SRES Fireside Chat: How Can Robot Safety Keep Up with AI?
What happens when a robot makes an unsafe decision even though its hardware and software are working as designed? How do we demonstrate that perception is safe enough to protect a person? And how do we emergency-stop a humanoid that needs active control to stay balanced?
In this Fireside Chat, SRES co-founders Jody Nelson and Gokul Krithivasan are joined by Marcus Rau, Head of the Certification Body for Functional Safety and Cybersecurity at SGS-TÜV Saar, to explore safety for industrial, mobile, and humanoid robots.
The discussion examines how established functional safety standards such as IEC 61508 and ISO 13849 connect with the developing ISO/IEC TS 22440 series and emerging robotics standards ISO 26058-1 and ISO 25785-1. They also discuss sensing standards, including IEC TS 62998-1, and how principles from ISO 21448 (SOTIF) can help teams consider unsafe behaviors that may occur without a hardware or software fault.
Together, they explore the role of system-level hazard and risk analysis, the challenges of demonstrating confidence in perception and machine learning, and how different standards can support a broader safety argument.
Drawing on perspectives from safety engineering, assessment, and certification, this discussion is relevant to teams developing, integrating, and deploying robotic systems.
Watch the full discussion below:
Selected Audience Q&A
The following questions were raised by attendees during the live session. Responses have been edited and condensed for clarity.
Can scenario-based safety analysis replace sensor-specific requirements? Could automotive sensors be used in robotics safety?
Marcus Rau: Existing sensors with high integrity may be useful in robotics applications, but they need to be evaluated for the new application and its operational design domain. Changes in operating conditions may require additional qualification.
Jody Nelson: The forthcoming third edition of IEC 61508 is expected to provide a path for reusing components developed to ISO 26262, such as automotive LiDAR. SOTIF principles offer a broader way to explore functional insufficiencies and triggering conditions, but teams need to establish their own acceptance criteria. Standards such as IEC 61496 and IEC TS 62998 provide more prescribed testing and thresholds within their respective scopes.
Gokul Krithivasan: SOTIF analyses, simulation, and scenario-based testing can help teams address known hazardous behaviors and discover additional edge cases at the system or robot level. They don’t replace the need to select sensors with appropriate performance for the operating concept, including range, field of view, and accuracy. Starting with system-level safety needs can also reveal where additional sensor modalities are needed.
How can safety keep pace with robots that receive continuous hardware and software updates?
Jody Nelson: A strong configuration management system is essential. Teams need to keep track of model versions, software versions, calibrations, and training data. More frequent updates may involve shorter test releases, but there still needs to be rigor around release decisions.
Gokul Krithivasan: Teams can establish risk acceptance criteria based on identified hazardous behaviors and use safety performance indicators to inform release decisions. Robust field monitoring and change management are also important, including attention to both leading and lagging indicators. A safety management system helps structure ongoing risk monitoring, response, and acceptance for the operational context.
How can a team be confident it has considered the right standards when preparing for a safety audit?
Jody Nelson: One option is to structure the safety argument using Goal Structuring Notation (GSN), an approach commonly used in autonomous vehicle development. Teams can also involve a third-party assessor for an independent perspective.
Gokul Krithivasan: Start with safety objectives and principles, then break them down into tangible safety claims. From there, explore which standards can serve as the right tools to support those claims. For a starting point, see our practical guide to safety standards for humanoid and general-purpose robots.
Meet the Speakers
Jody Nelson, Co-Founder & Managing Partner, SRES
Jody is a veteran safety leader with more than 24 years of experience advancing safety-critical automotive and autonomous systems for OEMs, suppliers, semiconductor companies, and autonomy developers. A former global assessment and certification lead at UL Solutions and co-founder of kVA and SRES, he combines hands-on development experience with consulting and independent assessment to deliver rigorous, defensible safety outcomes.
Gokul Krithivasan, Co-Founder & Managing Partner, SRES
Gokul is a senior safety expert with more than 14 years of experience in automated driving systems and safety assurance for complex interconnected products. He has supported technology companies, OEMs, and suppliers; led expert teams across safety, cybersecurity, and AI/ML; and brings experience spanning autonomous robotaxis, autonomous logistics, and robotics. He is a certified expert in ISO 26262, ISO 21448, and ISO 8800.
Marcus Rau, Head of the Certification Body for Functional Safety and Cybersecurity, SGS-TÜV Saar
Marcus is a functional safety assessor and technical certifier with experience across multiple industries. He founded the Global Competence Center for Functional Safety and Cybersecurity at SGS-TÜV and has led its certification body since 2025. He remains active as a lead assessor and supports the development of the organization’s work in humanoid robotics. Marcus has trained more than 10,000 people worldwide in the application of functional safety standards.
About the Series
The idea for a series of Fireside Chats traces back to the very early days of Jody and fellow SRES Partner Bill Taylor’s work together. In one of their first training sessions, a tricky FMEDA discussion left participants frustrated. That evening, the two of them stayed up late sketching a new way to explain it — an approach that ultimately found its way into the second edition of ISO 26262.
That moment sparked a tradition of technical discussions that has continued for years, now with the SRES team. These sessions have become a cornerstone of how we work — exploring difficult questions, challenging assumptions, and pushing toward better answers. With these Fireside Chats, we open that tradition to the broader safety community. Each session brings the same candid, technical conversations that shape our work to a public forum.
Explore Past Fireside Chats
Related Services
If you’re actively working on the topics discussed in this Fireside Chat, the following SRES services may be helpful.
SRES supports organizations developing and deploying AI-enabled physical systems, including industrial, mobile, and humanoid robots. Our experts help teams address functional safety, AI safety, validation, and safety assurance throughout development and deployment.
Our robotics safety services help teams identify applicable standards, develop safety architectures and safety cases, and address challenges such as perception limitations, controlled shutdown, and scenario-based validation.
SRES offers practical training for teams working on Physical AI and other safety-critical systems, covering IEC 61508, ISO/IEC TS 22440, safety-critical software, cybersecurity, and Responsible AI.
