Research White Paper
Agentic Graph RAG for Intelligent Mobility, Industrial IoT, Renewable Energy and Grid-Edge Systems
A Unified AI Architecture for Vehicle Diagnostics, EV/Hybrid Systems, Solar PV, BESS, EV Charging and Industrial Maintenance
Target Markets: India · United States · Canada · United Kingdom
Research Organizations: IAS-Research.com · KeenComputer.com
Technology/Hardware Ecosystem: KeenDirect.com
Primary Research Platform: OBD-AI / Mobility & Energy AI
Reference Development Environment: Kubuntu/Ubuntu Linux · Docker · RAGFlow · Neo4j · Ollama · Hugging Face · Python
Executive Summary
Industrial systems are undergoing a convergence of AI, IoT, edge computing, renewable energy, electrification and intelligent automation.
A modern automobile is no longer simply a mechanical machine. It contains:
- dozens of electronic control units,
- sensors and actuators,
- CAN networks,
- embedded software,
- batteries,
- power electronics,
- wireless connectivity,
- diagnostic interfaces.
Similarly, a modern energy installation may contain:
- solar PV,
- smart inverters,
- battery energy storage,
- EV chargers,
- meters,
- building-management systems,
- industrial controllers,
- distributed energy resources,
- grid interfaces.
Factories add another layer:
- PLCs,
- SCADA,
- robots,
- CNC machines,
- industrial sensors,
- motors,
- drives,
- machine vision,
- MES,
- ERP,
- predictive-maintenance systems.
The common problem is that the knowledge required to operate and maintain these systems is fragmented.
It exists in:
- service manuals,
- engineering documents,
- technical standards,
- wiring diagrams,
- PLC documentation,
- maintenance records,
- sensor data,
- CAN messages,
- SCADA data,
- IoT telemetry,
- alarm histories,
- engineering models,
- simulation results.
This paper proposes a unified architecture based on:
Industrial IoT + Edge Computing + RAG + Graph RAG + Agentic AI + Digital Twins + Engineering Knowledge Graphs.
The initial OBD-AI automotive platform therefore becomes the first implementation of a much broader Intelligent Engineering Maintenance Platform.
The architecture can support:
Automotive ↓ Hybrid / EV ↓ EV Charging ↓ Solar PV ↓ BESS ↓ Grid Edge ↓ Industrial IoT ↓ Industrial Equipment ↓ Predictive Maintenance
This creates a strategic technology platform in which IAS-Research provides research, architecture and intellectual-property development; KeenComputer provides software engineering, infrastructure and deployment; and KeenDirect provides hardware, components, edge-computing equipment and commercialization support.
1. Research Objective
The objective of this research is to investigate whether an open and modular AI architecture can provide a common intelligence layer across heterogeneous engineered systems.
The central research question is:
Can Agentic Graph RAG combine engineering documentation, equipment knowledge graphs, real-time Industrial IoT data and digital-twin information to support explainable diagnosis, repair, maintenance and optimization across vehicles, EVs, renewable-energy systems and industrial systems?
The proposed answer is an architecture built around five layers:
1. Document Intelligence 2. Knowledge Graph 3. Industrial IoT / Telemetry 4. Agentic AI 5. Digital Twin / Engineering Models
2. From OBD-AI to an Intelligent Engineering Platform
The original OBD-AI concept is:
Service Manuals ↓ RAGFlow ↓ Vector RAG ↓ LLM ↓ Diagnostic Assistant
The expanded architecture is:
INTELLIGENT ENGINEERING AI AI Agent │ ┌───────────────┼───────────────┐ │ │ │ Vehicle Energy Industrial │ │ │ OBD/CAN PV/BESS/EVSE IIoT │ │ │ └───────────────┼───────────────┘ │ Knowledge Graph │ Vector + Graph RAG │ Engineering Data │ Digital Twin
This allows the same core architecture to support multiple domains.
3. Why Graph RAG Is Important
Conventional RAG primarily retrieves semantically similar text.
Graph RAG adds relationships between entities.
Microsoft's GraphRAG documentation describes a pipeline that extracts entities, relationships and claims, performs community detection and generates summaries that can subsequently be used for retrieval. (GitHub)
For vehicle diagnostics:
P0171 ↓ Lean Condition ↓ Fuel System ↓ MAF ↓ Vacuum Leak ↓ Fuel Pressure ↓ Diagnostic Test
For solar:
Low PV Output ↓ PV String ↓ Combiner ↓ Inverter ↓ Grid
For industrial machinery:
High Motor Temperature ↓ Motor ↓ Bearing ↓ Lubrication ↓ Vibration ↓ Maintenance Procedure
The same knowledge-graph concept can therefore be reused across domains.
4. RAGFlow as the Initial Document Intelligence Layer
RAGFlow remains a logical starting point for the project.
The purpose is not to permanently lock OBD-AI into RAGFlow.
Instead:
RAGFlow becomes the initial document-understanding and RAG engine inside a modular architecture.
The document layer should process:
- service manuals,
- repair manuals,
- industrial equipment manuals,
- solar inverter manuals,
- BESS manuals,
- EV charger manuals,
- PLC documentation,
- maintenance manuals,
- engineering specifications,
- troubleshooting guides.
The architecture should allow later evaluation of:
- Haystack,
- LlamaIndex,
- LightRAG,
- Microsoft GraphRAG,
- R2R,
- custom retrieval pipelines.
Microsoft's current GraphRAG repository describes itself as largely in maintenance mode while continuing bug fixes and dependency updates, which is another reason to treat it as a research/reference component rather than making the entire commercial platform dependent on it. (GitHub)
5. Document Intelligence Architecture
PDF / Manual / Engineering Document │ ↓ Docling / OCR │ ↓ Structured Data │ ↓ RAGFlow │ ┌─────┴─────┐ ↓ ↓ Vector Metadata Store │ ↓ Retrieval
Documents should retain:
- manufacturer,
- model,
- revision,
- year,
- component,
- system,
- document type,
- page number,
- section,
- source,
- revision date.
