QERRA-v2 Classical adds ethical and physical safety layers to ROS 2 robots

QERRA-v2 Classical is an open-source, fully explainable safety layer that evaluates autonomous robot actions before they execute. Built for ROS 2 and Behavior Trees, it uses a two-layer architecture to combine physical and moral guardrails. The first layer, QERRA-HSR, is a pure Python, zero-machine-learning module that monitors physical hazard proximity and human distress signals, triggering immediate fail-closed safety halts when conditions are met. The second layer, SEMEV-12, is a hybrid semantic-pattern engine that assesses prospective natural-language tasks against 12 human-centered ethical dimensions—including coercion, gaslighting, and autonomy violations—while producing complete reasoning logs for auditability.

The middleware integrates directly into ROS 2 pipelines via Action Servers and PyTrees Behavior Tree condition nodes, with non-blocking background threading ensuring control loops never stall. The repository on GitHub includes a simulation prototype, unit benchmarks, and 50 Hz telemetry test suites, signaling production-ready maturity. The developer is now seeking technical feedback, code reviews, and collaboration on simulation or physical hardware trials. For robot developers, QERRA-v2 offers a deterministic, auditable way to enforce both physical safety and ethical constraints in autonomous systems, addressing growing regulatory and public scrutiny around AI and robotics.