Online Preventative Non-Destructive Evaluation in Automated Fibre Placement

The strict quality requirements for aerospace composite struc- tures give rise to costly quality control procedures. In automated bre placement (AFP) these procedures rely heavily on manual work and inspection. This research aims at performing preventative non-destructive evaluation of composite laminate quality based on an online geometric analysis of the placed bre. A robot mounted laser pro le sensor, in combination with robot positional data, is used to create a 3D model of the bre. These are fused using quaternion coordinate transfer operations with the Robot Operat- ing System, an open source robotics platform. The 3D model is converted into an image for fast processing using open source algorithms from OpenCV. Deviations in part-product quality are identi ed in real-time including geometric, positioning and buckling defects due to high-radius curvatures in the bre path. Currently the prototype system will give a non-conformance warn- ing to the operator, and in future work it is planned to develop