Uniquely Shaped Spaces: Object-Driven Algorithmic Shelf Design and Fabrication
Authors
Paper Title
Uniquely Shaped Spaces: Object-Driven Algorithmic Shelf Design and Fabrication
Publication Info
- Topic area: Computational design and fabrication of custom shelving based on object geometry.
- Keywords: Object-driven design, generative fabrication, algorithmic shelving, mixed-initiative systems, simulated annealing, cellular automata, custom furniture, subtractive fabrication, parametric joinery, human-computer interaction.
Background and Problem
- Problem / challenge: Conventional shelving systems rely on rectangular compartments that fail to accommodate the unique contours of objects. Existing computational design tools often prioritize designer-defined templates or abstract primitives, neglecting the potential of objects themselves to shape the design process.
- Significance: Shelving that adapts to the specific geometry of objects can enhance aesthetic, functional, and personal value, enabling storytelling and reflection through object display.
- Motivation and related work: Historical and contemporary practices in object display (e.g., Chinese display furniture, cabinets of curiosity) emphasize the interplay between objects and their containers. Prior computational tools focus on parametric constraints or sketch-based inputs but do not treat object geometry as a primary design driver. This paper addresses the gap by developing a system where object geometry dictates shelving design.
Solution
- Proposed approach: Uniquely Shaped Spaces, an algorithmic tool for generating custom shelving designs that respond to the contours of user-provided object silhouettes, ensuring manufacturability and aesthetic fit.
- Novelty:
- An object-driven generative pipeline that uses object geometry as the primary design constraint.
- A three-stage algorithmic workflow combining simulated annealing, cellular automata, and parametric joinery for manufacturable shelving.
- Empirical insights from a user study on how people interact with and adapt to mixed-initiative, object-driven systems.
- Procedure and key techniques:
- Object Input: Users photograph objects, mask their silhouettes, and upload them to the system.
- Layout Optimization: Simulated annealing arranges objects on a grid, optimizing spatial efficiency, structural validity, and aesthetic balance.
- Shelving Generation: Cellular automata grow walls around the layout, creating compartments tailored to object shapes.
- Fabrication Export: The system generates DXF files with parametric joinery for laser cutting and assembly.
Results
- Concrete findings:
- The system successfully generated geometrically fabricable shelving in 99.13% of 10,000 randomized trials.
- User study participants created shelving units ranging from 12–30 inches, fabricated in plywood or acrylic, and used them for curated object displays.
- Advantage over baselines:
- Unlike traditional shelving tools, the system centers object geometry, enabling unique, manufacturable designs that reflect the idiosyncrasies of users’ objects.
- Mixed-initiative workflows allow users to negotiate and adapt algorithmic outputs, fostering creativity and personalization.
- Experiments / evaluation:
- A 12-week user study with five participants, involving design workshops, fabrication, and two weeks of in-home use.
- Algorithmic stress tests with 10,000 trials to evaluate pipeline reliability across diverse object combinations.
- Limitations and future work:
- Current system guarantees geometric fabricability but not structural load-bearing performance.
- 2D silhouettes omit depth and surface details, limiting design precision.
- Rare failure cases in cellular automata wall growth (0.79% of trials) suggest the need for improved routing and recovery strategies.
- Future work could incorporate structural checks, richer shape acquisition, and mechanisms for incremental reconfiguration.
Summary
Uniquely Shaped Spaces introduces an object-driven approach to shelving design, using simulated annealing and cellular automata to generate manufacturable shelving layouts tailored to user-provided object geometries. A user study demonstrated how participants interpreted their objects as active agents in the design process, using the system’s outputs as starting points for negotiation and improvisation. The system reliably produced valid designs in stress tests, though limitations in structural modeling and rare algorithmic failures highlight areas for improvement. This work advances computational fabrication by foregrounding object constraints and enabling shared authorship between users, algorithms, and objects.
Research Questions / Practical Problems
Question signals indexed for this paper.
- 100%
StackMold: Rapid Prototyping of Functional Multi-Material Objects with Selective Levels of Surface Details
UIST '19· Desktop 3D Printing & Personal Fabrication +2
- 83%
ColorMod: Recoloring 3D Printed Objects using Photochromic Inks
CHI '18· Desktop 3D Printing & Personal Fabrication +2
- 83%
3D Printing Magnetophoretic Displays
UIST '23· Shape-Changing Interfaces & Soft Robotic Materials +3
- 80%
Forte: User-Driven Generative Design
CHI '18· Desktop 3D Printing & Personal Fabrication +1
- 80%
Grafter: Remixing 3D-Printed Machines
CHI '18· Desktop 3D Printing & Personal Fabrication +1
- 80%
Fabricaide: Fabrication-Aware Design for 2D Cutting Machines
CHI '21· Desktop 3D Printing & Personal Fabrication +1
- 80%
FabricatINK: Personal Fabrication of Bespoke Displays Using Electronic Ink from Upcycled E Readers
CHI '22· Desktop 3D Printing & Personal Fabrication +1
- 80%
Polagons: Designing and Fabricating Polarized Light Mosaics with User-Defined Color-Changing Behaviors
CHI '23· Laser Cutting & Digital Fabrication +1
- 80%
TensionFab: Fabrication of Room-scale Surface Structures From the Tension-Active Form of Planar Modules
CHI '24· Desktop 3D Printing & Personal Fabrication +1
- 80%
3D Printing Locally Activated Visual-Displays Embedded in 3D Objects via Electrically Conductive and Thermochromic Materials
CHI '24· Desktop 3D Printing & Personal Fabrication +1
Based on Jaccard similarity of research subtopics & professions (≥60%)