QERRA-v2 Classical brings deterministic ethics to robotics with 12-vector SEMEV-12 framework

Marussa Metocharaki, an independent researcher based in Greece, has developed QERRA-v2 Classical, a fully explainable ethical decision framework designed specifically for robotics and high-stakes autonomous systems. Unlike neural-network-based approaches that rely on opaque large language models, this system is completely deterministic and auditable. It operates on the SEMEV-12 framework, which defines 12 immutable, human-centred ethical vectors. The engine evaluates a given situation by activating relevant vectors and returns a traceable ethical score with full reasoning, activated vectors, and moral clarity signals. Currently, 9 out of 12 vectors are implemented and tested in version 1.8.5. The project is openly developed with full transparency about constraints — Marussa is a solo developer with limited resources, yet has delivered a public API on Hugging Face (free tier) and a ROS 2 bridge for real-world testing.

The framework is designed for practical integration into robotics stacks. The ROS 2 bridge allows developers to test the ethical layer on actual robots, and the architecture supports use as a Behavior Tree condition node for autonomous decision-making. The free public API on Hugging Face enables quick prototyping without local setup. The repository (GitHub: marunigno-ship-it/QERRA-v2-classical) is licensed under AGPL-3.0, and Marussa is actively seeking feedback from the ROS community on integration scenarios, the remaining three vectors, and real-world testing. This system addresses a critical gap: explainable, auditable ethical reasoning that can be deployed in autonomous vehicles, healthcare robots, and other high-stakes domains where black-box AI is unacceptable.