Architecture Overview

global architecture

Our Edge AI solution is a sophisticated hybrid cloud-edge architecture that brings enterprise-grade AI capabilities directly to the field. Here’s how all the components work together to create an intelligent, autonomous system:

🚂 At the Edge: the LEGO Train

Physical Components

  • LEGO Technic Motor: Provides precise movement control
  • LEGO Hub: Central control unit receiving commands via Bluetooth Low Energy (BLE)
  • On-board Camera: Captures real-time video feed for AI processing
  • Portable Battery: Powers the entire duration of the mission

Edge Computing Brain

The heart of our edge system is the NVIDIA Jetson Orin, a powerful System on Chip (SoC) that combines:

Component Specification Purpose
CPU ARM Cortex-A78AE System control & coordination
GPU NVIDIA Ampere AI inference acceleration
Memory Up to 64GB LPDDR5 High-speed data processing
Storage NVMe SSD Model storage & caching

Edge Software Stack

┌─────────────────────────────────────┐
│         Red Hat Device Edge         │  ← Enterprise Edge OS
├─────────────────────────────────────┤
│             MicroShift              │  ← Lightweight Kubernetes
├─────────────────────────────────────┤
│          Microservices Edge         │  ← AI & Control Logic
└─────────────────────────────────────┘

Red Hat Device Edge provides:

  • Performance: Optimized for resource-constrained environments
  • Reliability: Production-ready edge computing platform
  • Security: Enterprise-grade security features
  • OTA Updates: Over-the-air (OTA) deployment capabilities

MicroShift enables:

  • Container Orchestration: Kubernetes platform at the edge
  • Service Management: Automated deployment and scaling
  • Self-healing: Automatic recovery from failures
  • Monitoring: Real-time system health tracking

☁️ In the Cloud: the OpenShift Cluster

Infrastructure Components

Located in the AWS Cloud with 5G connectivity to the edge:

┌─────────────────────────────────────┐
│          OpenShift Cluster          │
├─────────────────────────────────────┤
│ ┌───────────────┐ ┌───────────────┐ │
│ │   OpenShift   │ │     CI/CD     │ │
│ │       AI      │ │   Pipelines   │ │
│ └───────────────┘ └───────────────┘ │
│ ┌───────────────┐ ┌───────────────┐ │
│ │     Video     │ │     GitOps    │ │
│ │   Monitoring  │ │   Deployment  │ │
│ └───────────────┘ └───────────────┘ │
└─────────────────────────────────────┘

Cloud Services

OpenShift AI Platform 🤖

  • Data Science Projects: Isolated environments for ML development
  • Jupyter Notebooks: Interactive development experience
  • Pipeline Servers: Automated ML workflow execution
  • Model Serving: REST API endpoints for inference
  • GPU accelerators: High-performance training infrastructure

CI/CD Pipelines 🔄

  • Tekton Pipelines: Cloud-native CI/CD workflows
  • Container Registry: Secure image storage and distribution
  • Multi-architecture Builds: Support for x86_64 and ARM64
  • Automated Testing: Quality assessment at every step

Video Surveillance System 📹

  • Real-time Streaming: Live camera feed from the train
  • Kafka Brokers: High-throughput message streaming
  • Web Interface: Remote monitoring capabilities
  • Alert System: Instant notifications for anomalies

📊 Data Flow

graph TD A[Train Camera] -->|Video Stream| B[NVIDIA Jetson Orin] B -->|AI Inference| C[Traffic Sign Detection] C -->|Control Commands| D[LEGO Hub] B -->|5G Connection| E[OpenShift Cluster] E -->|Training Data| F[OpenShift AI] F -->|Updated Model| B E -->|Monitoring| G[Dashboard] E -->|GitOps| H[Deployment] style A fill:#e1f5fe style B fill:#f3e5f5 style C fill:#e8f5e8 style D fill:#fff3e0 style E fill:#e3f2fd style F fill:#f1f8e9 style G fill:#fce4ec style H fill:#fff8e1

Real-time Processing Workflow

  1. Image Capture: Camera captures traffic sign images
  2. AI Inference: Jetson processes images using trained model
  3. Decision Making: AI determines appropriate action (stop/go)
  4. Command Transmission: BLE commands sent to LEGO Hub
  5. Feedback Loop: Results sent to cloud for continuous learning

🏢 Multi-Architecture Support

Build Infrastructure

Our system supports heterogeneous computing environments:

Architecture Use Case Platform
x86_64 AI Development & Training OpenShift Cluster
ARM64 Edge Deployment NVIDIA Jetson Orin
Multi-arch Universal Images Container Registry

Deployment Strategy

  • Cloud Development: AI models trained on powerful x86_64 clusters
  • Cross-compilation: Applications built for ARM64
  • Edge Deployment: Lightweight containers deployed to Jetson
  • Continuous Integration: Automated testing across all architectures

🛡️ Security & Reliability

Edge Security

  • Secure Boot: Verified system startup
  • Container Security: Isolated execution environments
  • Certificate Management: Mutual TLS authentication
  • Network Isolation: Segmented communication channels

Cloud Security

  • RBAC: Role-based access control
  • Secrets Management: Encrypted credential storage
  • Audit Logging: Comprehensive activity tracking
  • Network Policies: Micro-segmentation

This architecture demonstrates how Red Hat’s Edge AI stack enables sophisticated AI applications in resource-constrained environments while maintaining enterprise-grade security, reliability, and scalability!