The following documents are available in this folder.
- System Requirements Specification (SRS)
- Data Management Specification (DMS)
- System Architecture Description (SAD)
- Machine Learning Component Specification (MLCS)
- System Test Specification (STS)
- Deployment Specification (DS)
Supplementary documents
- Numbered ALTAI questions from the Assessment List for Trustworthy Artificial Intelligence. The European Commission published ALTAI to support self-assessment. The questions were discussed in a related paper on AI ethics.
- Instructions for data generation. Note that ESI Pro-SiVIC is needed to reproduce the SMIRK data set.
SMIRK evolves in compliance with Guidance on the Assurance of Machine Learning in Autonomous Systems (AMLAS) published by the University of York. The figure below shows an overview of the AMLAS process and the six corresponding stages.
The table below is an index to navigate the artifacts mandated (cf. [ID]) by the AMLAS process. The input/output columns indicate in which AMLAS stage the artifact is used. The final column lists the location of the corresponding artifact for SMIRK.
The set of artifacts listed constitutes the safety case for the ML-based object detection component of SMIRK, i.e., it instantiates the ML Assurance Scoping Pattern.
| ID | Title | Input to | Output from | Where? | Version |
|---|---|---|---|---|---|
| [A] | System Safety Requirements | 1, 6 | SRS Sec 3.1 | v1.0 | |
| [B] | Description of Operating Environment of System | 1, 6 | SRS Sec 4 | v1.0 | |
| [C] | System Description | 1, 6 | SRS Sec 2 | v1.0 | |
| [D] | ML Component Description | 1 | MLCS Sec 2 | v1.0 | |
| [E] | Safety Requirements Allocated to ML Component | 2 | 1 | SRS Sec 3.2 | v1.0 |
| [F] | ML Assurance Scoping Argument Pattern | 1 | SRS Sec 5 | v1.0 | |
| [G] | ML Safety Assurance Scoping Argument | 1 | SRS Sec 6 | v1.0 | |
| [H] | ML Safety Requirements | 3, 4, 5 | 2 | SRS Sec 3.3 | v1.0 |
| [I] | ML Safety Requirements Argument Pattern | 2 | SRS Sec 7 | v1.0 | |
| [J] | ML Safety Requirements Validation Results | 2 | SRS Sec 8 | v1.0 | |
| [K] | ML Safety Requirements Argument | 2 | SRS Sec 9 | v1.0 | |
| [L] | Data Requirements | 3 | DMS Sec 2 | v1.0 | |
| [M] | Data Requirements Justification Report | 3 | DMS Sec 3 | v1.0 | |
| [N] | Development Data | 3 | AI Sweden | v1.0 | |
| [O] | Internal Test Data | 3 | AI Sweden | v1.0 | |
| [P] | Verification Data | 3 | AI Sweden | v1.0 | |
| [Q] | Data Generation Log | 3 | DMS Sec 4 | v1.0 | |
| [R] | ML Data Argument Pattern | 3 | DMS Sec 5 | v1.0 | |
| [S] | ML Data Validation Results | 3 | DMS Sec 6 | v1.0 | |
| [T] | ML Data Argument | 3 | DMS Sec 7 | v1.0 | |
| [U] | Model Development Log | 4 | MLCS Sec 3 | v1.0 | |
| [V] | ML Model | 5, 6 | 4 | Model | v0.99 |
| [W] | ML Learning Argument Pattern | 4 | MLCS Sec 5 | v1.0 | |
| [X] | Internal Test Results | 4 | Protocols | v0.99 | |
| [Y] | ML Learning Argument | 4 | MLCS Sec 6 | v1.0 | |
| [Z] | ML Verification Results | 5 | Protocols | v0.99 | |
| [AA] | Verification Log | 5 | STS Sec 3 | v1.0 | |
| [BB] | ML Verification Argument Pattern | 5 | STS Sec 5 | v1.0 | |
| [CC] | ML Verification Argument | 5 | STS Sec 6 | v1.0 | |
| [DD] | Erroneous Behaviour Log | 6 | DS Sec 4 | v1.0 | |
| [EE] | Operational Scenarios | 6 | STS Sec 4.1 | v1.0 | |
| [FF] | Integration Testing Results | 6 | Protocols | v0.99 | |
| [GG] | ML Deployment Argument Pattern | 6 | DS Sec 5 | v1.0 | |
| [HH] | ML Deployment Argument | 6 | DS Sec 6 | v1.0 |