Non-Bypassable Execution Control in Autonomous Systems: Design Considerations for Runtime Safety Enforcement

This paper complements the architectural principles of non-bypassable execution control by providing design and implementation considerations. It is part of the execution control layer within a broader governance architecture for autonomous robotic systems. Related works: - Non-Bypassable Execution Control in Autonomous Systems: Architectural Principles for Runtime Safety Enforcement - A Governance Architecture for Safe and Bounded Autonomous Robotics Systems - Safety-Bounded Autonomy: Architectural Safety Enforcement for Distributed Robotic Systems This paper addresses a fundamental challenge in modern autonomous systems: how to ensure safe execution in the presence of adaptive, learning-based, and increasingly unpredictable components. Traditional safety approaches rely heavily on design-time validation and the assumption that system behavior can be fully anticipated. As autonomy increases, this assumption becomes progressively less reliable. This work introduces the concept of non-bypassable execution control as an architectural perspective for runtime safety enforcement. Instead of relying solely on the correctness of planning and control logic, the approach emphasizes the role of execution mediation as a mechanism to constrain system behavior at the point of action. The paper outlines key architectural considerations, including the separation of functional concerns, execution mediation, and runtime constraint validation. It frames execution control as an enforcement boundary between decision-making and actuation, enabling systems to maintain operational safety even under uncertain or evolving conditions. This work does not prescribe a specific technical implementation. Rather, it defines a conceptual framework that can be realized across different system architectures and application domains. The contribution is intended to support ongoing research and development in robotics, AI safety, and cyber-physical systems, particularly in contexts where robust, real-world operation is required. This work is part of a broader research program on governance and execution control in autonomous robotic systems, including capability governance, safety-bounded autonomy, and non-bypassable execution control architectures.