This metadata becomes critical when several generations of a product exist.
6. Vector RAG + Graph RAG
The recommended OBD-AI architecture is hybrid.
Query │ ┌──────────┴──────────┐ ↓ ↓ Vector Search Graph Search │ │ RAGFlow/Qdrant Neo4j │ │ └──────────┬──────────┘ ↓ Evidence Fusion ↓ AI Agent ↓ Response
Vector RAG answers:
"Where does the manual discuss P0171?"
Graph RAG answers:
"What components, symptoms, tests and procedures are related to P0171?"
Agentic reasoning answers:
"Given this vehicle's actual sensor readings, which diagnostic test should be performed next?"
7. Automotive Domain
The automotive domain remains the first OBD-AI implementation.
The system covers:
- ICE vehicles,
- hybrid vehicles,
- EVs,
- commercial vehicles,
- automotive electronics,
- OBD-II,
- CAN,
- ECU diagnostics.
8. ICE Diagnostics
The knowledge graph can model:
Vehicle ├── Engine ├── ECU ├── Sensor ├── Actuator ├── Fuel System ├── Ignition ├── Emissions ├── Transmission └── DTC
Example:
P0171 ↓ Lean Condition ↓ MAF / Vacuum / Fuel Pressure ↓ Test Procedure ↓ Repair
9. Hybrid Vehicle Intelligence
Hybrid systems combine:
ICE + Electric Motor + Battery + Inverter + BMS + Regenerative Braking + Control Software
A hybrid fault may cross several domains.
Therefore:
Engine ↕ Hybrid ECU ↕ Battery ↕ Inverter ↕ Motor
Graph RAG provides a natural mechanism for representing these relationships.
10. Electric Vehicle Intelligence
The EV architecture adds:
Battery Pack BMS Inverter Motor DC/DC Converter Onboard Charger Thermal Management Charging System
The diagnostic agent can correlate:
- SOC,
- SOH,
- cell voltage,
- temperature,
- charging rate,
- BMS alarms,
- inverter status,
- CAN messages.
The platform should initially remain read-oriented and advisory, with high-voltage repair decisions remaining under qualified human control.
11. EV Charging Infrastructure
EV charging extends the system beyond the automobile.
Grid ↓ Building ↓ EVSE ↓ EV
A charging fault could originate in:
- vehicle,
- charger,
- building,
- solar system,
- BESS,
- grid,
- communications,
- protection system.
Therefore, the diagnostic graph needs to cross system boundaries.
12. Solar PV
Solar becomes another knowledge domain.
PV Module ↓ String ↓ Combiner ↓ DC Protection ↓ Inverter ↓ AC Protection ↓ Meter ↓ Grid
The AI can combine:
- inverter alarms,
- power output,
- voltage,
- current,
- weather,
- irradiance,
- historical production,
- maintenance records.
13. Battery Energy Storage
BESS introduces:
Battery ↓ BMS ↓ PCS/Inverter ↓ Switchgear ↓ Grid
The knowledge graph can represent:
- cells,
- modules,
- packs,
- BMS,
- thermal systems,
- contactors,
- PCS,
- alarms,
- SOC,
- SOH.
14. Grid-Edge Renewable Energy
The grid is becoming increasingly distributed.
Instead of:
Central Generation ↓ Transmission ↓ Distribution ↓ Customer
the modern grid increasingly includes:
Grid │ ┌───────────┼───────────┐ │ │ │ Solar BESS EV │ │ │ └───────────┼───────────┘ │ Building
This creates a grid-edge knowledge problem.
The AI must understand relationships between:
- PV,
- BESS,
- EV,
- EVSE,
- building load,
- smart inverter,
- utility grid.
15. Industrial IoT
The next major expansion is Industrial IoT (IIoT).
NIST describes IIoT in the context of interconnected smart computing and networking technologies supporting industrial automation, reliability, control and intelligence. It also emphasizes the distinct requirements of industrial systems compared with consumer IoT, including control, networking, computing and quality-of-service considerations. (NIST)
Industrial systems can contain:
Sensors ↓ PLC ↓ SCADA ↓ MES ↓ ERP
while machines contain:
Motor Drive Bearing Pump Valve Actuator Robot CNC
The AI platform can become a knowledge layer across these systems.
16. Industrial IoT Data Pipeline
A scalable IIoT architecture can be:
Industrial Equipment │ ↓ Sensors / PLC / CAN / Modbus │ ↓ Edge Gateway │ ↓ MQTT / OPC UA │ ↓ Time-Series Database │ ├────────→ AI Analytics │ └────────→ Knowledge Graph │ ↓ AI Agent
NIST research on scalable IIoT data pipelines specifically addresses the challenge of integrating heterogeneous shop-floor data sources and distributing the resulting information to analytics applications. (NIST)
17. Industrial Equipment Diagnostics
Consider an industrial motor.
Telemetry:
RPM Current Voltage Temperature Vibration Power
Maintenance manuals provide:
Bearing specifications Lubrication procedures Temperature limits Vibration limits Maintenance schedules
The AI combines:
Telemetry + Manual + Historical data + Knowledge graph
and identifies potential maintenance conditions.
The same architecture used for:
P0171 → automotive
can be applied to:
High vibration → industrial motor
and:
Low output → solar inverter.
18. Industrial Predictive Maintenance
The evolution is:
Reactive Maintenance ↓ Preventive Maintenance ↓ Condition-Based Maintenance ↓ Predictive Maintenance ↓ AI-Assisted Prescriptive Maintenance
The AI system can monitor:
- vibration,
- temperature,
- current,
- pressure,
- flow,
- energy consumption,
- cycle time,
- alarms.
The objective is not simply to predict failure.
It should explain:
What changed, why it matters, what evidence supports the conclusion, and what inspection should be performed next.
