Solo researcher · founder of QERRA-v2 Classical · Greece
I am building an open-source, 100% classical, fully explainable ethical evaluation engine designed as a transparent safety layer for humanoid robots and high-stakes AI systems.
Built on the SEMEV-12 framework — 12 immutable, human-centred ethical vectors.
Every result is fully traceable: exact vectors activated, per-vector scores, and human-readable reasoning. No neural networks. No black boxes.
Current state: v1.8.8 — Stable Classical Research Engine. All 12 SEMEV-12 vectors active and verified. Live public API. Working ROS 2 bridge. Key features:
- Hybrid detection (semantic similarity + regular expression patterns)
- Weighted multi-vector scoring with nuance handling
- ROS 2 integration for Behaviour Trees and real robotics pipelines
Repositories:
- QERRA-v2-classical — active classical edition
- QERRA-v2 — original hybrid research project
- Prior Art (Zenodo): https://doi.org/10.5281/zenodo.20356394
Looking for researchers, engineers, and organizations in AI safety, ethical AI, and humanoid robotics who want to test the system, give feedback, or explore integration.
Contact: marunigno@gmail.com
If this vision resonates with you, consider starring the repositories or supporting via GitHub Sponsors.
This project exists because ethical reasoning should be a foundational, auditable layer in every system that acts alongside humans — not an afterthought.
QERRA-v2 Classical is an experimental research tool. It has not been independently validated by any academic institution or robotics lab. Commercial licenses are available for evaluation and non-safety-critical use.
Marussa Metocharaki · Greece · 2026