Overview
Simulated data
Overview
Every number here comes from a signed evidence package that was verified, and whose findings were re-derived from its raw bytes, when it was ingested.
Cases
4
1 flagship, 3 stress tests
Integrity verified
4/ 4
Ed25519 signature and every byte
Sources per case
2
Robot ROS 2 recording + facility log
Evidence sealed
5.59 MiB
Private, write-once storage
Flagship caseSIM-2026-0001 · SIM-AMR-01
Open casefile →Facility-triggered safety stop
The facility stop broadcast (E-004) precedes the robot's recorded stop input (E-005) by 11.754 to 17.976 ms on the aligned timeline.
Facility → robot
11.8–18.0 ms
Robot reaction
30 ms
Time to stationary
0.91 s
Clock alignment
±0.55 ms
All cases
View registry →| Case | Scenario | Sources | Cross-source order | Integrity |
|---|---|---|---|---|
| SIM-2026-0001 | Facility-triggered safety stop | 2/2 | Order proven | Verified |
| SIM-2026-0002 | Safety stop with congested clock sync (ordering ambiguity test) | 2/2 | Order not provable | Verified |
| SIM-2026-0003 | Safety stop with a 1.8 s facility clock error (alignment robustness test) | 2/2 | Order proven | Verified |
| SIM-2026-0004 | Safety stop with facility source missing (incomplete evidence test) | 1/2 | No cross-check | Verified |
What this demonstrates
- Two separately clocked sources joined by an offset derived from their own sync logs
- Event order proven on uncertainty intervals, including an honest “can’t tell”
- Tamper evidence down to a 64 KB chunk, checked by two independent verifiers
- Findings re-derived byte for byte from raw evidence, each citing its records
What it does not establish
- Anything about real robots or sites. Every recording here is simulated.
- Sensor truth. Integrity proves bytes are unchanged, not that they're accurate.
- Cause, fault, liability or compliance.
- Physically independent custody. Today the sources are separate generators.