Development of a Novel Model of Pilot Control Behavior in Balked Landings

This paper describes the development of a new pilot model to support simulations within the Balked Landing Study conducted under the auspices of the ICAO Obstacle Clearance Panel (OCP). The objective of the study is to determine the obstacle free zone (OFZ) balked landing surfaces for New Larger Aeroplanes (NLA). An earlier effort has identified the OFZ for landings under control of the auto-pilot. The purpose of the current study is to determine similar surfaces under the control of the pilot using the flight director for guidance during Instrumented Landing System (ILS) landings. A collaborative effort to develop a pilot model has been undertaken, as part of which Eurocontrol contracted QinetiQ Centre for Human Sciences (CHS) to contribute to the construction of an appropriate pilot model. This model will be used in conjunction with the Boeing 747-400 aircraft model, supplied by Boeing, to simulate balked landing behaviour in a Monte Carlo simulation study conducted by the US Federal Aviation Authority (FAA). The purpose of the development tasked by Eurocontrol, as part of the larger project, was to build a pilot model that is capable of generating the range of pilot pitch and roll control when landing under guidance from the flight director in the full range of environmental conditions, and includes individual pilot differences. A novel approach was adopted for the development of the QinetiQ pilot model, based on work undertaken with the Integrated Performance Modelling Environment (IPME). The model is based on a discrete event representation of pilot control movements for both pitch and roll control under the guidance of the flight director. At the heart of the model is a cycle of assessment of the current state of the aircraft. A control movement is performed as a stochastic function of the deviation of the current state from the desired state. If no control movement is made, there is a short wait and the cycle is repeated. The data used for calibration purposes were drawn from a trial conducted at the NASA Ames facility using a high fidelity full motion simulator of the Boeing 747-400 aircraft. Model parameters were calibrated for each pilot individually using the Nelder Mead simplex method to match the model behaviour to the observations. The pattern of variation of the predicted power spectrum of control movements was compared with the observed pattern to ensure that variation between pilots and environmental conditions was reflected correctly by the model. The results from the calibrations are described in the paper, and the observed and fitted variation in pilot control movement spectral activity is described.