Introduction of a Robotic System for Aerospace Manufacturing: the Case for Non-Destructive Inspection and Beyond

In this work, results and lessons learned with the introduction of a robotic system to inspect aerospace parts with both fluorescent penetrant (FP) and Eddy currents (EC) are presented. Non-destructive testing (NDT) is used in... [ view full abstract ]

In this work, results and lessons learned with the introduction of a robotic system to inspect aerospace parts with both fluorescent penetrant (FP) and Eddy currents (EC) are presented. Non-destructive testing (NDT) is used in the aerospace industry either during manufacturing or maintenance tasks to assert the presence of cracks. Usually done manually, NDT is highly dependent on the inspector skill, it is also very time consuming and tedious. In a project supported by NSERC and CRIAQ, the first author led a team of academics and industrial partners, namely L3-MAS and Pratt and Whitney Canada, aiming at adapting a common serial robot to NDT. This robot was used to handle the probes and camera associated with each inspection technique. Based on data provided by a CAD file, an initial visual inspection of the part which has been processed with FPI chemicals is realized. This inspection allows for the identification of areas of the part where potential defects have been identified. The latter are then automatically confirmed (or not in the case of a false positive) by an ECT probe. Eddy current testing is much more accurate than fluorescent penetrant inspection but very time consuming which prevent to use this technique exclusively. The establishment of the final diagnostics is based on a neuro fuzzy software which compare the signals received by the camera and the EC probe and compare them to the known database of samples stored in its memory. This database was bases on actual practical data provided by the industrial partners of the project and used for the training of human inspectors. This allows the system to use the previous expertise of the industrial partners developed along the years. The neuro fuzzy system also tries to extract additional characteristics of the defects such as depth, shape, etc. which can be used to remove the defect by machining if desired. The final system built by the team achieved all these objectives and has been successfully tested on sample parts but some limitations exist that will be discussed.