19. Digital Thread
This architecture also creates a digital thread.
NIST research describes digital-thread architecture as a way of connecting product-lifecycle systems so shared data can be used across heterogeneous systems and emphasizes the challenge of contextualizing data from those systems. (NIST)
For OBD-AI:
Design ↓ Manufacturing ↓ Vehicle ↓ Telemetry ↓ Service ↓ Repair ↓ Maintenance
For industrial equipment:
Engineering ↓ Manufacturing ↓ Installation ↓ Operation ↓ Maintenance ↓ Replacement
The knowledge graph becomes the connective tissue.
20. Digital Twin
The long-term architecture should incorporate digital twins.
Physical Asset ↕ Digital Twin ↕ Telemetry ↕ Knowledge Graph ↕ AI Agent
Possible digital twins include:
- vehicle digital twin,
- EV battery digital twin,
- solar PV digital twin,
- BESS digital twin,
- industrial-machine digital twin,
- microgrid digital twin.
21. Industrial AI Architecture
The resulting architecture is:
AI AGENT LAYER │ ┌──────────────┼──────────────┐ │ │ │ Vehicle Energy Industrial Agent Agent Agent │ │ │ └──────────────┼──────────────┘ │ KNOWLEDGE LAYER │ ┌─────────────┴─────────────┐ │ │ Vector RAG Graph RAG │ │ RAGFlow Neo4j │ │ └─────────────┬─────────────┘ │ DATA/EDGE LAYER │ ┌────────────┬───────┼────────┬───────────┐ │ │ │ │ │ OBD CAN PLC SCADA IoT │ │ │ │ │ └────────────┴───────┼────────┴───────────┘ │ PHYSICAL ASSETS
22. Edge Computing
Industrial AI cannot always depend upon the cloud.
Reasons include:
- latency,
- privacy,
- connectivity,
- reliability,
- cybersecurity,
- operational continuity.
Therefore:
Cloud │ Central AI │ Knowledge DB │ ──────┼────── │ Edge AI │ Industrial Site
The edge can perform:
- data filtering,
- anomaly detection,
- local inference,
- protocol conversion,
- buffering,
- safety-related monitoring.
Cloud infrastructure can provide:
- model management,
- fleet analytics,
- knowledge-base management,
- reporting,
- cross-site analytics.
23. Cybersecurity
The expanded system creates a larger attack surface.
Security must cover:
Sensor ↓ Edge Gateway ↓ Network ↓ API ↓ AI ↓ Database ↓ Cloud
Security requirements include:
- authentication,
- authorization,
- encryption,
- network segmentation,
- firewall,
- secure OTA updates,
- secrets management,
- audit logs,
- vulnerability management,
- backup,
- incident response.
For industrial environments, AI should initially remain outside safety-critical control loops unless it has been engineered, validated and certified for the specific application.
24. Role of IAS-Research.com
IAS-Research should function as the research and intellectual architecture organization.
Its responsibilities can include:
AI Research
- RAG
- Graph RAG
- agentic AI
- multimodal AI
- knowledge graphs
- AI evaluation.
Engineering Research
- MBSE
- SysML/UML
- digital twins
- SystemC/TLM
- embedded AI
- TinyML
- edge AI.
Energy Research
- PV
- BESS
- smart inverter
- DER
- grid edge
- power quality
- EV-grid integration.
Automotive Research
- OBD
- CAN
- EV
- hybrid
- battery diagnostics.
Industrial Research
- IIoT
- predictive maintenance
- industrial cybersecurity
- edge computing.
IAS-Research can produce:
- white papers,
- reference architectures,
- feasibility studies,
- prototypes,
- technical research,
- evaluation frameworks,
- engineering models.
25. Role of KeenComputer.com
KeenComputer should serve as the engineering, implementation and deployment organization.
Its responsibilities include:
Software
- Python
- FastAPI
- web applications
- APIs
- databases
- dashboards.
AI Infrastructure
- RAGFlow
- Neo4j
- Ollama
- Docker
- Linux
- vector databases.
Industrial Integration
- MQTT
- OPC UA
- Modbus
- REST
- CAN
- OBD-II.
Infrastructure
- VPS
- cloud
- edge servers
- firewalls
- monitoring
- backup.
DevOps
Development ↓ Testing ↓ Staging ↓ Production ↓ Monitoring
KeenComputer therefore becomes the implementation arm that turns IAS-Research architectures into deployable systems.
26. Role of KeenDirect.com
KeenDirect should provide the hardware and technology supply-chain layer.
This is particularly important as the project moves from software R&D to physical Industrial IoT deployments.
Potential hardware categories include:
Automotive
- OBD adapters
- CAN interfaces
- STM32 development boards
- edge computers
- diagnostic interfaces.
Industrial IoT
- industrial gateways
- sensors
- PLC accessories
- networking equipment
- industrial PCs.
Renewable Energy
- energy meters
- sensors
- gateways
- monitoring hardware.
EV
- EVSE components
- charging interfaces
- communication hardware.
Computing
- edge servers
- networking
- storage
- embedded computers.
The business relationship becomes:
IAS-Research Research / Architecture ↓ KeenComputer Software / Integration ↓ KeenDirect Hardware / Components / Supply Chain ↓ Customer System
27. The Three-Company Ecosystem
The combined model can be summarized as:
|
Organization |
Primary role |
Deliverable |
|---|---|---|
|
IAS-Research |
Research & innovation |
Architecture, IP, models, research |
|
KeenComputer |
Engineering & implementation |
Software, AI, IoT, infrastructure |
|
KeenDirect |
Hardware & supply chain |
Edge hardware, components, systems |
This provides a complete:
Research → Engineering → Hardware → Deployment → Operations
lifecycle.
28. Industrial IoT + KeenDirect Supply Chain
KeenDirect becomes especially relevant when customers need physical implementation.
