Matching CAD model and image features for robot navigation and inspection of an aircraft

Igor Jovančević (1) , Ilisio Viana (1) , Jean-José Orteu (1) , Thierry Sentenac (1, 2) , Stanislas Larnier (2)

1 ICA - Institut Clément Ader
2 LAAS-RAP - Équipe Robotique, Action et Perception

This paper focuses on the navigation of a moving robot equipped with cameras, moving around an aircraft to perform inspection of different types of items (probes, doors, etc.). Matching CAD model and image features is useful to provide meaningful features for localization and inspection tasks. In our approach two primitive sets are matched using a similarity function. The similarity scores are injected in the edges of a bipartite graph. A best-match search procedure in bipartite graph guarantees the uniqueness of the match solution. The method provides good matching results even when the location of the robot with respect to the aircraft is badly estimated. Inspection approaches on static ports and air inlet vent are presented.

Mots clés: feature matching; EDCircle; Hough transform; PTZ camera; computer-aided-inspection; image analysis

Domaines: Traitement du signal et de l'image [eess.SP]

Igor Jovančević, Ilisio Viana, Jean-José Orteu, Thierry Sentenac, Stanislas Larnier. Matching CAD model and image features for robot navigation and inspection of an aircraft. 5th International Conference on Pattern Recognition Applications and Methods (ICPRAM'2016), Feb 2016, Roma, Italy. ⟨10.5220/0005756303590366⟩. ⟨hal-01353317⟩