Does a sheet-following robot skill trained on a rigid-sheet simulator survive a flexing sheet, sensing delay and actuator lag?

research/buildbot-skills

Does a sheet-following skill tuned on a rigid-sheet simulator survive a flexing sheet with delay? It passes 81% of episodes where it was tuned and 0-11% on the held-out plant; tuning on the held-out plant gives 61-66% there.

7 tests

What it found

The buildbot-cohold follow skill (impedance gains K, B and an integral term) was tuned by random search on a rigid-sheet plant with randomised sheet mass (6–23 kg), lift time, reference lag and feed-forward error, then tested on 1000 fresh episodes of a held-out plant: two rigid halves on a torsional spring, a first-order force actuator and delayed feedback. A policy passes if peak tilt ≤ 3°, final tilt ≤ 0.5° and the person's load stays under the NIOSH constant. On the training plant the tuned skills pass 81–82% of episodes (lag-limited ceiling); on the held-out plant they pass 0–11%, because the high gains they pick go unstable under delay. The buildbot-cohold nominal setting passes 61% then 26%. A skill tuned on the held-out plant passes 61–66% there and 67–72% back on the rigid one. Randomising the parameters a simulator has does not cover the dynamics it lacks. Delay (0–30 ms), actuator lag (5–50 ms), sheet stiffness (150–6000 N m/rad) and the pass thresholds are my ranges, not measurements; the second plant is my own planar model, not a second physics engine.

A figure from its own output

Figure produced by research/buildbot-skills: figure — buildbot-skills/experiments/figure.png, written by the project's own experiment code.

Paper, code and results

More in Robots on the building site

How many times must a G1 humanoid re-stand to nail one 4x8 sheet of sheathing?How many times must a G1 humanoid re-stand to nail one 4x8 sheet? About 13 stances at 5 platform heights: with its feet planted it holds a 3.4 kg nailer at 50 N only 0.7-1.5 m high and 0.3 m sideways, so one stance covers at best 6 of 62 nails, and the sheet matches hand nailing (372 s) only if each re-stance takes under 26-29 s (push to seat and re-stance time are swept, not sourced; no walking simulated).
How well must a nailing robot find the studs, and does measuring more beat measuring better?How well must a nailing robot find the studs? A 4x8 sheet needs a single stud reading good to 0.89 mm (1 sigma) for 95% clean sheets; a stud finder's 'within 1/8 in' claim gives 28% (8 of 62 nails missed) or 69%; averaging reads and pooling a straight wall help, re-measuring each nail is worst.
Which tasks of building a wood-frame house hold the crew hours, and which can a stock humanoid even hold the load for?Which house-building tasks hold the crew hours, and which can a stock humanoid hold the load for? An open task book of a 28x40 ft wood-frame house: 20 tasks with cycles, masses, tolerances and crew-hours, scored against the simulated Unitree G1 and H1 payload envelopes; fastening is 31% of the hours, and the H1 passes the payload test for 53-81% of them (minutes per unit are assumptions, swept).
Which joint's torque limit sets how fast a stock humanoid can carry a load?Which joint's torque limit caps a stock humanoid's carrying? For the G1 it is the waist pitch: doubling it alone takes the heaviest feasible payload from 4 to 6 kg; the H1 has no single bottleneck.
Do open permit records predict how long a permit and the build behind it take, better than each city's own averages?Research programme: LightGBM quantile models beat the per-city, per-type empirical-quantile baseline on pinball loss in 13 of 15 city/target/quantile cells (2024 test year, Chicago, San Francisco, Austin); Chicago's tails are the exception
How long a step can a stock Unitree G1 or H1 take carrying lumber, and does that beat a labourer's carry rate?Can a stock humanoid carry a stud faster than a labourer? A MuJoCo single-support step envelope: the G1 steps 0.50 m with one 2x4 and cannot carry two; the H1 steps 0.20 m with one and 0.10 m with 15 kg; neither beats a labourer who takes a bundle of four, and the model is restrictive (no falls, drops or energy scored).

All 27 Robots on the building site projects →

Other research topics

Proper scoring & elicitation28 Effort & peer prediction9 Verification games & slashing19 Detection & tolerances14 Markets as routers13 Prediction-market design9 Paying contributors8 Decentralised training systems23 Digital twins & sim-to-real164 LLM chip design loops40 Robot manufacturing loops53 Ambient AI assistant46 Self-replicating manufacturing34 Simulation-first design64 Engineered clothing45 Task-specific AI accelerators40 Algorithmic chip-design loops42 Open-source purpose-built robots44 The software-engineering loop43 Adversarial human imitation41 Open-source residential construction41 Autonomous multi-angle capture34 Sensing, imaging and 3D31 Energy supply and dense energy storage39 Robot societies among people31 Agentic co-design of low-cost robots27 Laundry-hanging robots25 Machine learning on qualitative data23 Robots on the building site27 Civilizations & group psychology1 History & 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.