For example:
Customer ↓ Industrial AI requirement ↓ IAS-Research Architecture ↓ KeenComputer Software + AI ↓ KeenDirect Hardware ↓ Installation ↓ Monitoring ↓ Maintenance
This creates a potential full-lifecycle offering rather than a standalone consulting engagement.
29. Four-Market Strategy
The platform can be developed as a common core for:
India
- automotive workshops,
- EV service,
- solar,
- industrial IoT,
- distributed energy,
- cost-sensitive edge deployments.
USA
- independent automotive repair,
- EV fleets,
- EV charging,
- solar,
- BESS,
- industrial maintenance.
Canada
- automotive service,
- EVs,
- renewable energy,
- industrial systems,
- cold-weather applications,
- distributed energy.
UK
- EV charging,
- solar,
- home energy,
- BESS,
- industrial IoT,
- smart energy.
The technical core remains common while:
- privacy,
- regulatory,
- standards,
- data residency,
- product certification,
- market requirements
are configured by jurisdiction and application.
30. Commercial Product Family
The research can ultimately generate several products.
OBD-AI
Automotive diagnostic intelligence.
EV-AI
EV and battery diagnostics.
EVSE-AI
EV charger diagnostics.
Solar-AI
Solar PV maintenance intelligence.
BESS-AI
Battery-storage diagnostics.
GridEdge-AI
DER and smart-grid maintenance.
Industrial-AI
Industrial IoT and predictive maintenance.
Engineering-AI
Engineering-document and knowledge assistant.
All can share:
RAG + Graph + Agent + Telemetry + Digital Twin
31. Common Industrial Knowledge Graph
The most strategically important research asset could be a generic engineering knowledge graph.
Asset │ ├── Component ├── Sensor ├── Controller ├── Communication ├── Measurement ├── Fault ├── Symptom ├── Cause ├── Test ├── Procedure ├── Maintenance └── Repair
This ontology can represent:
Vehicle Solar Plant BESS EVSE Factory Motor Pump Robot Inverter Microgrid
That creates a reusable foundation for future products.
32. Agent Architecture
A multi-agent implementation could include:
Supervisor Agent │ ┌────────────────┼────────────────┐ │ │ │ Automotive Energy Industrial Agent Agent Agent │ │ │ OBD/CAN PV/BESS/EVSE PLC/SCADA/IIoT │ │ │ └────────────────┼────────────────┘ │ Knowledge Graph │ RAG Platform
Specialized agents should not independently invent conclusions.
They should return:
Evidence Source Observation Hypothesis Recommended test Confidence
to the supervisor.
33. Example Cross-Domain Use Case
Consider:
An EV fleet depot with rooftop solar, BESS and ten EV chargers.
The system has:
20 EVs Solar PV BESS 10 EVSE Building load Grid connection
One EV reports slow charging.
The AI investigates:
EV ↓ Onboard Charger ↓ EVSE ↓ Building ↓ Energy Management ↓ Solar ↓ BESS ↓ Grid
It discovers that the EV is healthy but the building energy-management system has reduced charging power because the building is approaching its grid-demand limit.
This is a fundamentally different diagnostic problem from:
"Which component of the EV is broken?"
The system needs cross-domain knowledge.
34. Industrial Example
A manufacturing plant reports:
"Production-line throughput has declined."
The AI can correlate:
MES ↓ Machine ↓ PLC ↓ Motor ↓ Vibration ↓ Temperature ↓ Maintenance history ↓ Manufacturer manual
It may discover that a motor is operating outside its historical vibration pattern.
The agent can retrieve the appropriate maintenance procedure and identify the next inspection.
This illustrates the transition:
RAG chatbot → engineering diagnostic agent.
35. Digital Thread Across the Ecosystem
The ultimate architecture becomes:
Design ↓ Simulation ↓ Manufacturing ↓ Installation ↓ Operation ↓ Telemetry ↓ AI Diagnostics ↓ Maintenance ↓ Repair ↓ Replacement
This is the engineering digital thread.
The knowledge graph provides the relationships.
RAG provides the documentary evidence.
IoT provides the real-world observations.
AI agents provide reasoning and workflow orchestration.
Digital twins provide system-level models.
36. Research and Commercialization Roadmap
Phase 1 — OBD-AI
RAGFlow + Service Manuals + Ollama + OBD
Phase 2 — Graph RAG
Neo4j + Vehicle Ontology + Vector RAG
Phase 3 — EV/Hybrid
Battery BMS Inverter EVSE
Phase 4 — Solar/BESS
PV Inverter BESS Energy Management
Phase 5 — IIoT
PLC SCADA MQTT OPC UA Industrial Sensors
Phase 6 — Digital Twin
Simulation + Telemetry + Knowledge Graph
Phase 7 — Commercial Platform
Mobility AI + Energy AI + Industrial AI
37. Recommended Technology Stack
Operating System
- Kubuntu/Ubuntu LTS
- Linux
Containers
- Docker
- Docker Compose
RAG
- RAGFlow
- Docling
- Vector database
- BM25/hybrid retrieval
Graph
- Neo4j
- Cypher
- Graph RAG
AI
- Ollama
- Hugging Face
- local/open models
- embedding models
- rerankers
Agent
- Python
- Haystack
- LlamaIndex
- LangChain where appropriate
- MCP/tool interfaces
IoT
- MQTT
- OPC UA
- Modbus
- CAN
- OBD-II
Edge
- ARM
- STM32
- industrial PCs
- embedded Linux
Engineering
- MATLAB/Simulink
- PSCAD
- SystemC/TLM
- SysML/UML
- digital twins
38. Security and Governance
The platform should follow a layered security model:
Physical Device ↓ Edge ↓ Network ↓ API ↓ AI Agent ↓ Knowledge Graph ↓ Database
The system should incorporate:
- authentication,
- authorization,
- encryption,
- network segmentation,
- audit trails,
- backup,
- vulnerability management,
- secure configuration,
- model governance,
- data retention.
