Skip to main content

Week 19 - Invention, Intellectual Property, And Income

1. IP and Income Plan

Licenses:

  • Firmware/application code → Apache 2.0

  • Hardware (PCB design, 3D models, laser-cut files, fabrication notes) → CERN OHL

  • Income paths (not a course goal, only as alternatives):

  • Teaching kits: Provide simplified AI Horse BOOK kits for schools, priced to cover costs only

2. Dissemination Plan

Distribution & Audience Strategy

AudienceMethod
Fab classmates/reviewersGitLab site + weekly pages + presentation.png/mp4
Maker communityGitHub mirror + LICENSE files + build thread
Schools/educatorsSTEAM kits (only upon request)

Timeline:

  • June 2026: Week 20 demo, images, video, site launch

  • July 2026: Repository cleanup, redact secrets, tag release, add licenses

  • August 2026: Public GitHub mirror

  • September 2026: Draft short article/blog post

  • H2 2026: Optional offline demo

3. Future Directions

DirectionDescription
LLM UpgradeSwitch to Gemini model for better response efficiency and intelligence
Graphical ProgrammingProvide more STEAM graphical programming features to lower the educational barrier

4. Done vs In Progress

✅ Done (subsystems independently verified):

  • Horse status → OLED display

  • Bluetooth broadcast → Horse receives

  • MQTT commands → Horse receives

  • Motion control (remote controller)

🔄 In Progress:

  • Mechanical + electrical full integration (Horse + BOOK frame) — key to passing Week 20

5. Pre-Week 20 Calendar

DateTask
June 5–6Mechanical assembly, remote controller assembly
June 7Power‑on debug, OLED display, AI voice dialogue assembly
June 8Freeze demo script, complete full rehearsal
June 9–10Shoot integrated photos/video, export presentation.png/mp4
~June 12Week 20 presentation

6. Key Takeaways

IP week forced me to distinguish between "what I will actually publish" and "what is just speculation"

  • Apache 2.0 + CERN OHL aligns with the existing Fab repository

  • Income planning was an honesty check, removing unconvincing crowdfunding/subscription ideas

  • Each subsystem works on the bench, but full integration remains the biggest risk

  • Every spoken output must trace back to sensor data or an "unknown" label