Using Machine Learning Algorithms to Develop Adaptive Man–Machine Interfaces

Automation has been introduced in aircraft cockpits to reduce pilot workload and increase safety. However, a number of reports mention “human factors” issues and misunderstandings of an automated-system behavior or its displays by the crew as major contributors leading to flight incidents. The interfaces should play, nevertheless, a crucial role in improving man–machine cooperation, by displaying “the right information at the right time.” This need of adapted displays and interfaces is more important than ever as the missions are becoming more and more complex, especially in the military domain. Moreover, many factors in workload mitigation are identified as crew or mission dependent and are highly variable from one flight to another, such as the cognitive demands of the current phase of flight or mission situation, the pilot's experience or “airmanship,” or individual physiological parameters. Thus, we can think of an adaptive intelligent interface that would monitor the automated system–pilot team as well as the mission operational context to provide the correct display and controls to the user and enable better cooperation between the human operator and the machine to match the current demands of the operational situation. This paper aims to investigate the potential of machine learning algorithms to develop these adaptive intelligent interfaces.