Material extrusion is the most widespread additive manufacturing method but its application in end-use products is limited by vulnerability to errors. Humans can detect errors but cannot provide continuous monitoring or real-time correction. Existing automated approaches are not generalisable across different parts, materials, and printing systems. We train a multi-head neural network using images automatically labelled by deviation from optimal printing parameters. The automation of data acquisition and labelling allows the generation of a large and varied extrusion 3D printing dataset, containing 1.2 million images from 192 different parts labelled with printing parameters. The thus trained neural network, alongside a control loop, enables real-time detection and rapid correction of diverse errors that is effective across many different 2D and 3D geometries, materials, printers, toolpaths, and even extrusion methods. We additionally create visualisations of the network’s predictions to shed light on how it makes decisions.

No abstract

3D printing could revolutionize manufacturing through local and on‐demand production while enabling uniquely complex and custom products. However, 3D printing's propensity for production errors prevents autonomous operation and the quality assurance necessary to realize this vision. Human operators cannot continuously monitor or correct errors in real time, while automated approaches predominantly only detect errors. New methodologies correct parameters either offline or with slow response times and poor prediction granularity, limiting their utility. A commonly available 3D printing process metadata is harnessed, alongside the video of the printing process, to build a unique image dataset. Regression models are trained to precisely predict how printing material flow should be altered to correct errors and this should be used to build a fast control loop capable of 3D printing parameter discovery and few‐shot correction. Demonstrations show that the system can learn optimal parameters for unseen complex materials, and achieve rapid error correction on new parts. Similar metadata exists in many manufacturing processes and this approach could enable the adoption of fast data‐driven control systems more widely in manufacturing.