Historical Runtime Governance Asset™
Overview
Historical Runtime Governance Asset™ (Historical RGA, abbreviated as RGA) is the organizational runtime knowledge asset of Atlas Runtime Governance™.
Every completed Runtime Investigation can preserve its authorized investigation results as Historical RGA for future investigation reuse.
Historical RGA preserves investigation knowledge, not runtime data.
It answers:
"In the past, how did we investigate similar Runtime Execution Failures (REF)?"
Rather than:
"What happened at that time?"
Historical RGA Architecture
Completed Investigation
│
▼
Investigation Result
│
▼
Lesson Learned
│
▼
Create Historical RGA
│
▼
Historical Repository
│
▼
Future Runtime Investigation
Historical RGA does not participate in Runtime Observation.
It belongs to knowledge retention after Runtime Investigation completion.
Historical RGA Domains
Atlas supports multiple Historical RGA Knowledge Domains.
Currently, there are two main categories.
Historical Runtime Governance Asset
│
├── OEM Historical RGA
└── Sensor Historical RGA
Both use the same data model.
The difference lies in investigation boundaries and knowledge scope.
OEM Historical RGA
OEM Historical RGA preserves historical experience from OEM Runtime Investigations.
It typically involves multiple Runtime Surfaces.
For example:
- Camera
- LiDAR
- IMU
- USB
- Ethernet
- CAN
- Linux Runtime
- ROS Runtime
- Multiple Sensor Interaction
OEM Historical RGA focuses more on:
- Runtime Execution Failure
- Cross-system Investigation
- Multiple Runtime Surface Correlation
- OEM Investigation Workflow
- OEM Lesson Learned
Typical case:
Unexpected Stop
↓
LiDAR
+
Ethernet
+
Linux Driver
+
ROS Topic
Sensor Historical RGA
Sensor Historical RGA preserves historical investigation experience from Sensor Manufacturers or FAEs.
The investigation target focuses on Sensor Runtime.
For example:
- Camera Runtime
- LiDAR Runtime
- IMU Runtime
- Driver Runtime
- Sensor Interface
- Firmware Runtime
The same Sensor issue may come from different deployment environments.
For example:
- Restaurant
- Factory
- Warehouse
- Hospital
- Hotel
Sensor Historical RGA retains:
Investigation experience with the same Sensor product across different runtime environments.
Shared Runtime Governance Model
OEM Historical RGA and Sensor Historical RGA use a unified data model.
A Historical RGA can contain:
Historical RGA
│
├── Investigation Summary
├── Runtime Pattern
├── Investigation Path
├── Excluded Path
├── Investigation Result (IR)
├── Lesson Learned (LL)
├── Runtime Surface References
├── Evidence References
├── Reuse Metadata
└── Authorization Metadata
Historical RGA does not store Runtime Dataset.
Evidence is always referenced through Evidence Pack.
Investigation Knowledge
Historical RGA preserves investigation knowledge.
Including:
Runtime Pattern
For example:
- Camera Frame Drop
- LiDAR Packet Loss
- USB Disconnect
- Driver Restart
- Runtime Timeout
Patterns help establish investigation direction for the future.
Investigation Path
Records investigation steps that were executed in the past.
For example:
Verify Connection
Review Timeline
Inspect Driver
Compare Historical Pattern
Excluded Path
Records directions that have been verified as irrelevant.
For example:
- Power Normal
- Ethernet Stable
- Driver Active
Avoids repeated troubleshooting in the future.
Investigation Result (IR)
References Investigation Result.
Maintains original engineering records.
Lesson Learned (LL)
References Lesson Learned.
Retains reusable engineering experience.
Historical Recall
New Runtime Investigations can retrieve Historical RGA.
Current REF
│
▼
Historical Recall
│
▼
Candidate Historical RGA
The goal of Recall:
- Provide investigation entry points
- Provide Runtime Pattern
- Provide Investigation Path
- Provide Lesson Learned
Atlas does not automatically generate investigation conclusions.
Partial Match
Historical Recall supports Partial Match.
Even if:
- Robot is different
- Sensor is different
- Deployment is different
- Software Version is different
Historical RGA may still have reference value.
Atlas always follows:
Partial Match takes priority over no recall at all.
Why Retrieved
Each Recall records:
why_retrieved
For example:
- Similar Runtime Pattern
- Similar Runtime Surface
- Similar Investigation Path
Helps engineers understand:
Why viewing this Historical RGA is recommended.
Historical Repository
Each organization maintains its own Historical Repository.
OEM
│
└── OEM Historical RGA Repository
Sensor Manufacturer
│
└── Sensor Historical RGA Repository
Atlas does not require both parties to share Repository.
Each organization has:
- Independent Repository
- Independent permissions
- Independent lifecycle
Cross-domain Collaboration
When OEM Runtime Investigation involves a Sensor Candidate:
OEM Investigation
│
▼
OEM Historical Recall
│
▼
Sensor Engagement Pack
│
▼
Sensor Historical Recall
│
▼
Sensor Investigation
OEM retrieves OEM Historical RGA.
Sensor Manufacturer retrieves Sensor Historical RGA.
What both parties share is:
- Evidence Pack Reference
- Investigation Context
- Sensor Engagement Pack
Rather than sharing the entire Historical Repository.
Design Principles
Historical RGA follows these principles:
- Preserve investigation knowledge, not Runtime Dataset
- Reference Evidence Pack, not copy Evidence
- OEM and Sensor use unified Schema
- OEM and Sensor have independent Repository
- Support Partial Match
- Provide why_retrieved for each Recall
- Do not automatically confirm Root Cause
- Do not automatically copy historical conclusions
- Respect customer authorization boundaries
Historical RGA vs Evidence Pack
Evidence Pack answers:
What happened at that time?
Historical RGA answers:
How were similar issues investigated in the past?
Evidence Pack preserves runtime evidence.
Historical RGA preserves investigation knowledge.
Together they form the foundation of Runtime Investigation.
Summary
Historical Runtime Governance Asset™ is the organizational knowledge asset of Atlas Runtime Governance™.
OEM and Sensor Manufacturer can each maintain their own Historical RGA Repository, preserving Investigation Pattern, IR, LL, and investigation experience under a unified data model.
Future Runtime Investigations no longer start from zero, but from the organization's long-term accumulated runtime knowledge.
Next Reading
- Investigation Context
- Investigation Tier Candidate
- Sensor Engagement Pack™
- Assist Vault™