Safe shared control between pilots and autopilots in the face of anomalies

Abstract

As societal drivers of sustainability, efficiency and quality of life become more urgent and intensive, analysis and synthesis of safety critical systems in the face of anomalies become extremely important. In applications such as fully autonomous ground or air transportation, in electrical grids, and healthcare, often there are two decision-makers, one of which is automation, and the other is a human expert. While there have been several studies undertaken to understand the role of automation, and that of human experts, how the two decision-makers can carry out a shared control in a safety-critical system in the face of anomalies has not been investigated in depth. Our focus in this chapter is on two different shared control architectures for a cyber–physical–human system (CPHS), where the decision-making of the human expert is judiciously combined with that of an advanced automation with cyber components of sensing, computation, communication, and actuation. These architectures are evaluated in the context of flight control problems when severe anomalies are present. It is argued that for a successful synthesis of CPHS a granularity assignment of task allocation and timeline has to be carried out which enables the coordination between human and automation, the specific tasks that they carry out, and the timeline associated with these tasks, all in the context of an anomaly. Models of the physical system, the automation, and the pilot using two different shared control architectures are discussed. Validation of these architectures with a simulation study with human-in-the-loop of flight control problems is reported, demonstrating the design of successful CPHS in the presence of severe anomalies.