AI system to solve the limited-data challenge for Robotic X-ray quality control of large objects – RoboQuality

Today, robotic tomography systems are able to inspect the internal and external parts of large objects, which opens up new possibilities for applications in sectors such as automotive or aeronautics.
In order to be able to carry out the 3D reconstruction of the inspected object, a sufficient number of acquisitions, corresponding to different angles of view, must be carried out. In many situations, some angles of view cannot be acquired due to the dimensions of the object and/or the kinematic limitations of the robots. This leads to a loss of quality in the 3D reconstruction and is one of the reasons why destructive methods are still commonly used in industry for the inspection of large objects.
Many recent research works have shown that artificial intelligence techniques are able to substantially improve 3D reconstruction results from incomplete datasets. The RobotQuality project aims to develop and evaluate AI techniques adapted to 3D reconstruction for the quality control of parts. It thus aims to make possible the non-destructive evaluation of large parts such as can be encountered in the automotive or aeronautical industries.