Safe-ROS: An Architecture for Autonomous Robots in Safety-Critical Domains

Deploying autonomous robots in safety-critical domains requires architectures that ensure both operational effectiveness and safety compliance.In this paper,we present the Safe-ROS architecture,designed to support the development of reliable and verifiable autonomous robots for such environments.Safe-ROS consists of two distinct subsystems:(1)Intelligent Control System(ICS)–responsible for normal and routine operations,and(2)a Safety System composed of formally verifiable Safety Instrumented Functions(SIFs)that provide independent safety oversight.We demonstrate the Safe-ROS framework on an AgileX Scout Mini robot performing autonomous inspection in a nuclear environment.One safety requirement is instantiated as a SIF:a cognitive agent programmed to stop the robot whenever it detects proximity to an obstacle.The SIF is formally verified to meet this safety requirement,and its integration within the overall autonomous inspection system is also verified.Validation is conducted in Gazebo simulation and laboratory testing,with evaluation performed in the context of the UK nuclear sector,where safety assurance and regulatory compliance are essential.Success criteria include the formalization of safety properties,implementation and verification of SIFs,and integration of the safety mechanism into the operational system.The results demonstrate that Safe-ROS enables safety-verifiable oversight for autonomous robotic systems in safety-critical domains,providing a robust,extendable framework applicable to additional requirements and diverse applications.