Cybernetics of Tunnel-in-the-Sky Displays

More Info


Consensus is growing that the flexibility gained with the introduction of programmable, electronic cockpit displays in the 1980s must be exploited to the full extent. An important candidate to become the primary flight display of future flight decks is the tunnel-in-the-sky display, a perspective flight-path display that shows the reference trajectory to be flown in a synthetic three-dimensional world. The usefulness of the tunnel display in the pilot manual aircraft control task is the subject of this thesis. The mainstream of tunnel display research is confined to empirical comparisons of the tunnel display with conventional displays. The approach taken in the present theoretical and experimental study is original and new as it is conducted from the perspective of cybernetics. A four-stage methodology is developed to study the fundamental characteristics of pilot/display interaction, based on a theoretical analysis of information, in particular the information used for control. The information analysis is conducted within the context of Gibson's ecological approach to visual perception. The information analysis provides novel insights into how the tunnel display geometric design variables can affect pilot behavior. To examine the validity of the theoretical hypotheses, six experiments have been conducted. Three experiments examined the effects of manipulating some of the main display design variables, such as the tunnel size, the viewing volume and the presence of guidance symbology. Another three experiments investigated the fundamental characteristics of the tunnel geometrical design in the tasks of following a trajectory that is either straight or circular, and in the task of conducting a curve-interception maneuver. The experiments show that the cybernetic, information-centered approach is indeed very successful in pin-pointing the important characteristics of pilot/display interaction. The experimental methodology employed in this thesis aimed at integrating the model-based approach with the common approach of collecting mainly performance-related data. It is described in detail how experiments can be designed with the objective of conducting a control-theoretic analysis. The limitations of some non-parametric identification methods in multi-axis, multiple loop tracking tasks are described. The use of criterion functions, in both the frequency and the time domain, in the parametric identification methods is also exemplified.