With the spread of coronavirus leading to many people working remotely, Automotive IQ has ramped up its free online event schedule. Our first is Autonomous Vehicles Online, a two-day lineup of speakers who will discuss just how far the self-driving vehicle industry has come, and highlight where it is going.
On May 26, our speakers will talk more generally about the state of autonomy, along with sessions from specially selected sponsors; while one May 27, we’ll get into more detail, diving in to some of the rules and regulations that govern self-driving vehicles.
Meet the speaker
He is a member of the Electronics, Controls and Software Process, Methods and Tools group at GM R&D, where he currently leads research in the area of vehicle level safety analyses including systems of systems and feature interactions.
We caught up with Rami ahead of his talk at Autonomous Vehicles Online to ask a few questions:
Automotive IQ The ISO standards that apply to autonomous vehicles effectively regulate the unfamiliar. Can you give us an idea of some of the challenges you have faced in agreeing what needs to be included in the standards?
Rami Debouk Developing a standard is not an easy task, let alone coming to a consensus or at least an agreement on the technical topics to be included.
In my opinion, this is not fully dependent on the technical topic. However, the fact that safety is the topic of our standards brings another dimension to the challenges faced.
We have worked diligently on developing these standards by providing a scientific and rigorous approach to address the issues with support from field data, best practices and many lessons learned.
AIQ How has the process of agreeing the size and scope of ISO 26262 and ISO 21448 differed?
RD There is a clear distinction between the scopes of the two standards. Even though their applicability is part of applying system safety analyses to road vehicles, ISO 26262 addresses hazards resulting from malfunctioning behaviors, while ISO 21448 analyses hazards caused by functional insufficiencies that is non-faulted behavior due to insufficiency of specification or performance limitation.
AIQ What are the main challenges in deciding how safe a self-driving vehicle needs to be, and how is that likely to be understood by the end consumer?
RD Safety is defined in the standards as absence of unreasonable risk, therefore arguments supported by evidence to demonstrate that, in addition to properly executing your processes, are required as part of your safety case.
Usually a safety case is not provided to the end consumer. However, a high-level summary of the safety methodology can be provided to the public, similar to the Voluntary System Safety Assessment (VSSA) recommended by the US Department of Transportation in the case of the development of autonomous vehicles.
AIQ Your session in Automotive IQ’s Autonomous Vehicles Online event will answer the question ‘What are the key trends and challenges in autonomous vehicle safety?’ Could you share some of the main themes of your talk?
RD In my opinion, ISO 26262 Second Edition requires further improvement to fully address the safety analysis for autonomous vehicles.
In my talk, I will discuss challenges in areas such as the hazard analysis and risk assessment, fail-operational behavior, operational safety, cybersecurity and verification and validation. ISO 21448 is trying to address some of these topics, namely the verification and validation of unknown hazardous behaviors.
To learn more about the full line-up of speakers, the confirmed agenda, and to register for the event, visit the dedicated Autonomous Vehicles Online site.