# RoboSolar Cleaner

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [John Ugbor](https://cerebralvalley.ai/u/JohnLarry)
- **GitHub:** https://github.com/johnugbor/RoboSolar
- **Demo video:** https://youtu.be/MCR5A5pzwpI
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4267

Solar panels lose a significant amount of power when dust, sand, bird droppings and other dirt build up on their surface, a dirty panel can lose 5–30% of its energy output, and the loss keeps growing until someone cleans it. Today that cleaning is done manually: it is slow, labor-intensive, wastes water, worse still operators usually clean on a fixed schedule thereby leaving dirty panels losing money for weeks because nobody knew they were dirty.


RoboSolar Cleaner is an autonomous solar-panel cleaning robot that cleans only when it is needed, and only the panels that need it. Five custom-designed circuit boards, designed with muse-wonder-1.0 specifically for Kicad work together as one system:

1. A Soiling Station compares a clean reference mini-panel against an exposed dirty one to measure exactly how much power soiling is costing.
2. A Panel Node on each solar array measures real voltage and current, confirming the loss.
3. A Weather Node checks rain and wind so the robot never cleans in unsafe conditions.
4. Meta's Muse Wonder (muse-wonder-1.0) receives this sensor data and recommends which panels to clean and in what order
5. The Robot then drives alongside the panels, extends a brush arm onto the panel surface, cleans it, returns to its Dock to recharge and refill its water tank, and verifies the panel's power has recovered.

## More from Meta Global AI Developer Hackathon

- [FollowCut](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4264)
- [Gameon Finance](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4265)
- [RobotRanger](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4266)
- [Team Abdul](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4268)
- [UNBIDDEN](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4269)
- [deloken](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4270)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4267. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt