PiPER grip clock v0.3 — frozen acquisition study

Distribution status: the tagged source, local installation kit, model, complete native evidence, and recorded 3D comparison have verified publication records. This closes distribution checks; it does not change the study's results or qualify a physical skill.

A compact learned clock pauses a frozen PiPER arm and jaw reference during grasp acquisition. This is an installable local CPU simulation prototype.

Frozen strict completion: 44/128 learned clock, 26/128 matched blind clock, 72/128 fixed timing, 67/128 sampled acquisition teacher. All 512 confirmation attempts are valid. The paired full-minus-blind interval crosses zero; no reliable feedback advantage is established. The learned clock performs worse than fixed timing. All qualification flags remain false.

The independent audit reconstructs all 768 main-study attempts with zero mismatches. Separate pilot failures remain public. The kit includes the wheel, exact data-only model documents, licenses, checksums and Terminal instructions. Python >=3.12; install with the arm extra. Native simulation and fitting use CPU. Camera, human-video motor learning, continuous learned slip recovery and hardware transfer are not established.

The recorded viewer retains all 512 confirmation results and replays 258 attempts: all 128 learned-clock and 128 matched-blind attempts, plus the first scheduled fixed and teacher controls. That replay selection is independent of outcome. It projects recorded native motion; the browser does not run a fresh physics experiment.

The prebuilt browser viewer was repaired separately after a large base64 validation pattern caused a JavaScript stack overflow. The display fix preserves canonical decoding, byte limits, hashes and decompression checks. The original v0.3 source tag and release kit remain unchanged; use this later display-fix source when regenerating the viewer locally. The separate hosted display overlay is recorded in the viewer proof below and does not rerun or improve the learned policy.

Research report · Recorded 3D comparison · Model · Complete native evidence · Local guide.

The native-evidence proof records anonymous byte verification of all 50 model files and 6,239 dataset files at the original pinned revisions. A later catalog overlay corrects the piper tag collision to agilex-piper, preserves original metadata, and verifies five changed files per repository while preserving every other Git/LFS identity. The full archive was not redownloaded at those later metadata heads. The recorded-viewer proof distinguishes the original anonymous verification of all 273 added Space files from the later display-only repair, with preserved scientific resources and separate exact revisions.

These compact public proofs bind the preserved original receipts by SHA-256. DVIDIA code, models and authored numeric evidence use MIT; bundled components retain their included licenses.