Levels of Aviation Autonomy

Abstract

Most modern airplanes include automated systems to improve both the efficiency and safety of flight. By the standards of the automotive industry, these avionics systems are both highly complex and highly automated. Ongoing advancements in computing systems and machine learning lead the SAE to publish a Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles that has become the standard for classifying levels of driving automation. This standardization has provided the automotive industry and the regulatory agencies a useful framework for exploring and promoting the development of more advanced automated and autonomous systems. The aviation industry is also benefiting from advancements in technology, with growing interest in even more advanced flight systems and ultimately fully autonomous solutions. The higher demands of the piloting task relative to the driving task and the more complex nature of modern airplane systems lead the authors to re-evaluate and tailor the SAE classifications of autonomy specifically for pilots and avionics solutions. This paper summarizes a taxonomy of flight automation system levels that Rockwell Collins has derived from the SAE driving automation system levels. The intent is to provide a similar framework for the exploration and development of advanced automation and autonomous solutions for avionics. Similar to the SAE classifications, this paper clarifies the role of both the (human) pilot, and the automated systems during flight operations at each level of automation. The objective is to provide scope definition and a useful framework for the development of technical specifications and related policies, regulations and standards.