Modeling Human Workload in Unmanned Aerial Systems

Unmanned aerial systems (UASs) often require multiple human operators fulfilling diverse roles for safe correct operation.  Although some dispute the utility of minimizing the number of humans needed to administer a UAS, minimization remains a long-standing objective for many designers.  This paper presents work toward understanding how workload is distributed between multiple human operators and multiple autonomous system elements in a UAS across time, with an ultimate goal to reduce the number of humans in the system. The approach formally models the actors in a UAS as a set of communicating finite state machines, modified to include a simple form of external memory. The interactions among actors are then modeled as a directed graph.  The individual machines, one for each actor in the UAS, and the directed graph are augmented with workload metrics derived from a review of the relevant literature. The model is implemented as a Java program, which is analyzed by the Java Pathfinder (JPF) model checker, which generates workload profiles over time.  To demonstrate the utility of the approach, this paper presents a case study on a wilderness search and rescue (WiSAR) UAS analyzing two different mission outcomes. The generated workload profiles are shown to be consistent with known features of actual workload events in the WiSAR system.