Conformance record
Snapshot of this document in the GEMs repository. The repository itself is the current version. Bản chụp tài liệu này trong repo GEMs. Bản đang chạy nằm ở chính repo.
This is a self-declared claim for a simulated body. Protocol §4 permits one provided the claim says so; this is that statement. No physical body exists, so no row below carries physical evidence.
Generated by assess.py — do not hand-edit.
| Protocol version | 0.1 |
| Assessment | self-declared, not third-party assessed |
| Date | 2026-09-26 |
| Commissioning baseline | none — no hardware to baseline |
| SHOULDs overridden | none |
Result
| State | Count |
|---|---|
| MET | 4 |
| MET (simulated) | 3 |
| not implemented | 2 |
| not demonstrable without hardware | 7 |
| Total | 16 |
This body does not conform. Protocol §4 is explicit that partial conformance is not a conformance claim but a gap list, and that is what this document is. It is published because a precise gap list is more use than a vague disclaimer.
Requirements
| # | Requirement | Evidence | State | Note |
|---|---|---|---|---|
| C-1 | Inside/outside separable | C | MET (simulated) | the distinction is constructible from proprioceptive returns; demonstrated against synthetic data, not a physical push |
| C-2 | Effective action | M | not demonstrable without hardware | needs a physical region that can return something other than predicted; a simulator returns what it was written to |
| C-3 | State persists | T | MET | 20 records across one sealed batch, chain continuous, no findings |
| C-4 | Readable emission | T | MET | emission written and read back with its anchored context intact |
| C-5 | Distinguishable action | M | not demonstrable without hardware | needs ambient physical fluctuation to be distinguishable from |
| C-6 | History accrues | T | MET | sequence gapless and monotonic; an out-of-order write is refused |
| C-7 | Withstands resistance | D, M | not demonstrable without hardware | needs a structure to resist with |
| C-8 | Agency classification precedes interpretation | C, T | MET (simulated) | procedure §7.1 run on synthetic returns: commanded motion and applied force distinguished, tags written pre-fusion, residuals retained so the tag can be re-derived |
| C-9 | Traced appraisal | C, T | MET | an emission without anchored context is refused at write, as is one still unclassified; no bypass path exists in the writer |
| C-10 | Low-power trace | M | not demonstrable without hardware | needs a radio to transmit from, a loop to write loop_state, and a meter to measure quiescent power with |
| C-11 | Integrity attestable | C | not implemented | tier 1 needs a secure element, tier 2 a commissioning baseline taken from real sensors, tier 3 actuators to challenge; none exists |
| C-12 | Contact amplitude recorded | M, T | MET (simulated) | the recording requirement is enforced at write; the measurement it records needs instrumented contact surfaces that do not exist |
| C-13 | Balance rate ≥ 500 Hz | M | not demonstrable without hardware | needs firmware on a target board |
| C-14 | Reflex latency ≤ 10 ms | M | not demonstrable without hardware | needs firmware on a target board |
| C-15 | Shared time base | M | not implemented | the shared time base is specified and the bus chosen; nothing implements it |
| C-16 | Supported failure state | C | not demonstrable without hardware | needs a body that can fall |
Declaration (protocol §8)
Figures are derived from the specification and the budget model, never measured.
| Declared | Value |
|---|---|
| Simulated or physical | simulated — no physical body exists |
| Mass | 96 kg (derived, not weighed) |
| Free-running endurance | 2 h at the declared operating point, with a dock |
| Endurance ceiling | 10.7 h |
| Sleep states supported | none implemented |
| Protection level and coverage | NIJ IIIA + stab, 65% coverage — selected, not built |
| Actuator specific torque | 80–90 Nm/kg peak over module mass — selected, not procured |
| Summed joint torque | 2479 Nm across 31 joints |
| Peak power at the source | 19.2 kW at 5C |
| Link bandwidth and latency | 8 Gbps, ~1 ms PHY — specified, no radio |
| Sensor configuration | 41 logged joints; no sensor exists |
| Aggregate raw rate | 15.1 Gbps at the conservative configuration |
| Compression ratio achieved | none — not implemented |
| Compute split | specified; nothing running |
| Log rates, full tier | 2.6 Mbps, 1.18 GB/h |
| Contact surfaces instrumented | none |
What the gaps cluster into
The requirements that cannot be demonstrated fall into four groups, and each group is one purchase away from being testable:
| Group | Requirements | What it would take |
|---|---|---|
| A target board and firmware | C-13, C-14, C-15 | one compute module and a bus interface |
| Sensors and contact surfaces | C-2, C-5, C-12 | one instrumented joint with force sensing |
| Structure | C-7, C-16 | one limb that can be loaded and dropped |
| Root of trust and radio | C-10, C-11 | a secure element and a low-power radio |
The cheapest group is the last, and the one that unblocks the most rows is the first.