Industrial systems require particular caution because IIoT can interact with command-and-control environments. NIST's IIoT research emphasizes the distinct requirements around control, networking and computing compared with consumer IoT. (NIST)
39. AI Safety Principle
The platform should follow:
Observe → Analyze → Explain → Recommend → Human Approve → Act
rather than:
Observe → AI decides → Automatically controls equipment.
For safety-critical systems:
AI Recommendation ↓ Engineering Validation ↓ Human Authorization ↓ Controlled Action ↓ Verification
This is particularly important for:
- high-voltage EV systems,
- BESS,
- grid equipment,
- industrial machinery,
- safety systems.
40. Research Metrics
The project should establish quantitative evaluation.
RAG
Graph
Agent
IIoT
Business
- diagnostic time reduction
- maintenance downtime reduction
- technician productivity
- system availability.
41. Intellectual Property Strategy
The open-source infrastructure should be treated as a foundation rather than the complete product.
Potential proprietary assets include:
- automotive ontology,
- industrial ontology,
- mobility-energy ontology,
- diagnostic workflows,
- agent architectures,
- evaluation datasets,
- domain-specific prompts,
- telemetry-processing algorithms,
- digital-twin models,
- integration software,
- customer dashboards.
The commercial architecture becomes:
Open Source + IAS-Research IP + KeenComputer Engineering + KeenDirect Hardware = Commercial Solution
42. Strategic Positioning
The combined organization can position itself around:
AI Engineering for Connected Physical Systems
Vehicles EVs Hybrid Vehicles EV Chargers Solar BESS Microgrids Factories Industrial Equipment
AI + RAG + Graph + IoT + Edge + Digital Twin
43. The IAS-Research / KeenComputer / KeenDirect Flywheel
The three organizations can create a continuous development cycle:
This creates a research-to-commercialization feedback loop.
44. Final Architecture
The complete vision can be represented as:
45. Conclusion
The combined research changes the strategic scope of OBD-AI.
It begins with a very practical problem:
How can AI help a technician diagnose a vehicle using service manuals and OBD/CAN information?
How can AI understand, diagnose and maintain complex connected physical systems?
That second question encompasses:
- ICE vehicles,
- hybrid vehicles,
- EVs,
- batteries,
- EV chargers,
- solar PV,
- BESS,
- smart inverters,
- grid-edge renewable energy,
- industrial machinery,
- Industrial IoT,
- predictive maintenance,
- digital twins.
The technology foundation is correspondingly broader:
RAGFlow provides a practical starting point for document intelligence.
Docling can be evaluated for complex engineering-document extraction.
Neo4j provides the vehicle/energy/industrial knowledge graph.
Graph RAG provides relationship-aware retrieval; Microsoft's GraphRAG documentation demonstrates the general approach of extracting entities and relationships and using graph/community structures to augment retrieval. (GitHub)
Ollama and Hugging Face can provide local/private model infrastructure.
Industrial IoT supplies real-world operational data.
Edge computing provides local processing.
Digital twins and MBSE provide engineering-system context.
Agentic AI connects all of these components into diagnostic and maintenance workflows.
The strategic roles are equally clear:
IAS-Research = Research, Architecture, Innovation and IP KeenComputer = Software Engineering, AI, IoT, Infrastructure and Deployment KeenDirect = Hardware, Components, Edge Computing and Supply Chain
Together, these can form a complete:
Research → Design → Prototype → Software → Hardware → Deployment → Operations → Feedback → Research
ecosystem.
The most important long-term opportunity is therefore not a standalone vehicle chatbot. It is a common AI engineering platform for connected physical systems, where the same knowledge, RAG, graph, IoT and agent technologies can be progressively extended from OBD-AI to EV-AI, Solar-AI, BESS-AI, GridEdge-AI and Industrial-AI.
NIST's work on intelligent manufacturing architectures, industrial IoT and digital-thread integration provides a useful engineering foundation for this broader direction. (NIST)
Selected technical references
- Microsoft GraphRAG documentation and architecture. (GitHub)
- Microsoft GraphRAG current repository status and maintenance guidance. (GitHub)
- NIST, A Survey on Industrial Internet of Things: A Cyber-Physical Systems Perspective. (NIST)
- NIST, Scalable Data Pipeline Architecture to Support the Industrial Internet of Things. (NIST)
- NIST, Reference Architecture to Integrate Heterogeneous Manufacturing Systems for the Digital Thread. (NIST)
- NIST, Reference Architecture for Smart Manufacturing. (NIST)
- NIST, Intelligent Systems Architecture for Manufacturing (ISAM). (NIST)
Engineering References
A. Retrieval-Augmented Generation, Graph RAG and Knowledge Graphs
- Microsoft Research, GraphRAG: From Local to Global—A Graph RAG Approach to Query-Focused Summarization of Large Documents, Microsoft Research, 2024. The Microsoft GraphRAG documentation describes a structured, hierarchical approach that extracts entities and relationships, builds community hierarchies and summaries, and uses these structures for retrieval-augmented generation. (Microsoft GitHub)
- Edge, D., Trinh, H., Cheng, N., Bradley, J., Chao, A., Mody, A., Truitt, S., and Larson, J., From Local to Global: A Graph RAG Approach to Query-Focused Summarization, arXiv, 2024.
