Jeffery A. Schroeder

Effect of Simulator Motion on Pilot Behavior and Perception

A set of experiments were conducted on the University of Toronto Institute for Aerospace Studies flight research simulator to determine the effects of translational and yaw motion on pilot performance, workload, fidelity, pilot compensation, and motion perception for three helicopter yaw control tasks. The three control tasks were a yaw capture, a disturbance rejection task, and a tracking task. The yaw capture experiment was a duplication of an experiment previously run at a different simulator facility. The results of the yaw capture task were in general agreement with the previous study with the exception that, in the current study, yaw motion had a larger impact on pilot performance than the previous study. The current study found that translational motion improves performance and increases fidelity for all three tasks. Yaw motion increased performance for the yaw capture and disturbance rejection tasks. Translational motion generally improved fidelity and was easier to detect than yaw motion for all three tasks. Finally, if translational motion was present, the addition of yaw motion usually provided little additional benefit to performance, workload, compensation, or fidelity for all three tasks.

Pilot Behavioral Observations in Motion Flight Simulation

This paper summarizes key findings on how simulator platform motion affects pilotvehicle performance and workload. These measures are a subset of the broad topic of pilot behavior, namely those associated with fine motor skills. The findings include the effects of specific tasks, vehicle dynamics, interaction with visual cues, and cockpit manipulator cues. The results are in the context of vehicle systems development rather than training. A section reviews a set of motion fidelity criteria and compares it with the literature. The criteria allow an assessment of a simulator’s motion cueing strengths and weaknesses for a given task. The paper offers best practices for achieving pilot behavior in a flight simulator that is more akin to in-flight behavior. Three practices are most important: quantify all of the available cues analytically, evaluate the motion cues against available criteria, and strive to find relevant tasks that permit the motion cues to have at least adequate fidelity.

Simulator Platform Motion Effects on Pilot-Induced Oscillation Prediction

Simulator motion platform characteristics were examined to determine if the amount of motion affects pilotinduced oscillation prediction. Five test pilots evaluated how susceptible 18 different setsof pitch dynamics were to pilot-induced oscillations with three different levels of simulation motion platform displacement: large, small, and none. The pitch dynamics were those of a previous in-e ight experiment, some of which elicited oscillations. These in-e ight resultsserved as truth data for the simulation. As such, the in-e ight experiment wasreplicated asmuch as possible. Objective and subjective data were collected and analyzed. With large motion, pilot-induced oscillation and handling qualities ratings matched the e ight data more closely than with small motion or no motion. Also, regardlessof the aircraft dynamics, large motion increased pilot cone dencein assigning handling qualitiesratings, reduced safety pilot trips, and lowered touchdown velocities. Whereas both large and small motion provided a pitch rate cue of high e delity, only large motion presented the pilot with a high e delity vertical acceleration cue.

Evaluation of a motion fidelity criterion with visual scene changes.

An experiment examined how visual scene and platform motion variations affected a pilots ability to perform altitude changes. Pilots controlled a helicopter model in the vertical axis and moved between two points 32-ft apart in a specified time. Four factors were varied: visual-scene spatial frequency, visual-scene background, motion-filter gain, and motion-filter natural frequency. Drawing alternating black and white stripes of varying widths between the two extreme altitude points varied visual-scene spatial frequency. The visual-scene background varied by either drawing the stripes to fill the entire field of view or by placing the stripes on a narrow pole with a natural sky and ground plane behind the pole. Both the motion-filter gain and natural frequency were varied in the motion platform command software. Five pilots evaluated all combinations of the visual and motion variations. The results showed that only the motion-filter natural frequency and visual-scene background affected pilot performance and their subjective ratings. No significant effects of spatial frequency or motion system gain were found for the values examined in this tracking task. A previous motion fidelity criterion was found to still be a reasonable predictor of motion fidelity.

PILOT-INDUCED OSCILLATION PREDICTION WITH THREE LEVELS OF SIMULATION MOTION DISPLACEMENT

Simulator motion platform characteristics were examined to determine if the amount of motion affects pilot-induced oscillation (PIO) prediction. Five test pilots evaluated how susceptible 18 different sets of pitch dynamics were to PIOs with three different levels of simulation motion platform displacement: large, small, and none. The pitch dynamics were those of a previous in-flight experiment, some of which elicited PIOs. These in-flight results served as truth data for the simulation. As such, the in-flight experiment was replicated as much as possible. Objective and subjective data were collected and analyzed. With large motion, PIO and handling qualities ratings matched the flight data more closely than did small motion or no motion. Also, regardless of the aircraft dynamics, large motion increased pilot confidence in assigning handling qualities ratings, reduced safety pilot trips, and lowered touchdown velocities. While both large and small motion provided a pitch rate cue of high fidelity, only large motion presented the pilot with a high fidelity vertical acceleration cue.

