Increasing Pilot's Understanding of Future Automation State – An Evaluation of an Automation State and Trajectory Prediction System

Abstract

A pilot in the loop flight simulation study was conducted at NASA Langley Research Center to evaluate a trajectory prediction system. The trajectory prediction system computes a five-minute prediction of the lateral and vertical path of the aircraft given the current and intent state of the automation. The prediction is shown as a graphical representation so the pilots can form an accurate mental model of the future state. Otherwise, many automation changes and triggers are hidden from the flight crew or need to be consolidated to understand if a change will occur and the exact timing of the change. Varying dynamic conditions like deceleration can obscure the future trajectory and the ability to meet constraints, especially in the vertical dimension. Current flight deck indications of flight path assume constant conditions and do not adequately support the flight crew to make correct judgments regarding constraints.The study was conducted using ten commercial airline crews from multiple airlines, paired by airline to minimize procedural effects. Scenarios spanned a range of conditions that provided evaluation in a realistic environment with complex traffic and weather conditions. In particular, scenarios probed automation state and loss of state awareness. The technology was evaluated and contrasted with current state-of-the-art flight deck capabilities modeled from the Boeing 787.Objective and subjective data were collected from aircraft parameters, questionnaires, audio/video recordings, head/eye tracking data, and observations. This paper details findings about the trajectory prediction system including recommendations about further study.

Usability Evaluation of Indicators of Energy-Related Problems in Commercial Airline Flight Decks

“…Each crew began with a few runs for familiarization with the simulator and the new technologies. Data collection runs followed these training runs [16]. Data were collected individually from the pilot flying (PF) and the pilot monitoring (PM).…”