If a twin gives every scene the same inlier fraction, how many RANSAC iterations does it say we need, and is that enough when real scenes vary?

What it found

With N = 100 correspondences, minimal sample 2 and 50% mean inliers, the twin's budget for 99% success is 17 iterations. If the real scene inlier count is beta-binomial with concentration κ, the exact failure at 17 iterations is 1.3% (binomial variation only), 7.0% (κ = 10) and 14.1% (κ = 4); restoring 99% takes 19, 58 and 458 iterations, and for κ ≤ 2 no budget exists because scenes with fewer than 2 inliers cannot succeed. Exact values match a sampler, and an end-to-end noisy line fit goes from 0.9% to 12.6% slope failures at κ = 4 (1.8% with the larger budget, still above 1%). Stylised: assumed scene law, line fitting only, no real data.

A figure from its own output

Paper, code and results

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

More in Digital twins & sim-to-real

• Can a lidar twin give each beam an independent visibility probability, or do neighbouring beams share occluders?Lidar twin with independent per-beam visibility vs a Poisson occluder field: beams closer than an occluder diameter share occluders, exact blind probability by inclusion–exclusion over union lengths, 8 beams at 25 cm blind 0.0815 vs the twin's 0.0039 (20.9×), the twin is exact once spacing ≥ 2ρ, an occluder-field ensemble matches. Stylised.
• A battery twin with no internal resistance: how much endurance and how large a burst does the real, sagging cell actually give?Battery twin with no internal resistance vs a real sagging cell: endurance exactly (1+√(1−u))/2 of the twin's (0.5 at the fold), 9–94% of charge stranded at the voltage cutoff, a burst of 0.5 fails below 51% charge where the twin's floor is 0 (a flat 20% reserve covers only a third of the full-charge limit), and a constant-derate refit errs +12% to +473% off its fit power. Stylised.
• How many real trajectories does a classifier two-sample test need to reject a wrong digital twin, compared with a paired error test?How many real trajectories does a classifier two-sample test need to reject a wrong digital twin? Stylised pendulum: for a 20% Coulomb error a paired error test has power 0.98 at 100 trajectories vs 0.25 for the C2ST (which needs ~200), but for a 10% error the paired test stalls at 0.09–0.20 up to 800 while the C2ST reaches 1.00 at 400, and for same-variance noise-shape error the paired test has power ≤0.07 at every size. Null calibration approximate (60 replicates). Preliminary, simulated only.
• Was the paired digital-twin test weak, or was its statistic?Was the paired twin test weak, or was its statistic? Stylised pendulum, Student-t noise: the mean squared residual has power ~0.1 at a 10% Coulomb error up to n = 400, while median, trimmed, absolute and Huber statistics reach 0.94–1.00 from n = 50; Gaussian noise barely cares.
• Which robust paired test still catches a wrong digital twin when noise has infinite variance or whole runs go bad?Which robust paired test still catches a wrong digital twin when noise has infinite variance or whole runs go bad? Stylised pendulum, 10% Coulomb error: median and trimmed residuals reach power 1.00 under Student-t(2) cells but only 0.17–0.40 under bad runs, where a per-run median reaches 0.89–1.00; no statistic wins both.
• Which paired test of a digital twin survives both heavy-tailed cells and whole bad runs?Which paired test of a digital twin survives both heavy-tailed cells and whole bad runs? Stylised pendulum, 10% Coulomb error: a two-level statistic (trimmed mean of squared residuals within each run, then trimmed mean across runs) has power 0.99–1.00 at n = 50 under t(2) cells, bad runs at 0.86 and both at once, where the earlier single-level statistics each fail one case. Costs power under Gaussian noise (0.84 against 0.91).

All 164 Digital twins & sim-to-real 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.