Hi everyone,I’m Marussa marunigno-ship-it (MARUSSA METOCHARAKI) · GitHub
, a solo self-taught founder based in Greece.
I’m building QERRA-v2 — an open-source hybrid quantum-classical ethical decision engine that acts as a safety layer for high-level robot deliberation (the “mission control / planning” part of autonomous robots).
Main mission / goal:
To help robots (especially humanoids) make decisions that are not only efficient but also ethically safe and transparent. The system checks inputs for toxicity and manipulation, evaluates proposals against 12 ethical dimensions (SEMEV-12 vectors), applies region-aware safety rules (EU / USA / UAE), and generates audit trails.
The long-term vision includes a quantum-inspired W-state layer for richer exploration of decision possibilities, combined with post-quantum cryptography for secure access.
Recent improvements (April 2026):
-
Significantly polished the classical /v1/analyze endpoint — now more stable, with better toxicity + deception detection (using Detoxify) and safety kernel logic.
-
Live public API is running and documented (demo key available).
-
Updated documentation: README, ARCHITECTURE.md, WHITEPAPER.md, API-DEMO.md, and QUANTUM-STATUS.md (including the real 8-qubit W-state run on IBM hardware from January).
-
Added a basic ROS2 stub as a starting point for integration with deliberation frameworks.
-
The quantum/hybrid part is still simulated due to resource limits, but the classical safety layer is fully functional and open for testing.
The project is early-stage / experimental (AGPL-3.0), fully on GitHub:
https://github.com/marunigno-ship-it/QERRA-v2
I’m mainly here in the Deliberation group to listen and learn how deliberation tools work in practice, especially around safety, visualization, and ROS 2 integration.
Any feedback on the ethical safety approach or ideas for connecting it to existing deliberation frameworks (Behavior Trees, planners, etc.) would be very welcome.
Happy to answer questions or share the public API demo!Best,
Marussa