EFFECTS OF VEHICLE BANDWIDTH AND VISUAL SPATIAL-FREQUENCY ON SIMULATION CUEING SYNCHRONIZATION REQUIREMENTS

Results of a recent flight simulation study suggested criteria for visual and motion cueing synchronization, but only for one vehicle bandwidth and one visual scene. The purpose of the study reported here was to determine if those synchronization criteria may be generalized. In particular, a complete factorial design was used to examine the flight simulation effects of the following five experimental factors: visual time delay, roll motion time delay, lateral motion time delay, vehicle bandwidth, and visual spatial frequency. Five experimental test pilots completed the full experimental matrix. The results show that the more limited set of synchronization criteria generalize for the variations examined. That is, regardless of the vehicle bandwidths or the visual cueing spatial frequencies examined, the same synchronization criteria are applicable for the visual, roll, and lateral cues. Thus, the results add further confidence to the recently suggested criteria, which suggested three guidelines. First, roll and lateral motion cues should be synchronized, but if they cannot be, the asynchronization should be no more than 40 msec. Second, the visual and roll motion cues should also be synchronized, but if they cannot be, the asynchronization should also be no more than 40 msec. Third, the synchronized roll and lateral motion cues can be allowed to lag, but should not lead, the visual cues. Nomenclature

Pre and Post Pilot Model Analysis Compared to Experimental Simulator Results

‡In 2003, a workshop on the use of pilot models to analyze the influence of visual and vestibular motion perceptio n on a pilot’s control behavior was held. As an example, an earlier experiment on the Vertical Motion Simulator ( VMS ) at NASA Ames was used to demonstrate that a pilot model analysis could reasonably explain the trends of the experimental results. Mo re importantly, the same analysis suggested an experimental configuration that, if tested, might have changed the general conclusions found in the NASA Ames study. Useful progress in research is enhanced when new studies and analyses build upon previous st udies and analyses. A new simulator experiment built upon the NASA Ames study and also evaluated all the new conditions analyzed with the pilot model during the workshop noted above. The results roughly correspond with the analysis. Although, the utmost w as done to produce the same motion cues in the University of Toronto research flight simulator as in the VMS experiment before, differences in the experimental results were found between the results of the NASA experiment and the University of Toronto experiment. The paper reviews the initial analysis, compares the analysis results with the new results from the new experiment, and readjust s the pilot mode l analysis with the new results. † ‡ p y

The Effect of Simulator Motion on Pilot's Control Behavior for Helicopter Yaw Control Tasks

A set of experiments were conducted on the UTIAS flight research simulator to determine the effects of translational and yaw motion on pilot performance, workload, fidelity, pilot compensation, and motion perception for three helicopter yaw control tasks. The three control tasks were a yaw capture, a disturbance rejection task, and a tracking task. The yaw capture experiment was a duplication of an experiment previously run at a different simulator facility. The results of the yaw capture task were in general agreement with the previous study with the exception that in the current study yaw motion had a bigger impact on pilot performance than the previous study. The current study found that translational motion improves performance and increases fidelity for all three tasks. Yaw motion increased performance for the yaw capture and disturbance rejection tasks. Translational motion generally improved fidelity and was easier to detect than yaw motion for all three tasks. Finally, if translational motion is present, the addition of yaw motion usually provided little benefit to performance, workload, compensation or fidelity, for all three tasks. p x f = specific force at pilots location in x-direction of pilot body frame, m/s 2 p y f = specific force at pilots location in y-direction of pilot body frame, m/s 2

Spatial frequency and platform motion effects on helicopter altitude control

An experiment examined how visual scene and platform motion variations affected a pilot’s ability to perform altitude changes. Pilots controlled a helicopter model in the vertical axis and moved between two points 32-ft apart in a specified time. Four factors were varied: visual scene spatial frequency, visual scene background, motion filter gain, and motion filter natural frequency. Drawing alternating black and white stripes of varying widths between the two extreme altitude points varied visual scene spatial frequency. Visual scene background varied by either drawing the stripes to fill the entire fieldof-view or by placing the stripes on a narrow pole with a natural sky and ground plane behind the pole. Both the motion filter gain and natural frequency were varied in the motion platfom command software. Five pilots evaluated all combinations of the visual and motion variations. The results showed that only the motion filter natural frequency and visual scene background affected pilot performance and their subjective ratings. No significant effects of spatial frequency or motion system gain were found for the values examined in this tracking task. A previous motion fidelity criterion was found to still be a reasonable predictor of motion fidelity.

Transfer of Training on the Vertical Motion Simulator

This paper describes a quasi-transfer-of-training study in the NASA Ames Vertical Motion Simulator (VMS). Sixty-one general aviation pilots trained on four challengingcommercial transport tasks under one of four different motion conditions: no motion, small hexapod, large hexapod, and VMS motion. Then, every pilot repeated the tasks in a check with VMS motion to determine if training with different motion conditions had an effect on task performance. New objective motion criter ia guided the selection of the motion parameters for the small and large hexapod conditions. Considering resu lts that were statistically significant, or marginally significant, the motion condition used in training affected 1) longitudinal and lateral touchdown position; 2) the number of secondary stall warnings in a stall recov ery; 3) pilot ratings of motion utility and maximum load factor obtained in an overbanked upset recovery; an d 4) pilot ratings of motion utility and pedal input reaction time in the engine-out-on-takeoff task. Sinc e the training motion conditions revealed statistical differences on objective measures in all the tasks pe rformed in the VMS motion check, with some in the direction not predicted, trainers should be cautious not to oversimplify the effects of platform motion. Evidence suggests that the new objective motion criteria may off er valid standardization benefits, as increases in the training motion fidelity, as predicted by the two conditions covered by the criteria, resulted in expected trends in pilot ratings and objective performance measures afte r transfer.