Automatic, multicopter-based indoor inspection of large surfaces - AMIIGO

Key Info

Basic Information

01.05.2017 to 31.08.2019
Organizational Unit:
Chair of Production Metrology and Quality Management, Model-based Systems
German Federation of Industrial Research Associations AiF

Research partner

  • Institute of Control Engineering (IRT)



Aline Kluge-Wilkes

Research Assistant


+49 241 80 20237


  Aircraft Copyright: © IRT/WZL

Every commercial aircraft is struck by lightning on average once a year. In order to rule out possible defects, a time-consuming visual inspection of the aircraft's outer skin by maintenance personnel is necessary. In the research project "Automatic Multicopter-Based Indoor Inspection of Large Surfaces (AMIIGO)" an automated multicopter is used for the inspection. The automated, non-destructive defect detection of pin-sized damage on the up to 4,400 m² fuselage surfaces (e.g. Airbus A380) will drastically reduce maintenance costs and AOG (aircraft on ground) time.

At IRT, a flight path and a departable trajectory for the multicopter are generated from a CAD model of the aircraft to be inspected. During the flight, collision avoidance takes place in real time, based on the evaluated measurement data of a LIDAR on board. In this project, the Chair of Production Measurement Technology and Quality Management of the WZL realizes the automated acquisition of location-indexed images as well as the automated identification of maintenance-relevant surface defects.

Central challenges are the automated defect recognition and the precise indoor localization of the highly dynamic multicopters during image acquisition. Industrial camera technology mounted on the multi-copter is aligned perpendicular to the aircraft surface by a gimbal suspension depending on the current pose. For the traceability of the downstream defects to be identified by the autonomous offline defect detection to the real position on the fuselage surface, the image data at the time of data acquisition are referenced with position data from a sensor fusion. The image data are first reduced by a hybrid of classical image processing with machine learning algorithms and then examined for potential defects. A so-called defect map is extracted from the found and traceably located defects, in which the maintenance personnel are interactively presented with the identified defects on the aircraft model.