Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Parallelobox: Improved Decomposition for Optimized Parallel Printing using Axis-Aligned Bounding Boxes

Hayley Hatton, Muhammed Khalid, Umar Manzoor, John Murray

Abstract

Much contemporary research in additive manufacturing focuses on breaking down models into constituent parts in the pursuit of various factors, such as printability of large models in smaller printing volumes, or reduction of support structures. Newer research has begun to focus on using these decomposition processes for printing models across multiple printers in parallel. We present a novel approach to this that incorporates axisaligned bounding boxes as height fields to improve the characteristics of decomposition, including printing time, feasibility, and aesthetics. By expanding these bounding boxes according to a parallel printing objective, with additional improved efficiency from a metaheuristic process, these boxes can then be used for rapid decomposition using simple out-of-the-box mesh clipping operations. This algorithm is experimentally evaluated across a range of models against two other contemporary approaches to parallel printing that use more rudimentary techniques, such as recursive symmetry and cube skeletonization. Parallelobox outperformed each of these across a range of sample models on the basis of a parallel printing time metric using simulated 3D printing to compute the results

Parallelobox: Improved Decomposition for Optimized Parallel Printing using Axis-Aligned Bounding Boxes

Abstract

Much contemporary research in additive manufacturing focuses on breaking down models into constituent parts in the pursuit of various factors, such as printability of large models in smaller printing volumes, or reduction of support structures. Newer research has begun to focus on using these decomposition processes for printing models across multiple printers in parallel. We present a novel approach to this that incorporates axisaligned bounding boxes as height fields to improve the characteristics of decomposition, including printing time, feasibility, and aesthetics. By expanding these bounding boxes according to a parallel printing objective, with additional improved efficiency from a metaheuristic process, these boxes can then be used for rapid decomposition using simple out-of-the-box mesh clipping operations. This algorithm is experimentally evaluated across a range of models against two other contemporary approaches to parallel printing that use more rudimentary techniques, such as recursive symmetry and cube skeletonization. Parallelobox outperformed each of these across a range of sample models on the basis of a parallel printing time metric using simulated 3D printing to compute the results

Paper Structure

This paper contains 34 sections, 5 equations, 16 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Decomposition for Additive Manufacturing
  4. Decomposition in Parallel
  5. Axis-Aligned Bounding Boxes
  6. Contribution
  7. Problem Overview
  8. Parallel Printing
  9. Application of Axis-Aligned Bounding Boxes
  10. Method
  11. Symmetry Optimization
  12. Orientation Optimization
  13. Grid-Based Fine Decomposition
  14. Surface Region Extraction
  15. Initial Block Determination
  16. ...and 19 more sections

Figures (16)

  • Figure 1: Overview of the Parallelobox process, from model to printing and assembly.
  • Figure 2: Flow chart depicting the overview of the critical stages of the algorithm.
  • Figure 3: 2D Cross-sectional example of region classification for extraction. Grey cells are the "External" voids, Black cells are the "Internal" spaces, White cells are the "Boundary" cells.
  • Figure 4: 2D cross-sectional demonstration of how the initial "seed" blocks are determined. Left: Cluster centroids determined from k-means++ on vertices; Middle: Association of cluster centroids to nearest surface boundary boxes; Right: Output initial blocks in cyan.
  • Figure 5: Demonstration of the block growth phase on 3DBenchy.
  • ...and 11 more figures