software provided a full simulation and offline programming environment for the team at Access e.V. to work with. Credit: Access e.V.
Remote programming and virtual validation with RoboDK
Access e.V. uses RoboDK as the central offline programming and simulation platform for the robotic cell.
The team imports the CAD data for each component and creates a virtual model of the robot, component and inspection environment. RoboDK then automatically calculates the handling trajectories required to present the component to the camera from the necessary viewpoints.
“The biggest advantage for us is the ability to program the robot remotely,” says Geibel. “We don’t need to physically access the inside of the machine every time we want to change or adjust a robot position.”
This approach eliminates the need to program every movement manually inside the inspection cell. Before a program is transferred to the RB5-850, the complete image-capture cycle can be simulated and checked in RoboDK.
RoboDK also calculates the movements so that collisions between the robot arm, casting and surrounding infrastructure are ruled out from the beginning.
“Thanks to RoboDK, we were able to fully validate our handling trajectories and collision avoidance virtually,” says Geibel. “This allows us to ensure the system runs smoothly before the robot makes a single real-world movement.”
RoboDK enables high-fidelity simulations of complex cells. Credit: Access e.V.
Virtual testing also reduces the consequences of programming errors.
“Last but not least, if we have a collision in RoboDK, we can simply reset the program without risking damage to the actual parts or the robot,” Geibel explains.
Cutting documentation time by around one third
Automated image capture delivers a considerable time saving. Instead of relying on an inspector to reposition and examine the component manually, the robot moves through a repeatable sequence of camera positions.
“There is definitely a considerable time-saving aspect to having automated image capture,” explains Geibel. “By having the robot cycle through the different positions in a controlled manner, we can save roughly one third of the time, depending on the complexity of the part and the number of images required.”
Offline programming also allows Access e.V. to prepare upcoming components while the FPI cell continues capturing images.
“A second major benefit is the time savings, as we can run the machine while programming the next parts in parallel,” Geibel says.
Programs for new components can therefore be generated, simulated and validated on a PC. They do not need to be developed through trial and error on the physical system. Inspection work can continue without interruption, eliminating the downtime traditionally associated with programming a new application directly on the robot.
By automatically generating trajectories from component data, Access e.V. can integrate new parts in record time while maintaining 100% collision safety. Virtual validation protects the robot, camera equipment and valuable cast components from costly crashes inside the confined cell.
Improving conditions for human inspectors
Traditionally, inspectors spend extended periods in fully darkened rooms while examining fluorescent indications. The work is repetitive and unergonomic. The dark environment also limits how long an operator can work effectively.
In the Access e.V. cell, the robot performs the component manipulation and image-capture work inside the darkened environment. The resulting high-resolution UV images are transferred to a standard, well-lit workstation.
An AI system performs the first stage of image analysis and identifies potential flaws for further review. A human inspector then examines the AI’s preliminary findings and verifies the captured images. The inspector no longer needs to remain inside the dark inspection environment.
“The main benefit for us is that the operator can work in a lit environment instead of a dark inspection chamber,” says Geibel. “This allows us to increase the operator’s effective working time by around four hours per day.”
The system does not remove the human inspector from the process. Instead, it transfers the repetitive component handling and image capture to the robot. This frees the inspector to focus on reviewing outputs and making informed quality decisions.
RoboDK enables faster cell development and set up, while ensuring there are no collisions when the robot is handling parts with complex geometries. Credit Access e.V.
From research project to public demonstrator
Access e.V. presented the completed prototype at the ILA Berlin Air Show in June 2026. Throughout the event, the RoboDK-programmed RB5-850 manipulated the parts flawlessly.
The demonstration showed aerospace specialists how automatic trajectory generation, robotic image capture and AI-assisted analysis can be combined in a flexible quality-control system.
The project also demonstrates how RoboDK can bridge the gap between component CAD data and reliable physical inspection. It automatically calculates robot trajectories, validates movements virtually and prevents collisions before deployment. This enables Access e.V. to automate a difficult inspection task while improving flexibility, productivity and working conditions.
Access e.V. now plans to the technology beyond research demonstrations.
“Next step for the automated FPI will be the validation at an industrial partner,” says Geibel.
The automated FPI system was developed within the GATE and FAST research projects, funded by the German Federal Ministry for Economic Affairs and Energy on the basis of a decision by the German Bundestag.
