Increasing 3D Printing Accuracy Through Convolutional Neural Network-Based Compensation for Geometric Deviations
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
As Additive Manufacturing (AM) evolves from prototyping to full-scale production, improving geometric accuracy becomes increasingly critical, especially for applications requiring high dimensional fidelity. This study proposes a machine learning-based approach to enhance the geometric accuracy of 3D printed parts produced by Fused Filament Fabrication (FFF), a widely used material extrusion process in which thermoplastic filament is heated and deposited layer by layer to form a part. Our method relies on a Convolutional Neural Network (CNN) trained to predict a systematic deviation field based on 3D scan data of a sacrificial print. These scans are acquired using a structured light 3D scanner, which provides detailed surface information on geometric deviations that arise during the printing process. The predicted deviation field is then inverted and applied to the digital model to generate a compensated geometry, which, when printed, offsets the errors observed in the original part. Experimental validation using a complex reference geometry shows that the proposed compensation method achieves an 88.5% reduction in mean absolute geometric deviation compared to the uncompensated print. This significant improvement underscores the CNN’s ability to generalize across geometric features and capture systematic deformation patterns inherent to FFF. The results demonstrate the potential of combining 3D scanning and deep learning to enable adaptive, data-driven compensation strategies in AM. The method proposed in this paper contributes to reducing trial-and-error iterations, improving part quality, and facilitating the broader adoption of FFF for precision-demanding industrial applications.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it