Skip to content

architecture_notes.md

Latest commit

 

History

History
50 lines (39 loc) · 1.67 KB

architecture_notes.md

File metadata and controls

50 lines (39 loc) · 1.67 KB

Camera Control Architecture Notes

MCP vs Connector Separation

Current Implementation

  • MCP Server: /Users/j/Code/mcp/camera_mcp_server.py
    • Complete PTZ control implementation
    • Hardware-specific fixes (cameracaptured process handling)
    • Input validation and safety controls
    • Direct integration with webcam-ptz executable and imagesnap
    • JSON-RPC interface for local control

Proposed Connector Architecture

[Multiple Cameras] → [Camera MCP Servers] → [Protocol Connectors] → [athena/External Systems]
     USB/WiFi            Local Control         MQTT/REST/WS         Integration Layer

Benefits of Separation

  1. Reusability: Single MCP implementation works for all camera types
  2. Protocol Flexibility: MQTT, REST, WebSocket connectors can all use same MCP
  3. Hardware Abstraction: MCP handles device-specific quirks once
  4. Scalability: Easy to add more cameras without duplicating PTZ logic
  5. Maintainability: PTZ bugs fixed once in MCP, benefits all connectors

Implementation Strategy

Phase 1: MCP Server (✅ Complete)

  • PTZ control via webcam-ptz
  • Camera discovery
  • Screenshot capture
  • Input validation

Phase 2: MQTT Connector (Next)

  • Lightweight bridge between MQTT and Camera MCP
  • Protocol translation only
  • No PTZ logic duplication

Phase 3: athena Integration

  • Use MCP directly OR via MQTT bridge
  • Focus on AI vision pipeline, not camera control

Tools Available via MCP

  • list_cameras: Discover connected cameras
  • take_screenshot: Capture with base64 encoding
  • ptz_control: Pan/tilt/zoom commands with validation
  • get_camera_status: System status and health checks

Date: 2025-06-07