Applicability of UL 4600 to Unmanned Aircraft Systems (UAS) and Urban Air Mobility (UAM)

Abstract

Aviation is a conservative field with a focus on deterministic systems that continuously builds on known risks. Currently, available aviation design assurance methods using safety assurance standards are not suited for regulatory assurance of autonomous systems Aviation is facing enormous growth in Unmanned Aircraft Systems (UAS) and Urban Air Mobility (UAM) technology. The challenge for applicants and regulators is to assure system safety of autonomous systems. The automobile industry, which is similarly challenged by autonomous systems, is finding some possible solutions. This paper explores whether the aviation industry can apply those solutions to UAS and UAM assurance. ANSI/UL 4600, Standard for Safety for the Evaluation of Autonomous Products, is the first and only publicly available autonomous safety standard. This standard is backed by a certified non-profit standards organization, Underwriters Laboratories. This standard has been recognized as an international American National Standards Institute (ANSI) approved standard. The goal-based, technology neutral features of ANSI/UL 4600 make it adaptable to other industries and applications. It assures safety through monitoring safety performance, which enables the desired aviation regulatory approach. It reflects stakeholder views from Original Equipment Manufacturers (OEMs) who have contemplated the product liability for the design and manufacturing challenges associated with these new technologies. The applicability to the aviation industry can be understood in terms of these features: Goal-Based and Technology Neutral: ANSI/UL 4600 does not require a specific design approach or specific technology. The applicant company must present an auditable and defensible evidence-based safety case. Those central features of safety and auditability follow the Federal Aviation Administration (FAA) compliance philosophy and could be adapted to aviation. Continued Airworthiness with Monitored Safety Metrics: The FAA's desire to move to performance-based standards with risk-based strategy is supported by ANSI/UL 4600. Further, ANSI/UL 4600 recognizes that continued safety is even more important than one-time initial approval/acceptance, especially for learning systems used in problem-solving Artificial Intelligence (AI). Functional transformations of learning systems are continuously monitored by OEMs. Continued airworthiness could be assessed using this method even though the system is presumably fielded with unknown risks due to new and novel technology. Modernization of Regulatory Approaches: This methodology supports performance-based regulations with compliance required to enforce system-of-systems interoperability. Applicants could design and test appropriate to their technology while regulators can audit and check for safety case completeness. Over time, industry can standardize some portions of testing as the technology matures. Different standards and codified standard approaches could also be independently designed and adapted as technology becomes standardized across the industry. Aviation regulators are already on this path to make performance-based regulations where applicants have the choice of technology and standards that can support innovation. Industry and Public Stakeholder Engagement: The standard has undergone industry and public scrutiny in the automobile segment with considered reviews from technical, legal, and actuarial perspectives. The FAA and The MITRE Corporation (MITRE) have been working to implement a safety case umbrella that encompasses all facets of UAS and UAM development, operation, and maintenance. The safety case umbrella is a systems approach to combine specific UAS and UAM nuances as they relate to current regulatory structure. While the approach affords a performance-based and risk-based oversight of aircraft design and development, production, maintenance and operation, the current regulations were not written to address AI and specifically machine learning. Even when those advanced techniques are not used, small UASs often use commercial-off-the-shelf (COTS) technology and technology of unknown pedigree could also be assured by the regulators using that approach. The codified approach of ANSI/UL 4600 could be superimposed on the safety case umbrella approach based on current regulations. The exercise of creating an auditable safety case also increases applicant safety awareness and safety culture. For approval of UAS and UAM, this paper recommends a structured and practical framework with an evidence-based safety case, which also takes into consideration the current regulations for applicability to UAS and UAM safety.