Automated Online Determination of Pilot Activity Under Uncertainty by Using Evidential Reasoning

The objective of a workload-adaptive associate system is to support human pilots in critical workload situations to avoid excessive demands on their mental capacity. Since workload strongly depends on the current activity of the pilots, i.e. the tasks the pilots are performing in a specific task situation, a key feature of an adaptive associate system is to determine the activity of the pilots online in real-time. This contribution presents a method for determining the activity of helicopter pilots automatically in an uncertain and complex environment like military manned-unmanned teaming missions (MUM-T), where multiple unmanned aerial vehicles (UAVs) are commanded from the cockpit of a manned helicopter. We use a pilot observation system with different measurement sensors to collect evidences and apply an evidential reasoning algorithm to draw conclusions on the actual activity of the pilots. Our method is based on a simplified version of Dempster-Shafer theory and is capable of collecting and combining even contradictory evidences. By providing a means of implicit deliberative communication, this method lays a foundation for improving human-machine team performance in complex task situations. The implementation of this method in a helicopter mission simulator is explained in detail.