# Quberon

- **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:** [Rasmus Lundqvist](https://cerebralvalley.ai/u/Quberon)
- **GitHub:** https://github.com/quberon/quberon-core
- **Demo video:** https://youtu.be/ov0RTx9BHWE
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4208

Quberon is a prototype robot identity and human-intervention interface. A Raspberry Pi and e-ink QR display provide the physical entry point. The web application lets authenticated users inspect a robot's state and request a software STOP, with separate viewer, operator and admin permissions.

The robot is a software state-machine simulator, not a physics simulation. The application records STOP requests, retains the latch across normal restarts, and blocks simulated tasks while stopped in the single-process demo workflow. An admin reset clears the simulation latch without starting a task. The prototype has known retry and error-handling limitations. There is no physical humanoid control or safety-rated emergency-stop function in this version.

I built the application with a Hermes coding agent using Meta Muse Wonder 1.0. The code also contains an advisory planner with a labelled mock provider and a configurable live-model adapter. A successful live planning call is not yet demonstrated in the repository's evidence. Proposed actions pass through a local allow-list; the planner cannot reset the latch or command hardware.

This is an early step toward a common intervention interface across robot manufacturers. The current delivery is the Pi interface, authenticated web application and simulator. Gesture recognition, voice-triggered STOP and Isaac Sim integration are next steps, not implemented capabilities in the reviewed version. Phone access still uses a private certificate authority, so frictionless public onboarding remains unfinished.

## More from Meta Global AI Developer Hackathon

- [SightSeen](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4205)
- [Ronda_By_Ciro](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4206)
- [paneller](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4207)
- [Anthony Dawson (solo)](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4209)
- [Peer](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4210)
- [StackForge](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4211)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4208. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt