Building Trust for Self Driving Vehicles
Edge Case Research is working alongside the insurance industry to quantify risk
The “drivers of tomorrow” will likely be a dramatic change from what we experience on our roads today. Replacing hundreds of thousands of human drivers, the driver of the future may be hundreds of millions of lines of software code. In the future, software “drivers” will transport us to work, carry our families and friends to see each other, and safely deliver the goods we count on every day. In preparation for this innovation, Edge Case Research is working with Liberty Mutual Insurance to help ensure that everyone stepping into a self-driving vehicle gets a safe ride and that every self-driving vehicle (SDV) traveling through our neighborhoods is built safely from the ground up.
Autonomous mobility relies on cutting edge artificial intelligence to drive safely. However, when will the artificially intelligent “drivers” of the future be safe for deployment? Edge Case Research and Liberty Mutual helped answer this question in April 2020 with the release of ANSI/UL 4600, Standard for Safety for the Evaluation of Autonomous Products, the world’s first standard specifically addressing autonomous products. Edge Case Research, Liberty Mutual, and partners including full-stack autonomy developers, autonomous and traditional vehicle manufacturers, rideshare operators, the U.S. Department of Transportation, and other government agencies, teamed up to draft the standard. With the official publication of ANSI/UL 4600, the industry now has a comprehensive framework for autonomy risk management. Built around a system-level safety case approach, ANSI/UL 4600 includes defined metrics, engineering processes, and assessment procedures, and it incorporates critical concepts from existing standards including ISO 26262 and SOTIF (ISO/PAS 21448). As the autonomous industry progresses, we will continue to improve ANSI/UL 4600, with revisions already underway.
A focused approach to risk management will help SDV developers bring their products to market faster and with a higher degree of safety assurance. Traditional SDV developer risk management solutions rely heavily on expensive and time-consuming road testing or on simulations. While on-road testing and simulation are undoubtedly necessary, savvy developers know they need more robust methods; the real world is simply too complex and unpredictable to be modeled perfectly, even in the most accurate simulations. Taking a safety case approach and utilizing safety metrics outlined in ANSI/UL 4600 provides a roadmap developers can follow to minimize risk and to compensate for simulation and on-road testing shortfalls.
When the crucial decision that self-driving vehicles are ready for deployment is made, lagging metric data will likely be inadequate to verify with sufficient confidence that the technology is safe. Nonetheless, companies have an obligation to make responsible deployment decisions. The ANSI/UL 4600 approach suggests a practical amount of testing combined with engineering rigor, safety culture, and a strong commitment to utilizing post-deployment field monitoring to continuously increase product safety. This can enable faster deployment than traditional safety validation methods, by reducing the need for confidence created through on-road testing and simulation, in exchange for a more comprehensive case-based validation approach and stringent requirements for utilizing safety performance indicator field feedback to continuously enhance safety.
At Edge Case Research, we are pioneers in safety engineering and autonomy, and understand how to manage risks faced in our autonomous future. We are now providing Risk Management Services which helps SDV developers to quickly identify and mitigate risks. Our straightforward approach to risk management will mitigate the total costs of risk throughout the self-driving vehicle development and deployment lifecycles, ultimately reducing a developer’s time to market while improving their profitability and enabling SDVs to safely integrate into society. We analyze both existing and potential risks, by identifying their root causes and exploring their probabilities and severities. Our risk control services are far more than a one-size-fits-all product. Our experts address your most pressing issues and follow up with a flexible array of integrated risk management solutions and services. To learn more, you can email us at riskmgt@ecr.ai.
