Quantifying the Safety Benefits of a Digital Copilot in General AviationMay 2020
Topics: Aviation and Aeronautics, Aviation Safety, Artificial Intelligence, Software Engineering (General)
Digital Copilot is a prototype information system providing cognitive assistance to pilots in general aviation. This paper presents methods and results of an analysis quantifying the expected safety benefits of the system.
The methods include: Cognitive Performance Analysis to quantify pilot workload in flight activities; Human Reliability Analysis to quantify pilot error probabilities; and Probabilistic Risk Assessment to quantify accident sequence frequencies. These methods are used to model flight risk with and without pilots' use of Digital Copilot, and the comparison provides a quantitative measure of expected safety benefits assuming the prototype system is adopted by pilots.
The results suggest that Digital Copilot technology has the potential to significantly reduce the rates of total and fatal accidents in general aviation. The same methods also enable a detailed analysis of individual pilot actions and system features, to quantify their risk importance and identify the most promising opportunities for further system improvements.