Velocity tracking at 0.1 m/sSIMULATION
microduck/results.tsv:9
2026-10-23 slowwalk E2B CPU stageB final.onnx B_floor0.12 target0.1 3 10/seed tracking% mean±sd 101.5±2.1 0 3/3
Where each number comes from
Every stat on this site is copied from a results file in our working repo. Below is the file, the line and the text of that line, so you can check each one.
Measured speed at a 0.3 m/s commandSIMULATION
microduck/results.tsv:5
2026-10-23 stageB_eval E2B CPU stageB final GFMIN1.8 nominal_DR 0.2/0.3/0.4 1 10 vx m/s 0.192/0.259/0.293 0
Scripted towel two-folds passedSIMULATION
laundrybot/rounds.md:41
Results (SIMULATION): 16/20 = 0.80 (Wilson95 0.584–0.919), mean stage 3.55, harness milestone1_pass=true.
10-step action error, single datasetOFFLINE
laundrybot/rounds.md:104
L1/std (CoRL-train std): CoRL-only k1 0.0736 / k10 0.1321 / k25 0.2039; pooled 0.0839 / 0.1442 / 0.2223; hold-state 0.0735 / 0.1744 / …
Clip: Stage CSIMULATION
Backlash 1.5°, 0.3 m/s command. Stress eval 87.5% tracking at 214.6M steps (gate: 90%). Did not pass.
microduck/results.tsv:84
Clip: Stage DSIMULATION
Backlash 1.5°, 0.3 m/s command. Stress eval 81.4–83.4% across two checkpoints, 0 falls. Did not pass.
microduck/results.tsv:89-90
Clip: Stage ESIMULATION
Backlash 1.5°, 0.3 m/s command. Stress eval 85.0% (64M) and 92.1% (107M), but the 107M checkpoint lost turning. Did not pass.
microduck/results.tsv:92
Clip: Scripted two-foldSIMULATION
Newton VBD cloth sim, 30 cm towel, trial 0 of 20. Scripted motions, not a learned policy.
laundrybot/rounds.md:41