- Microsoft Research, GraphRAG Documentation and Getting Started Guide, Microsoft GraphRAG Project. The documentation provides implementation guidance, indexing architecture, configuration and query workflows. (Microsoft GitHub)
- Neo4j, GraphRAG User Guide and Developer Documentation, Neo4j. The Neo4j GraphRAG framework provides vector retrieval, graph retrieval and LLM-based generation using graph databases. (Neo4j Graph Intelligence Platform)
- Neo4j, Enhancing the Accuracy of RAG Applications with Knowledge Graphs, Neo4j Developer Blog, 2024. This reference discusses combining vector retrieval with graph-based relationships to improve contextual retrieval. (Neo4j Graph Intelligence Platform)
- Neo4j, RAG Tutorial: How to Build a RAG System on a Knowledge Graph, Neo4j Developer Blog, 2025. (Neo4j Graph Intelligence Platform)
- Neo4j, What is GraphRAG?, Neo4j Developer Blog, 2026. The article discusses knowledge graphs as a structured context layer for RAG systems and their use in complex, connected information domains. (Neo4j Graph Intelligence Platform)
- Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H., Lewis, M., Yih, W., Rocktäschel, T., Riedel, S., and Kiela, D., “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks,” Advances in Neural Information Processing Systems, Vol. 33, 2020.
- Gao, Y., Xiong, Y., Gao, X., Jia, K., Pan, J., Bi, Y., Dai, Y., Sun, J., Wang, H., and Wang, H., “Retrieval-Augmented Generation for Large Language Models: A Survey,” arXiv, 2023/2024.
- Hogan, A., Blomqvist, E., Cochez, M., d'Amato, C., de Melo, G., Gutierrez, C., Kirrane, S., Gayo, J. E. L., Navigli, R., Neumaier, S., Ngomo, A.-C. N., Polleres, A., Rashid, S. M., Rula, A., Schmelzeisen, L., Sequeda, J., Staab, S., and Zimmermann, A., “Knowledge Graphs,” ACM Computing Surveys, Vol. 54, No. 4, 2021.
- Neo4j, GraphRAG Python Package and Knowledge Graph Retrieval Documentation, Neo4j. The package provides workflows for knowledge-graph creation, retrieval and GraphRAG applications. (Neo4j Graph Intelligence Platform)
B. Large Language Models, Embeddings and AI Frameworks
- Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., and Polosukhin, I., “Attention Is All You Need,” Advances in Neural Information Processing Systems, 2017.
- Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K., “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” Proceedings of NAACL-HLT, 2019.
- Reimers, N. and Gurevych, I., “Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks,” Proceedings of EMNLP-IJCNLP, 2019.
- Hugging Face, Transformers Documentation and Open-Source Machine Learning Platform, Hugging Face.
- Ollama, Ollama Documentation: Running Large Language Models Locally, Ollama.
- RAGFlow Project, RAGFlow Documentation and Open-Source Retrieval-Augmented Generation Engine, RAGFlow.
- LlamaIndex, LlamaIndex Documentation: Data Framework for LLM Applications, LlamaIndex.
- LangChain, LangChain Documentation: Framework for Developing Applications Powered by Language Models, LangChain.
- LangGraph, LangGraph Documentation: Stateful Agent and Workflow Framework, LangChain.
C. Industrial Internet of Things and Cyber-Physical Systems
- Xu, H., Yu, W., Griffith, D. W., and Golmie, N. T., “A Survey on Industrial Internet of Things: A Cyber-Physical Systems Perspective,” IEEE Access, Vol. 6, 2018, pp. 78238–78259. DOI: 10.1109/ACCESS.2018.2884906. This survey addresses IIoT architecture, industrial control, networking, computing, cloud/edge computing and industrial automation. (NIST)
- Lee, J., Bagheri, B., and Kao, H.-A., “A Cyber-Physical Systems Architecture for Industry 4.0-Based Manufacturing Systems,” Manufacturing Letters, Vol. 3, 2015, pp. 18–23.
- Kagermann, H., Wahlster, W., and Helbig, J., Recommendations for Implementing the Strategic Initiative INDUSTRIE 4.0, acatech, Germany, 2013.
- National Institute of Standards and Technology, Reference Architecture for Smart Manufacturing Part 1: Functional Models, NIST Advanced Manufacturing Series 300-1, 2016. The reference architecture addresses manufacturing functions, information flows and integration of manufacturing software and hardware components. (NIST)
- Lu, Y., Riddick, F. H., and Ivezic, N., The Paradigm Shift in Smart Manufacturing System Architecture, National Institute of Standards and Technology, 2016. (NIST)
- NIST, Reference Architecture to Integrate Heterogeneous Manufacturing Systems for the Digital Thread, 2017. This work addresses integration of heterogeneous manufacturing systems and lifecycle information through a digital thread. (NIST)
D. Digital Thread, Digital Twin and Systems Engineering
- Grieves, M. and Vickers, J., “Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems,” in Transdisciplinary Perspectives on Complex Systems, Springer, 2017.
- Tao, F., Qi, Q., Liu, A., and Kusiak, A., “Data-Driven Smart Manufacturing,” Journal of Manufacturing Systems, Vol. 48, 2018, pp. 157–169.
- Tao, F., Zhang, H., Liu, A., and Nee, A. Y. C., “Digital Twin in Industry: State-of-the-Art,” IEEE Transactions on Industrial Informatics, Vol. 15, No. 4, 2019, pp. 2405–2415.
- National Institute of Standards and Technology, Reference Architecture to Integrate Heterogeneous Manufacturing Systems for the Digital Thread, NIST. (NIST)
- International Council on Systems Engineering (INCOSE), Systems Engineering Handbook: A Guide for System Life Cycle Processes and Activities, Wiley/INCOSE.
- Object Management Group, Systems Modeling Language (SysML) Specification, OMG.
E. Industrial Communication and IoT Protocols
- OASIS, MQTT Version 5.0, OASIS Standard, March 2019. MQTT is a lightweight publish/subscribe protocol designed for machine-to-machine and IoT environments. (OASIS Documentation)
- OPC Foundation, OPC Unified Architecture — Part 1: Overview and Concepts, IEC 62541 / OPC UA. OPC UA provides information modeling, communication and interoperability capabilities across industrial systems from devices through enterprise and cloud systems. (OPC UA Online Reference)
- Modbus Organization, Modbus Application Protocol Specification V1.1b3. The specification defines the Modbus application protocol and transaction function codes. (Modbus)
- Modbus Organization, Modbus Security Protocol Specification. The specification extends Modbus with TLS and certificate-based authentication and message integrity mechanisms. (Modbus)
- International Electrotechnical Commission, IEC 62541 — OPC Unified Architecture, IEC.
