runtime_integrity bridges ROS 2 cmd_vel audit gap for EU AI Act compliance

ZC_Liu has released runtime_integrity, a ROS 2 package that refactors the earlier ros2_kinematic_guard into a broader framework for execution-integrity evidence targeting autonomous mobile robots using Nav2. The core engineering problem remains concrete: /cmd_vel commands are published, but the robot's physical motion diverges silently. With the EU AI Act's logging and human-oversight obligations approaching for high-risk AI systems (which include AMRs in certain applications), runtime evidence becomes critical. The package does not claim to make a robot fully compliant, nor does it replace safety-rated PLCs or hardware E-stops. Instead, it focuses on providing machine-readable logs that feed into engineering diagnostics, incident reconstruction, operator dashboards, audit trails, and human-oversight workflows.

The integration model is deliberately non-invasive: runtime_integrity sits between the controller server, planner, or teleop node and the base driver. It can start in observe mode, meaning no changes are required to existing Nav2 controllers, behavior trees, or base-driver code. The package now emphasizes runtime accountability and provides hooks for enterprise audit event schemas and HMI integration. For the ROS 2 community, this addresses a persistent accountability gap—where does the answer to "why did the robot do that?" come from? runtime_integrity offers a transparent, evidence-based layer that bridges the software navigation stack and physical execution, enabling better diagnostic and compliance processes.