Hello everyone,
My name is Marussa Metocharaki. I’m an independent researcher from Greece working on QERRA-v2 Classical, a fully explainable ethical evaluation engine designed for robotics and autonomous systems.
The system uses the SEMEV-12 framework (12 ethical vectors) and returns a transparent score with reasoning. It is completely classical — no neural networks involved.
I recently released v1.8.8, where all 12 vectors are now active. The public API on Hugging Face has also been updated to this version.
I’ve written a short technical brief for the robotics community here:
QERRA_FOR_ROBOTICS.md
I would really appreciate some guidance from people with more experience in ROS 2 and Behavior Trees.
My main question is: How would you recommend integrating something like QERRA as a Condition node in a Behavior Tree?
Are there any common patterns, best practices, or things to watch out for when using an external ethical check before executing an action?
I’m especially interested in real-world testing ideas or examples from people who have worked with decision-making nodes in ROS 2.
Any feedback or suggestions would be very helpful.
I’m still learning and would really value input from the community.
hank you in advance!
Best regards,
Marussa Metocharaki