EXN [ExoSpaceLabs] – Advanced Satellite Telemetry, Recognition & AI

EXN is a modular open-source demonstration payload that combines a Raspberry Pi, STM32 (RTOS), and FPGA into an integrated Earth Observation system.
The goal is to showcase onboard image capture, CCSDS packetization, AI inference, and system-level coordination – all within a realistic space payload architecture.

---

🌌 Project Overview

EXN demonstrates how different embedded platforms can work together as a cohesive payload:

• Pi (Camera Node): Captures images, pre-packets or streams raw data.
• STM32 RTOS (Control Node): Supervises the system, handles CCSDS TeleCommands/TeleMetry, performs health checks, and routes control.
• FPGA (Processing Node): Performs preprocessing (resize/normalize), inference acceleration, and postprocessing (argmax).
• CCSDSPack: Provides CCSDS packet definitions and handling across nodes.

All interfaces (CCSDS, SPI/UART/CAN links, FPGA register maps) are defined in the central EXN interfaces specification, ensuring consistency across hardware.

---

📂 Repository Structure

This is the meta-repo that ties everything together.

• interfaces/ → Packet interfaces: CCSDSPack .cfg, JSON mirrors, and MCU header-only definitions
• docs/ → architecture docs, diagrams, and roadmaps
• system/ → integration scripts, manifests, CI/CD workflows
• submodules/:
  • exn-pi-cam
  • exn-mcu-rtos
  • exn-fpga-ai
  • ccsdspack

---

🧭 Modular System Graph

graph LR
  %% GitHub-friendly Mermaid: quoted subgraph labels and explicit directions
  subgraph "Ground Station"
    direction TB
    GSNode[GS App APID 0x0F0 SrcID 0x10]
  end

  subgraph "EXN Payload"
  
      subgraph "MCU RTOS"
        direction TB
        MCUNode[MCU Controller APID 0x100 SrcID 0x01]
      end
    
      subgraph "PI CAM"
        direction TB
        PiNode[Camera Node APID 0x101 SrcID 0x02]
      end
    
      subgraph "FPGA AI"
        direction TB
        FpgaNode[Processing Node APID 0x102 SrcID 0x03]
      end
      
  end

  GSNode --> MCUNode
  MCUNode --> GSNode

  MCUNode --> PiNode
  MCUNode --> FpgaNode

  PiNode --> MCUNode
  FpgaNode --> MCUNode

  PiNode --> FpgaNode

Loading

Notes:

• GS may request individual HK (Service 3/1) via MCU and may request a System HK aggregation (Service 3/10). MCU bridges/proxies; devices do not depend on GS.
• All APIDs are fixed: GS=0x0F0, MCU=0x100, PI=0x101, FPGA=0x102. Source IDs: MCU=0x01, PI=0x02, FPGA=0x03, GS=0x10.

➕ Additional Diagrams

The sequence diagrams have been moved to a dedicated page for clarity. See:

• docs/diagrams.md — Mermaid sequences for System HK aggregation and the GS-driven end-to-end flow.

---

📑 Interfaces & ICD

• Master ICD: docs/ICD.md
• Device/Service ICDs:
  • GS (Ground Station Simulator): docs/icd/gs.md
  • MCU-RTOS: docs/icd/mcu-rtos.md
  • PI-CAM (Raspberry Pi 5 + Camera Module 3): docs/icd/pi-cam.md
  • FPGA-AI: docs/icd/fpga-ai.md

🧩 Interfaces

• CCSDSPack interfaces (.cfg):
  • TeleCommands: interfaces/ccsdspack/tc/
  • TeleMetry: interfaces/ccsdspack/tm/
• JSON mirrors (for tooling):
  • TeleCommands: interfaces/json/tc/
  • TeleMetry: interfaces/json/tm/
• MCU-RTOS (no filesystem):
  • Header-only definitions: interfaces/mcu-rtos/

See also: docs/diagrams.md for the sequence diagrams moved from this README.

---

🛠 Roadmap

1. Phase 1 – Pi Camera
   • Define ACK / NACK / TM and packets structures.
   • Bring up Pi camera, capture single frames
   • Export to CCSDS packets with CCSDSPack
2. Phase 2 – STM32 RTOS
   • Implement TC/TM service handler
   • Add health monitoring + basic CAN/SPI link
3. Phase 3 – FPGA Integration
   • AXI register map + resize block
   • Add inference model (ImageNet classifier)
4. Phase 4 – Full System Integration
   • Pi → STM → FPGA workflow
   • End-to-end CCSDS packet chain
   • Demo: classification result sent as telemetry

---

📜 License

Apache 2.0 (open source, research & demonstration use).