Pilot task load prediction for human AI teaming

Abstract

Pilots typically experience varying task load during a flight, with high task loads, e.g., during take-off and landing, and lower task loads during parts of the cruise phase. The German Aerospace Center (DLR) developed within the SESAR project DARWIN a system to allocate specific tasks between the automation and the pilot, ensuring that the pilot is neither over- nor unchallenged during the flight. The allocation mechanism requires a prediction of the expected task load along the planned flight trajectory. The Task Load Monitor (TLM) calculates the expected future task load of a flight. This paper will concentrate on the TLM by first explaining how task load is used within the system and how the acceptable level of task load in is defined in this context, and the way the TLM calculates the future task load for a specific flight. The final part will describe TLM’s performance in the first successful real flight validation with the complete system.