Determinism in autonomy for applications in the National Airspace System (NAS)

Collaborative efforts between U.S. government agencies and industry have been underway to address certification and operational issues associated with integrating Unmanned Aircraft Systems (UAS) into the National Airspace System. This fact, combined with the obvious potential synergies between UAS and high-level autonomous systems, indicates a high likelihood that in the near future, pressure from industry, the DoD, and research organizations will emerge to certify increasing levels of autonomy for UAS operations in the NAS. The complex and sensitive nature of the NAS will require that the characteristics of and requirements for these autonomous systems, some of which may involve complex decision-making or interacting with humans, be carefully evaluated. A characteristic of autonomous systems that is highly relevant for NAS applications is that of determinism. This note defines and discusses two types of determinism. "Ideal determinism" is exhibited when, given exactly repeatable inputs and initial conditions, a system has exactly repeatable outputs. "Functional determinism," a subtly different concept, is exhibited when, given inputs and initial conditions that are repeatable within the measurement limitations of an observer, the system has repeatable outputs. The need for functional determinism in autonomous systems is crucial for their safe and effective integration into the NAS, both to maximize the performance and potential benefits of such systems and to ensure that the operational environment of the NAS is not degraded for any stakeholders with respect to safety, organization, or ease of operation.