Towards Zero-Waste Furniture Design

Bongjin Koo, Jean Hergel, Sylvain Lefebvre, Niloy J. Mitra

Abstract

In traditional design, shapes are first conceived, and then fabricated. While this decoupling simplifies the design process, it can result in inefficient material usage, especially where off-cut pieces are hard to reuse. The designer, in absence of explicit feedback on material usage remains helpless to effectively adapt the design -- even though design variabilities exist. In this paper, we investigate waste minimizing furniture design wherein based on the current design, the user is presented with design variations that result in more effective usage of materials. Technically, we dynamically analyze material space layout to determine which parts to change and how, while maintaining original design intent specified in the form of design constraints. We evaluate the approach on simple and complex furniture design scenarios, and demonstrate effective material usage that is difficult, if not impossible, to achieve without computational support.

Introduction

In traditional furniture manufacturing, a significant "decoupling" exists between form and fabrication. Designers typically create a 3D shape first, which is only later decomposed into parts to be cut from flat sheets, such as plywood. Because designers lack real-time visualization of how 3D modifications affect 2D material packing, they often produce designs that result in substantial "off-cut" waste—typically ranging from 5% to 15% in mass production.

Optimizing this process is inherently difficult because the "bin packing problem" is computationally NP-hard. When the shapes themselves are allowed to vary to improve fit, the search space expands exponentially. This research introduces a computational framework that bridges this gap, suggesting subtle design adjustments—such as slightly widening a shelf or shortening a leg—to fill gaps in the material sheet while preserving the designer's intent.

Methodology

The system treats furniture design as a function $D(X)$, where $X$ is a vector of design parameters (e.g., lengths, widths). The optimization process occurs in two primary stages:

1. Discrete Optimization (Ordering)

The system employs a "docking algorithm" to pack parts onto a sheet. Since the packing sequence significantly impacts the final layout, the algorithm stochastically explores various sequences to establish an efficient starting point.

2. Continuous Optimization (Shape Adjustment)

This "guided design" phase involves two types of moves:

  • Grow Step: The system attempts to increase part sizes to fill empty spaces in the 2D layout, adhering to user-defined constraints like symmetry or minimum functional heights.
  • Shrink Step: If parts form a "locking chain" that prevents further growth, the system identifies and shrinks specific components to break the chain, unlocking new optimization possibilities.

To maintain an interactive user experience, the authors developed a "sliding" heuristic. Rather than re-calculating the entire 2D packing for every 3D change—which would cause jarring visual jumps—the algorithm "slides" parts in four directions to resolve overlaps and close gaps smoothly.

Results and Evaluation

The framework was evaluated on custom furniture designs of varying complexity, including curved pieces like a "Parrot" shelf and parametric items like coffee tables and storage chests.

Figure 2: Comparison of (a) input design, (b) optimized final design, and (c) the fabricated coffee table, showing improved material utilization.

Key results include:

  • Significant Waste Reduction: In the "Parrot shelf" example, wastage was reduced from 34% to 8%. A coffee table (illustrated in Figure 2) saw a reduction from 22% to 11%.
  • Interactive Performance: The system operates at interactive rates, processing complex designs (16+ parts) in a few seconds to four minutes, facilitating a "human-in-the-loop" workflow.
  • User Validation: A study with six design students indicated that the tool effectively removed the frustration of switching between 2D and 3D representations, allowing for more sustainable choices without hindering creativity.

Discussion and Future Work

While the system significantly improves material efficiency, it currently has limitations. It can adjust dimensions but cannot alter the "topology" of a design (e.g., adding a new support beam). Additionally, the current docking algorithm does not support "nesting" small parts inside the hollows of larger ones.

Future work will focus on integrating structural analysis, such as the Finite Element Method (FEM), to ensure that optimized designs remain structurally sound, and scaling the system to handle designs exceeding 30 planks efficiently.

Summary updated

Abstract

In traditional design, shapes are first conceived, and then fabricated. While this decoupling simplifies the design process, it can result in inefficient material usage, especially where off-cut pieces are hard to reuse. The designer, in absence of explicit feedback on material usage remains helpless to effectively adapt the design -- even though design variabilities exist. In this paper, we investigate {\em waste minimizing furniture design} wherein based on the current design, the user is presented with design variations that result in more effective usage of materials. Technically, we dynamically analyze material space layout to determine {\em which} parts to change and {\em how}, while maintaining original design intent specified in the form of design constraints. We evaluate the approach on simple and complex furniture design scenarios, and demonstrate effective material usage that is difficult, if not impossible, to achieve without computational support.
BibTeX
@article{koo2017towards,
  author = {Bongjin Koo and Jean Hergel and Sylvain Lefebvre and Niloy J. Mitra},
  title = {{Towards Zero-Waste Furniture Design}},
  journal = {IEEE Transactions on Visualization and Computer Graphics},
  year = {2017},
  url = {https://openreview.net/forum?id=d6tNRiNFEd},
}
Figure