Hi everyone,
I’m Marussa Metocharaki, working solo on QERRA-v2 Classical — a 100% classical and fully explainable ethical evaluation engine based on the SEMEV-12 framework (12 immutable human-centred vectors).
The long-term goal is to create a lightweight, auditable ethical safety layer that can run as a Condition node in Behavior Trees for robots, especially humanoid systems that interact with humans.
Current status:
-
Live public API on Hugging Face
-
Stable v1.8.1 core with improved nuance handling (moral pressure + strong commitment cases)
-
ROS2 bridge stub included (works standalone, becomes full node with rclpy)
-
All regression tests passing
I am now looking for a small number of serious early testers, especially from the ROS 2 and robotics community.
What would be most helpful:
-
Robotics scenarios with conflicting human instructions or high-stakes decisions
-
Medical, caregiving, or industrial ethical dilemmas
-
Any prompts where you think the score or reasoning feels wrong
Live API (free, demo key available): https://qerra-v2-api-classical-qerra-v2-api-classical.hf.space/docs
Full Call for Testers + details: https://github.com/marunigno-ship-it/QERRA-v2-classical/blob/main/CALL_FOR_TESTERS.md
GitHub repo: https://github.com/marunigno-ship-it/QERRA-v2-classical
If you’re interested, please reply here or email me at marunigno@gmail.com with subject “QERRA Tester” and a short note about your background.
All honest feedback is extremely valuable at this stage.
Thank you!
Best regards,
Marussa Metocharaki