The extent of automation and autonomy used in general aviation (GA) has been
steadily increasing for decades, with the pace of development accelerating
recently. This has huge potential benefits for safety given that it is
estimated that 75% of the accidents in personal and on-demand GA are due to
pilot error. However, an approach to certifying autonomous systems that
relies on reversionary modes limits their potential to improve safety.
Placing a human pilot in a situation where they are suddenly tasked with
flying an airplane in a failed situation, often without sufficient
situational awareness, is overly demanding.
This consideration, coupled with advancing technology that may not align with
a deterministic certification paradigm, creates an opportunity for new
approaches to certifying autonomous and highly automated aircraft systems.
The new paths must account for the multifaceted aviation approach to risk
management which has interlocking requirements for airworthiness and
operations (including training and airspace integration). They occur across
a variety of different operational paradigms with varying roles for the
human and the systems in question. If implemented properly, autonomy can
take GA safety to the next level while simultaneously increasing the number
and variety of aircraft and transportation options they provide.