#	Pagina
attuale pagina	/open-h2020/projects/213589/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - VISTA (Vision-based Inspection Systems for automated Testing of Aircraft interiors)

Teaser

Summary

Post-assembly testing of aircraft interior installations in both cabin and cargo units is fundamental to ensure quality and reliability of located components. Currently, most of aircraft interior testing activities are carried out by human surveyors, often in non-ergonomic conditions. Process chains are extremely complex and do not guarantee a sufficient level of transparency of the assembly status at all times.
Innovative automated testing systems aimed to support human work force are therefore needed, in order to reduce time and costs and to increase reliability, flexibility and transparency of tolerance control and quality check of the cabin and cargo interior final assembly.
The overall goal of the project is the development and implementation of a multi-sensor platform and sensor processing algorithms to be integrated on-board a mobile system (AGV) to provide efficient automated testing of cabin and cargo installations. In more detail, the use of multiple optical devices, like 3D LIDAR, omnidirectional laser profilometers, RGB-D, and (stereo) vision sensors, will be able to generate quality data, namely 3D reconstruction of surfaces in space and corresponding texture and color information. All sensors will be integrated on-board an automated guided vehicle (AGV), with a specifically designed imaging hardware, that will collect and store sensor measurements, while traversing the aircraftâ€™s interior, process them and report possible defects to a human officer. The latter will continue to be in charge for the final assessment and decisions, and will use innovative and immersive techniques, such as augmented or virtual reality, depending on the specific needs of the human operator.

Work performed

The work for the reported period has been mostly devoted for performing a close assessment of the measurement needs to reach the overall objectives and in devising the inspection strategy, the imaging sensor needed and in general the overall system architecture.
In particular it has addressed the requirements of the automated measurement system and the related test strategy needed to detect the panel defects that might arise because of an automated assembly of representative part of the internal aircraft linings. In order, this has required to analyze the standard procedures already in place while assembling a single isle civil aircraft, with the objective of identifying a set of guidelines useful for the automated assembly of ceiling panels, side walls panels, hatracks and cargo panels, that are considered in the VISTA project.
Following a classification already being used in Airbus, this list of requirements has been split between cabin and cargo area. Additionally, these requirements have been identified either as related to geometrical defects or to surface defects.
A test strategy has been devised, aiming at decoupling an initial â€œcoarseâ€ maneuvering toward the aircraft area to inspect and a fine localization strategy gaining control soon after that for pointing the imaging sensors toward the zone to inspect. In particular, this early analysis has shown that laser triangulation could provide the sufficient accuracy for detecting geometrical defects, while passive imaging sensors (including but not limited to RGB sensors) can acquire the necessary data required to find surface defects.
The requirements of the automated measurement system have then been used to compare different competing technologies for finding geometrical or surface defects. Suitable technologies that have been examined included laser profile scanners, structured light, time of flight, stereo and color cameras. They have been evaluated and compared based on their expected capability in detecting geometrical defects, such as issues with respect to positioning, gap and flush, and the ability to perform viable texture and color analysis for detecting and characterizing surface defects. Additional criteria being used have included cost, weight and transportability, power and illumination constraints, along with computational requirements. This comparison identified a much smaller subset of suitable technologies. These sensors have then been obtained for evaluation by hardware distributors. Following this, some days of testing have taken place at Fraunhofer IFAM facilities in Stade (Topic Manager facilities), where different kind of defects have been simulated on the panels considered in this project. These early experiments have therefore enabled the selection of two specific sensors for the detection of geometric and surface defects, usable both in the cabin and cargo area.
The overall architecture of the system, and specifically the hardware necessary for automatically moving the sensors inside the cabin and cargo area, have been considered as well. In particular, the study has addressed how the chosen enabling technologies for defect detection are integrated together in a sensing payload and then moved inside the cabin and cargo areas by means of an automated guided vehicle. In doing this, both the hardware and software architecture are considered, along with the necessary localization and acquisition strategy being employed. Moreover, data exchange requirements related to the visualization of the inspection results using an AR-capable interface have been analyzed, along with safety, integration constraints and requirements for the test environment to be configured in the IFAM premises.

Final results

The project focus is now shifting towards the definition of suitable processing techniques able to analyze images and 3d profiles, looking for geometrical and surface defects. An important expected result from defects assessment done by human operators to defects assessment made by an AGV and then reported to a human operator is related to more objective assessment. Although keeping the human in the loop might suggest that subjectivity could still be an issue, it is expected that this approach will increase the reliability of the results for several reasons: first, it will provide reliable results that are less affected by measurement tools positioning mistakes done by humans; second, since possible defects reporting is performed after the whole aircraft have been inspected, all defects will be considered in a shorter time-span, enabling human operators to provide more consistent decisions since all the possible defects can be compared together in a shorter time and when human focus has not yet been affected by fatigue. Last but not least, the planned ability to operate in a virtual reality or augmented reality settings will provide additional flexibility: the inspector is no longer constrained to operate and take decisions near the assembly facilities. Instead it can work from everywhere and even share the burden of a decision with another inspector, connected elsewhere. Indeed, there are different benefits from operating by using in an augmented reality scenario, where the inspector can over-impose important information on the part of the aircraft panels being inspected and come to a more informed decision. The potential impact of this vision for the factory of the future is therefore considerable, with benefits in the accuracy of the measurements, the consistency of the decisions, the flexibility of being able to operate everywhere, overcoming any geographical constraints.