- International Organization for Standardization, ISO 9506 — Manufacturing Message Specification (MMS), ISO.
- Eclipse Foundation, Eclipse Paho MQTT Client Documentation, Eclipse IoT.
F. Automotive Diagnostics, CAN and Vehicle Networks
- International Organization for Standardization, ISO 11898 — Road Vehicles — Controller Area Network (CAN), ISO.
- International Organization for Standardization, ISO 14229 — Road Vehicles — Unified Diagnostic Services (UDS), ISO.
- International Organization for Standardization, ISO 15765 — Road Vehicles — Diagnostic Communication over Controller Area Network (DoCAN), ISO.
- SAE International, SAE J1979 — E/E Diagnostic Test Modes, SAE International.
- SAE International, SAE J1939 — Serial Control and Communications Heavy-Duty Vehicle Network, SAE International.
- SAE International, SAE J1772 — SAE Electric Vehicle and Plug-in Hybrid Electric Vehicle Conductive Charge Coupler, SAE International.
- ISO/IEC, Road Vehicles — Functional Safety, ISO 26262.
- ISO/SAE, ISO/SAE 21434 — Road Vehicles — Cybersecurity Engineering, ISO/SAE.
- SAE International, SAE J3068 — Electric Vehicle Power Transfer System Using a Three-Phase AC Connection, SAE International.
G. Electric Vehicles, Batteries and EV Charging
- International Electrotechnical Commission, IEC 61851 — Electric Vehicle Conductive Charging System, IEC.
- International Electrotechnical Commission, IEC 62196 — Plugs, Socket-Outlets, Vehicle Connectors and Vehicle Inlets — Conductive Charging of Electric Vehicles, IEC.
- ISO, ISO 15118 — Road Vehicles — Vehicle to Grid Communication Interface, ISO.
- Open Charge Alliance, Open Charge Point Protocol (OCPP), Open Charge Alliance.
- International Electrotechnical Commission, IEC 62660 — Secondary Lithium-Ion Cells for the Propulsion of Electric Road Vehicles, IEC.
- International Organization for Standardization, ISO 12405 — Electrically Propelled Road Vehicles — Test Specification for Lithium-Ion Traction Battery Packs and Systems, ISO.
- Plett, G. L., Battery Management Systems, Volume I: Battery Modeling, Artech House.
- Plett, G. L., Battery Management Systems, Volume II: Equivalent-Circuit Methods, Artech House.
- Plett, G. L., Battery Management Systems, Volume III: Battery State Estimation, Artech House.
H. Renewable Energy, Solar PV, BESS and Grid-Edge Systems
- International Electrotechnical Commission, IEC 61724-1 — Photovoltaic System Performance — Part 1: Monitoring, IEC.
- International Electrotechnical Commission, IEC 62109 — Safety of Power Converters for Use in Photovoltaic Power Systems, IEC.
- International Electrotechnical Commission, IEC 62477-1 — Safety Requirements for Power Electronic Converter Systems and Equipment, IEC.
- International Electrotechnical Commission, IEC 62933 — Electrical Energy Storage Systems, IEC.
- International Electrotechnical Commission, IEC 61850 — Communication Networks and Systems for Power Utility Automation, IEC.
- International Electrotechnical Commission, IEC 61850-7-420 — Communication Networks and Systems for Power Utility Automation — Distributed Energy Resources, IEC.
- IEEE, IEEE Std 1547 — Standard for Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power Systems Interfaces, IEEE.
- IEEE, IEEE Std 2030.5 — IEEE Standard for Smart Energy Profile Application Protocol, IEEE.
- International Energy Agency, Renewables, IEA.
- National Renewable Energy Laboratory, Distributed Energy Resources and Grid-Interactive Efficient Buildings Research, NREL.
- U.S. Department of Energy, Energy Storage Grand Challenge, U.S. Department of Energy.
I. Predictive Maintenance and Industrial AI
- Jardine, A. K. S., Lin, D., and Banjevic, D., “A Review on Machinery Diagnostics and Prognostics Implementing Condition-Based Maintenance,” Mechanical Systems and Signal Processing, Vol. 20, No. 7, 2006, pp. 1483–1510.
- Lei, Y., Li, N., Guo, L., Li, N., Yan, T., and Wang, J., “Machinery Health Prognostics: A Systematic Review from Data Acquisition to RUL Prediction,” Mechanical Systems and Signal Processing, Vol. 104, 2018, pp. 799–834.
- Si, X.-S., Wang, W., Hu, C.-H., and Zhou, D.-H., “Remaining Useful Life Estimation — A Review on the Statistical Data Driven Approaches,” European Journal of Operational Research, Vol. 213, No. 1, 2011, pp. 1–14.
- Mobley, R. K., An Introduction to Predictive Maintenance, 2nd ed., Butterworth-Heinemann, 2002.
- ISO, ISO 17359 — Condition Monitoring and Diagnostics of Machines — General Guidelines, International Organization for Standardization.
J. Edge Computing and Distributed AI
- Shi, W., Cao, J., Zhang, Q., Li, Y., and Xu, L., “Edge Computing: Vision and Challenges,” IEEE Internet of Things Journal, Vol. 3, No. 5, 2016, pp. 637–646.
- Satyanarayanan, M., “The Emergence of Edge Computing,” Computer, Vol. 50, No. 1, 2017, pp. 30–39.
- Abbas, N., Zhang, Y., Taherkordi, A., and Skeie, T., “Mobile Edge Computing: A Survey,” IEEE Internet of Things Journal, Vol. 5, No. 1, 2018, pp. 450–465.
