Is a hanger designed in 3D really better than one designed on the 2D centre line, once seed noise is counted?

What it found

hanger-3d found that a 3D design loop beat hangers designed on the 2D centre line by 0.2–0.4 cm at 3 cm half-depth on two seeds, within seed noise. Here are eight seeds per route (M tee, 3 cm half-depth arm, 64 drapes per 3D loop, 150 per 2D loop). Whole-sheet error: 2D-designed 1.80 ± 0.13 cm (mean ± sd), 3D-designed with hanger-fit's bounds 1.54 ± 0.28 (difference −0.26, bootstrap 95% CI −0.44 to −0.03, exact permutation p = 0.033), 3D-designed with bow up to 5 cm and end radius up to 6 cm 1.12 ± 0.25 (−0.68 against 2D, CI −0.85 to −0.49, p < 0.001; −0.42 against the original bounds, p = 0.006). The 3D optimum sits on the radius limit in three of eight wide-bound runs, so the bound still binds. The wide-bound designs also drop the crease from about 25 to 11 deg/cm and the edge sag from 4.6 to 2.9 cm but have a worse centre-line error (0.77 against 0.40 cm), so the 2D score and the whole-sheet score disagree. Raising the radius limit to 9 and 12 cm (8 seeds each) leaves no seed on the limit at 12 cm and changes the error by only −0.10 to −0.11 cm (p > 0.4); the optimum radius ranges 4.8–10.4 cm, so the error is flat in it. Stylised: assumed fabric constants, one 20 cm strip, coarse grid, no friction; one garment and one arm depth; the budget (64 drapes) is small, so a longer search may change the gap (untested).

A figure from its own output

Paper, code and results

• README and method
• Experiments and results
• Code
• Tests

More in Laundry-hanging robots

• Does a second laundry-hanging arm lose less if it picks a free garment instead of waiting?Does a second laundry-hanging arm lose less if it picks a free garment instead of waiting? At 0.30 m clearance blind pinching recovers 9.5 of its 28 lost garments an hour (151 to 161), aiming about 2 of 10; at 0.45 m nothing. Stylised: clearance and spacing assumed.
• Is a laundry-hanging arm's gain from picking a free garment real? A paired test on 2000 loadsIs a laundry-hanging arm's gain from picking a free garment real? Paired on 2000 loads at 0.30 m: aiming +1.25 an hour (0.60 to 1.92), sweeping +4.4, pinching +10.0; at 0.45 m aiming gains nothing. Stylised: inputs are placeholders.
• How many dryers can one laundry-hanging arm serve once the pick is simulated, not assumed?How many dryers can one laundry-hanging arm serve once the pick is simulated, not assumed? A one-arm cell drops from 169 to 147 garments an hour (5.1 to 4.4 dryers) with look-then-aim picking and to 104 with blind pinching; with two arms the fold table is the limit. Stylised.
• How many dryers can a laundry-hanging arm serve when the fold takes as long as published robots take?How many dryers can a laundry-hanging arm serve when folding takes as long as published robots take? At SpeedFolding's 90–120 s per fold the learned-rule cell falls from 169 to 58–44 garments an hour; always hanging holds 146. Only the fold time is sourced.
• How many dryers can a laundry-hanging cell serve when its fold station runs at published robot speeds?How many dryers can a laundry-hanging cell serve when its fold station runs at published robot speeds? Timed from the SpeedFolding paper (68 to 167 s per garment) rather than an assumed 25 s, one arm and one fold table serve 0.8 to 2.1 dryers instead of 5.1, and with two arms five dryers need 3 to 10 fold tables. Fold times are sourced; every other station time is still assumed.
• Does the hanger-fit width rule hold on fabrics with published KES-FB bending rigidity and weight?Does "arm 1 cm wider than the half shoulder" hold on four fabrics with published KES-FB bending rigidity (0.0066–0.125) and weight (12.5–24.6 mg/cm²)? Best offset is +1 cm in all 12 cases (with stretch swept 40–800 N/m); heavy suiting at 40 N/m nearly ties with +2 cm. B and weight sourced; stretch swept; tee geometry assumed.

All 25 Laundry-hanging robots projects →

Other research topics

Proper scoring & elicitation28Effort & peer prediction9Verification games & slashing19Detection & tolerances14Markets as routers13Prediction-market design9Paying contributors8Decentralised training systems23Digital twins & sim-to-real164LLM chip design loops40Robot manufacturing loops53Ambient AI assistant46Self-replicating manufacturing34Simulation-first design64Engineered clothing45Task-specific AI accelerators40Algorithmic chip-design loops42Open-source purpose-built robots44The software-engineering loop43Adversarial human imitation41Open-source residential construction41Autonomous multi-angle capture34Sensing, imaging and 3D31Energy supply and dense energy storage39Robot societies among people31Agentic co-design of low-cost robots27Laundry-hanging robots25Machine learning on qualitative data23Robots on the building site27Civilizations & group psychology1History & human sciences1

Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.