A Preliminary Study of Human Pilot Dynamics in the Control of Time-Varying Systems

Abstract

An exploratory examination of time-varying human pilot behavior is presented. A procedure for creating a real-time, pilot-in-the loop simulation environment using Simulink® is discussed. A series of single-axis compensatory tracking tasks are conducted with a single, well-trained subject. Identification of pilot dynamics is undertaken using the output error model estimation algorithm. The results suggest that the human pilot is capable of generating dynamics associated with the crossover model of the human during controlled element transitions. A model matching exercise is included in which the parameters of a simplified pursuit control model of the human pilot are varied to provide an acceptable match to the identified open-loop pilot/vehicle transfer function. A brief discussion of the statistical properties associated with identification techniques is presented followed by a discussion of alternate identification schemes. The latter includes a second estimation algorithm (gettf1) in which the optimal regressor order is found based upon a user-defined upper limit on the number of free parameters in the numerator and denominator. The study continues with the exercise of gettf1 in analyzing pilot-in-the-loop three-axis tracking data involving the attitude control of a helicopter model at hover. Finally, a pilot-in-the-loop simulation was conducted that utilized a simple model-reference adaptive control system that attempted to compensate for changes in vehicle dynamics. A discussion and conclusion sections complete the study.

Multimodal Pilot Behavior in Multi-Axis Tracking Tasks with Time-Varying Motion Cueing Gains

“…[1][2][3] In many situations, however, it is the human controller's adaptation to changes in important task variables that is actually of interest. 4 An example of a situation where such time-varying control behavior occurs, and for which interest has been steadily increasing, [5][6][7][8][9] is one where a pilot is faced with a sudden change in aircraft dynamics. Such a change in the controlled system dynamics can occur, for instance, in the case of a damaged aircraft and typically requires significant adaptation of the pilot's control behavior to retain stability and satisfactory performance.…”

“…For studying such time-varying characteristics of manual control behavior, progress has been made recently in the modeling and identification of time-varying pilot control behavior. [5][6][7][8][9] Another situation in which time-varying manual control behavior may be induced is in flight simulators with adaptive motion cueing. To make more efficient use of a flight simulator's motion workspace, it is common to apply motion drive algorithms with time-varying dynamics.…”

Adaptive State Predictor Based Human Operator Modeling on Longitudinal and Lateral Control

“…In an attempt to simplify the model while still capturing the essential pilot behavioral characteristics, Hess proposed a simplified model that has been recently applied to systems with time-varying dynamics. 11,12 In addition to several relevant model structures to describe the pilot-as-a-controller behavior, there has also been significant work on techniques used to identify the pilot model of a given structure. [13][14][15][16][17][18][19] These included various applications of parameter identification ranging from wavelets-based approach to frequencybased system identification techniques.…”

Wetware, Hardware, or Software Incapacitation: Observational Methods to Determine When Autonomy Should Assume Control

“…while still capturing the essential pilot behavioral characteristics, Hess proposed a simplified model that has been recently applied to systems with time-varying dynamics 12,13 .…”