- NIST, Fog Computing Conceptual Model, NIST Special Publication 500-325.
K. AI Safety, Cybersecurity and Trustworthy AI
- Tabassi, E., Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1, National Institute of Standards and Technology, 2023. (NIST)
- National Institute of Standards and Technology, NIST Cybersecurity Framework (CSF) 2.0, NIST Cybersecurity White Paper 29, 2024. (NIST Computer Security Resource Center)
- National Institute of Standards and Technology, NIST Cybersecurity Framework 2.0: Informative References Quick-Start Guide, NIST SP 1347, 2026. (NIST Computer Security Resource Center)
- National Institute of Standards and Technology, Guide to Industrial Control Systems (ICS) Security, NIST SP 800-82.
- National Institute of Standards and Technology, Cybersecurity Framework for Improving Critical Infrastructure Cybersecurity, NIST.
- IEC, IEC 62443 — Security for Industrial Automation and Control Systems, International Electrotechnical Commission.
- ISO/IEC, ISO/IEC 27001 — Information Security Management Systems — Requirements, International Organization for Standardization.
- ISO/IEC, ISO/IEC 27002 — Information Security Controls, International Organization for Standardization.
L. Industrial Automation and Manufacturing
- International Society of Automation, ANSI/ISA-95 — Enterprise-Control System Integration, ISA.
- International Society of Automation, ANSI/ISA-88 — Batch Control, ISA.
- International Society of Automation, ANSI/ISA-99 / IEC 62443 — Security for Industrial Automation and Control Systems, ISA/IEC.
- National Institute of Standards and Technology, Smart Manufacturing Systems Design and Architecture Research, NIST.
- National Institute of Standards and Technology, Intelligent Manufacturing Systems Architecture and Reference Models, NIST.
- Kagermann, H., Change Through Digitization — Value Creation in the Age of Industry 4.0, acatech.
M. Software Architecture, DevOps and Open-Source Infrastructure
- Newman, S., Building Microservices, 2nd ed., O'Reilly Media.
- Richards, M. and Ford, N., Fundamentals of Software Architecture, O'Reilly Media.
- Kleppmann, M., Designing Data-Intensive Applications, O'Reilly Media.
- Burns, B., Beda, J., and Hightower, K., Kubernetes: Up and Running, O'Reilly Media.
- Merkel, D., “Docker: Lightweight Linux Containers for Consistent Development and Deployment,” Linux Journal, 2014.
- Open Container Initiative, Open Container Initiative Specifications, OCI.
- Docker, Docker Documentation and Docker Compose Documentation, Docker.
- Python Software Foundation, Python Documentation, Python Software Foundation.
N. Engineering Modeling and Simulation
- Object Management Group, Systems Modeling Language (SysML) Specification, OMG.
- Object Management Group, Unified Modeling Language (UML) Specification, OMG.
- Object Management Group, Model Driven Architecture (MDA) Guide, OMG.
- Delligatti, L., SysML Distilled: A Brief Guide to SysML, Addison-Wesley.
- Friedenthal, S., Moore, A., and Steiner, R., OMG Systems Modeling Language Tutorial, INCOSE/OMG.
- IEEE, IEEE 1516 — High Level Architecture (HLA) for Modeling and Simulation, IEEE.
- IEEE, IEEE 1666 — SystemC Language Reference Manual, IEEE.
- Accellera Systems Initiative, SystemC Language Reference and Modeling Resources, Accellera.
- MathWorks, MATLAB and Simulink Documentation, MathWorks.
- Siemens/PSCAD, PSCAD Power Systems Simulation Documentation, Manitoba Hydro International.
O. Industrial Data, Semantic Models and Interoperability
- W3C, RDF 1.1 Concepts and Abstract Syntax, World Wide Web Consortium.
- W3C, Web Ontology Language (OWL) 2 Web Ontology Language Document Overview, W3C.
- W3C, SPARQL 1.1 Query Language, W3C.
- Gruber, T. R., “A Translation Approach to Portable Ontology Specifications,” Knowledge Acquisition, Vol. 5, No. 2, 1993, pp. 199–220.
- Studer, R., Benjamins, V. R., and Fensel, D., “Knowledge Engineering: Principles and Methods,” Data & Knowledge Engineering, Vol. 25, Nos. 1–2, 1998, pp. 161–197.
P. Recommended Primary Online Technical Sources
- Microsoft Research / Microsoft GraphRAG Project — GraphRAG architecture, indexing and query documentation. (Microsoft GitHub)
- Neo4j — GraphRAG, knowledge graphs, vector search and graph database documentation. (Neo4j Graph Intelligence Platform)
- NIST — Industrial Internet of Things and cyber-physical systems research. (NIST)
- NIST — Smart Manufacturing reference architectures and digital-thread research. (NIST)
- OASIS — MQTT Version 5.0 Standard. (OASIS Documentation)
- OPC Foundation — OPC Unified Architecture and industrial interoperability specifications. (OPC UA Online Reference)
- Modbus Organization — Modbus and Modbus Security specifications. (Modbus)
- NIST — AI Risk Management Framework. (NIST)
- NIST — Cybersecurity Framework 2.0. (NIST Computer Security Resource Center)
Q. Project-Specific Research and Development Sources
- IAS-Research.com, OBD-AI: Intelligent Vehicle Diagnostics and Maintenance Research Platform, internal research and development program.
- KeenComputer.com, AI, RAG, Graph RAG, Industrial IoT and Digital Transformation Engineering Projects, engineering and implementation resources.
- KeenDirect.com, Industrial Computing, Embedded Systems, Networking, Sensors, Automotive Diagnostic and Energy-System Hardware Supply Chain, technology and hardware sourcing resources.
- IAS-Research.com / KeenComputer.com / KeenDirect.com, Agentic Graph RAG for Intelligent Mobility, Industrial IoT, Renewable Energy and Grid-Edge Systems, consolidated research framework, 2026.