Project summary

Transparency Note: English is not my native language and I use AI assistants as co-pilots for coding and drafting. Every commit, Webots simulation run, and 140+ unit tests in the repository are built and verified by me. You should first check the repository not the AI assistant usage!

QERRA-v2 Classical is an open-source, deterministic safety gate that prevents AI planners from directly controlling mobile robot wheels. It evaluates sensor data and human distress on a strict 32 ms loop, instantly clamping motor speeds to zero if an obstacle appears, the planner lags, or a hazard is detected. The core architecture is already built, fully unit-tested, and verified in Webots simulation on a TIAGo robot with 100% deterministic replay.

What are this project's goals? How will you achieve them?

First is to make mobile robots safe and predictable around people in shared spaces like hospitals, small / mid businesses, shops and homes. The long-term vision is connecting all three layers: physical stopping reflexes at the wheels (Layer 1), instruction screening (Layer 2), and social etiquette/ ranker (Layer 3).
The Achievement Part: Right now I am working with zero budget. This grant gives me basic living runway, a reliable laptop, and paid cloud hosting so I can work on this full-time without bottlenecks. With that stability, I will turn the simulation code into clean, native ROS 2 packages, open the architecture to external peer review, and use this open-source baseline to find a technical co-founder or robotics lab to test QERRA on physical hardware.

How will this funding be used ?

1). Rent and basic living in Greece : Approximately: 1.800 a month.
Covers a modest apartment, bills, and food so I can move into my own space and work on this full-time with zero distractions.

2). A New Laptop : A reliable machine that can handle Webots simulations, ROS 2 builds, and local models without freezing or crashing. Approximately: 2.500

3). Test Hardware : Buying a Raspberry Pi / Jetson board and a basic LiDAR sensor so I can test this on physical hardware instead of just simulation. Approximately : 2.500 a month.

4). Moving Deposit & Setup Buffer : Rental deposits to move out, plus basic workstation setup. Approximately: 3.600

5). Cloud Hosting, AI Tools & Subscriptions : Covers a paid cloud server to eliminate Hugging Face cold starts, plus the daily AI coding assistants I rely on as co-pilots (Claude, ChatGPT, GitHub) and essential dev subscriptions.

Who is on your team? What's your track record on similar projects?

I am a self-taught solo builder based in Athens, Greece. I have no lab backing or team; I work independently using AI tools as my co-pilots. I don't have past robotics projects to show, QERRA is my track record.
I’ve pushed 670+ commits, 140+ passing tests, and verified sub-32 ms motor clamping with bit-for-bit deterministic replay in Webots. Everything I do is public on GitHub, open-source, and honestly documented.

What are the most likely causes and outcomes if this project fails?

CAUSES: The main risk is running out of survival runway before finding the right collaborators or a technical co-founder. Working in isolation without lab access makes the jump from simulation to physical robot hardware the hardest bottleneck.
OUTCOMES: Even if I fail to get this onto physical robots, the work is not lost. The entire architecture, 140+ tests, simulation worlds, and documentation remain 100% open-source (Apache-2.0) on GitHub for any robotics engineer or safety researcher to build on.

How much money have you raised in the last 12 months, and from where?

I haven't raised any kind of money.