G-code Assisted Phase Unwrapping For Autonomous Error Detection and Correction in 3D Printing

Autonomous error detection in 3D printing allows users to save time, material, and cost by monitoring where defects may occur throughout the fabrication process and alerting the user or stopping the print if damage is imminent. Coupling this detection with a closed-loop controller can enable the autonomous correction of certain defects within the 3D print, resulting in high reliability without the need for manual parameter tuning. However, the inefficient implementation of either of these processes in-situ can potentially put the user at a detriment, with increased processing times or material wastage in between layers while computations are being performed. Existing structured light (SL) 3D imaging approaches try to address this challenge, yet for such a closed-loop pipeline, they fall short in their computationally expensive phase unwrapping, a step required in SL for 3D reconstruction of the measured geometry. As such, there is a need for a high speed phase unwrapping methodology for use with hardware for real-time closed loop error correction in 3D printing. This thesis presents a novel method for phase unwrapping that utilizes the G-code of the printed geometry in order to significantly improve upon the computational speed. By modelling each G-code layer of the printed geometry as an ideal plane shared between the camera and projector within the SL imaging system, optical relationships can be used to mathematically constrain the system such that less total fringe images need to be captured and processed for phase unwrapping. This works by transforming these G-code planes into the projector coordinate frame, where absolute phase is sampled along the plane, thereby providing a reference at which to find fringe order without the need for significant spatial or temporal processing that most unwrapping algorithms rely on. This method has been shown to calculate unwrapped phase both accurately and quickly compared to a well-established multifrequency unwrapping baseline. Furthermore, by integrating this method with an existing phase-based error detection and correction algorithm, high speed in-situ closed-loop control has been successfully achieved.