Students:Kay Blokland, Jason Browne, Quintus Kesler, Fadil Ihsan Fadilah (2025)
GKN – Fokker unveils an automated scanning system for detecting FOD
Overview
Fokker Aerostructures, a leader in advanced aerospace manufacturing, specializes in high-performance composite and metallic structures. Serving top aircraft manufacturers and defense organizations, the company continues to find new standards in efficiency and manufacturing.
Problem
However, a challenge has emerged in the manufacturing process. Fokker Aerostructures produces fuselage panels by bonding aluminum sheets with epoxy. Over time, this epoxy can leave residue on the mold, affecting the next batch. In some cases, the adhesive remains loosely attached, while in others, it must be manually scraped off. This epoxy labelled as FOD (foreign object debris) could cause imperfections in products that can be detrimental to the unforgiving tolerances of the aerospace industry
Solution
Based on this problem, our group began developing a solution to scan the tool for adhesive. Initially, we built a small rover designed to move over the tool and scan its surface. However, during testing, we encountered significant challenges with this design, including traction issues on the tool’s curved sections and difficulties in localizing the rover.
Given these challenges, we went back to the drawing board to develop an alternative solution. We designed a moving gantry scanner that captures images of the mold surface by traveling along its edge using a motor-driven wheel. This design offers ease of use, as the scanner can be repositioned effortlessly.
The gantry’s wheelbase functions similarly to train wheels, running along the edge of the tool. On the left side, the wheelbase is adjustable to accommodate a double-curved mold. A spring-loaded wheel is added the prevent the system from falling outward while adapting to the mold’s curvature. To move the camera sideways, a rack-and-pinion mechanism mounted on a slider, driven by a stepper motor, is used.
The web-based GUI allows the user to input the mold size, which at Fokker can vary, with molds reaching up to 6 meters in length. The gantry then generates a movement plan for the mold and begins scanning across it while capturing multiple images. These images are subsequently cropped and stitched together to form a complete image of the mold.
Once the gantry reaches the end of its track, it moves to the next row and continues scanning until the entire mold has been covered. Epoxy detection is performed using a YOLOv8 AI model, with the results schematically displayed on the screen. The GUI presents the detected epoxy locations and the number of epoxy debris found, allowing the user to quickly identify and remove them
Customer Quote
“The problem statement called for a solution that would be the least disruptive to current manufacturing process step and factory floor area. This location is already made to be as lean as possible with tools and logistical equipment occupying every available space. The solution the students came up was modular, could easily be placed on a tooling mould to be allow for scanning, and then removed by 2 operators. This solution is the least disruptive to the current manufacturing process, relatively inexpensive, easy to implement and most importantly doesn’t require any additional fixtures or installations to be implemented. The solution thought of by the students was truly out-of-the-box and we are extremely satisfied by their work. We look forward to continue R&D on this solution and make it fit for industrialisation.”
-Shashank Amin
Future improvement
This gantry has a lot of room for improvement, some improvements that could be made in the future are:
- Movement feedback to track real world position
- Better image processing
- Higher resolution camera
- Better lighting setup
- A mechanism to indicate the detected epoxy
- Rotation in camera and lighting to keep perpudicular to surface
- Possibility of rotating wheel base for double curved tools
Besides these areas for improvement, the prototype was a great success in demonstrating the viability of using a gantry to scan the tool’s surface and showcasing the mechanics behind the solution.