Risk-management of UAS Robust Autonomy for Integration into Civil Aviation Safety Frameworks

Perez, T., Clothier R. A. and Williams, B.

This paper discusses a model of the civil aviation regulation framework and shows how the current assessment of reliability and risk for piloted aircraft has a limited applicability for Unmanned Aircraft Systems (UAS) as technology moves towards higher levels of autonomous decision making. Then, a new framework for risk management of robust autonomy is proposed, which arises from combining quantified measures of risk with normative decision making. The term Robust Autonomy describes the ability of an autonomous system to either continue or abort its operation whilst not breaching a minimum level of acceptable safety in the presence of anomalous conditions. The term combines reliability, safety, and robustness. The decision making associated with risk management requires quantifying probabilities associated with the measures of risk and also consequences of outcomes related to the behaviour of autonomy. The probabilities are computed from an assessment under both nominal and anomalous scenarios described by faults, which can be associated with the aircraft’s actuators, sensors, communication link, changes in dynamics, and the presence of other aircraft in the operational space. The consequences of outcomes are characterised by a loss function quantifies the desirability of the outcomes.

Cite as: Perez, T., Clothier R. A. and Williams, B. (2013). Risk-management of UAS Robust Autonomy for Integration into Civil Aviation Safety Frameworks. In Proc. Australian System Safety Conference (ASSC 2013) Adelaide, Australia. CRPIT, 151. Cant, T. Eds., ACS. 37-45

(from crpit.com) (local if available)