Automated recognition and correction of warp deformation in extrusion additive manufacturing

Warp deformation is a common error encountered in additive manufacturing. It typically caused by residual internal stresses the manufactured part that arise as material cools. These errors are challenging to prevent or correct they build over time and thus only visible long after actions them. As result, existing work extrusion manufacturing has attempted warp detection but not correction prevention. We report hybrid approach combining deep learning, computer vision, expert heuristics warp. train convolutional neural network using diverse labelled images recognise real-time. compute five metrics from candidates predict severity of proportionately update print settings. This enables first demonstration automated both during printing for future prints.