The Composite Automation Development Centre (CADC) of the Delft University of Technology, is trying to bring more smart automation in the composite manufacturing- and processing industry. This project is executed as a pilot for automating drilling of fuselage panel laminate layups.
The assignment originates from Fokker who manufacture these panels major aerospace manufacturers. Currently the pilot holes are drilled out manually, which is labour intensive and prone to errors. The goal of this pilot is to demonstrate that, the for aerospace required accuracy is, achievable using smart robotics.
For this project a Kuka KR 210 robot and a MicroEpsilon scanCONTROL 2950-25 linescanner were made available. Additionally, Fokker supplied a folding mould for fuselage panels, fuselage panels with pilot holes already drilled out and CAD drawings for these panels.
A requirement for this project is that only commercially backed software can be used. For computer vision Halcon had to be used. RoboDK had to be used to control the Robot.
Using the abovementioned tools, an end of arm tool (EOAT), as well as software, had to be developed to achieve the required accuracy.
To validate the accuracy of the system, three concepts in different price and complexity were made.
• A simple solution, which would have a pin penetrate a hole in the panel and push into foam. The foam would be mounted on the back of the panel behind the hole and after a test the foam could be removed to measure the accuracy by hand.
• A somewhat complex solution, where a joystick would be used to touch the pilot hole. If the joystick would not be in the centre of the hole, the joystick would be angled and the accuracy can be read out.
• A complex solution, where four digital micro meters would be mounted in a diamond pattern around a central one. This would be able to accurately compute the angle of the End of Arm Tool relative to the panel as well as the positioning on the pilot hole.
After consultation with Fokker, it was determined the simple solution would be further developed. To compensate for possible mistakes, parts of the solution had to be cheaply replaceable. The first prototype was attempted with as material for the pin. This was deemed not feasible as no wood was found which was straight enough. The final solution for the End of arm Tool was then developed using spring loaded metal pin through a guide hole.
In RoboDK a panel with holes can be loaded in. The holes are then extracted by a script and targets at their positions are made. Whilst running to program, the robot will move toward each hole and stop when the distance between the line scanner and the panel is optimal. There the robot will sweep over the hole whilst the scanner is recording. Halcon will identify the hole and extract its exact coordinates. The coordinates are then sent back to RoboDK and the robot will penetrate the panel and the foam with the pin through pilot hole.
Using the current setup not all holes can be reached. This has the following causes and potential solutions.
• The end of arm tool collides with stringers on the panel when trying to scan certain holes.
A different line scanner with more range should be used.
• The external track on which the robot is mounted is not supported by RoboDK.
Convince RoboDK to build in support for this setup.