Enhancing Safety and reliability in nuclear cleanup robots

AI Safety Cages Ensure Confidence in Autonomous Nuclear Cleanup Robots

d-risq - modelworks image

In response, D-RisQ embarked on a ground-breaking project, collaborating with Vaarst® to harness AI in the form of Simultaneous Localization and Mapping (SLAM). The project centred around Vaarst's SubSLAM underwater camera system, renowned for its machine learning-based capabilities demonstrated in hazardous environments, including offshore industries. The overarching goal was to devise a strategy for ensuring the safety of these autonomous robotic systems.

The heart of the solution lay in D-RisQ's development of the Last Response Engine (LRE), aptly nicknamed the "AI Cage." This innovative safety monitoring software tackles two primary safety hazards: collision avoidance and the prevention of propeller-induced splashes. When predefined safety thresholds are breached, the LRE seamlessly takes control. What sets the LRE apart is its adherence to industry-recognized standards, making safety-related software development more accessible and cost-effective.

Kapture®, D-RisQ's tool, facilitated the process of distilling user requirements into verifiable system requirements, thereby establishing a solid foundation for software implementation. Subsequently, D-RisQ ModelWorks® enabled an exhaustive verification of design elements in Simulink®/Stateflow®, ensuring alignment with established requirements.

The process culminated in the automatic coding of the Simulink® model using dSPACE's TargetLink® autocoder. The resulting source code underwent further validation using another tool in the D-RisQ Toolsuite, CLawZ®, a formal methods-based tool. This robust approach enhances both rigour and efficiency, yielding significant cost savings compared to conventional methods.

The project, assisted greatly by D-RisQ’s Last Response Engine software, has ushered in a new era of safety and reliability for AI-integrated robotics. By effectively mitigating collision risks and propeller-induced splashes, this safety cage for AI ensures autonomous systems' operation at the highest levels of assurance. The amalgamation of automated formal analysis tools not only enhances precision but also accelerates the process, revolutionizing safety practices in critical domains.

The success of this endeavour transcends the nuclear decommissioning sector, showcasing potential applications in various domains, including maritime and aviation industries. As demonstrated by Mr. Nick Tudor of D-RisQ Ltd. at the Embedded World Conference in June 2022, this innovative approach marks a pivotal step towards harnessing the full potential of AI in safety-critical contexts.

Need help with control system software? Contact us today to find out more

Get in touch
D-Risq - Logo