Declan Coster

Student: Declan Coster

Project: Data Driven Correction of Distored Images

View: Research Poster (PDF)

Institution: Lehigh University

Major: Electrical and Computer Engineering

Advisor: Yaling Liu

Abstract

While resin-based 3D printing machines (Stereolithography (SLA) or Digital Light Processing (DLP)) have advantages in printing accuracy and print intricacy compared to filament-based printers,  the projected pattern of resin-based printers usually have distortion at large area leading to inaccuracies in the machine output. Current corrections for this process require hardware solutions which can be expensive and time-consuming or have to limit the printing size to a small area. This project proposes a solution for SLA or DLP printing distortion using a data-driven software approach. The solution measures the distortion of the printer by processing photos of calibration runs on the machine. Information from the measurements is then used to reconstruct the distortion for the whole domain. Using the information of the distortion over the domain, the G-code coordinates can be read and corrected by the inverse mapping of the distortion. This calculates the commands needed to output the intended shape and corrects the gcode to account for the machine’s distortion. With the large growth in laser additive manufacturing year after year, effective and practical solutions to existing problems can have an impact on how this field progresses. Our solution builds on the trend of cloud computing to bring the solution to users independent of computer capabilities. Additionally, our results show that a data-driven approach is effective at measuring and reverting distortion in SLA/DLP machines. With companies creating bigger and bigger machines, our results show a way in which maintenance can be minimized or how distortion could be corrected in additional machine lasers.