Analysis of Skill Development in Manual Ramp Tracking Tasks: Using a Feedforward Pilot Model

Analysis of Skill Development in Manual Ramp Tracking Tasks: Using a Feedforward Pilot Model

Willems, M.E.

Mulder, M. (mentor)
Pool, D.M. (mentor)
Van Paassen, M.M. (mentor)
Damveld, H.J. (mentor)

Aerospace Engineering

Control and Simulation

Human Machine Interaction

2012-03-22

A study is performed analyzing the behavioral adaptation of pilots in skill based ramp tracking tasks performed with pursuit display. From data analyses of previous ramp tracking experiments an indication is found that pilots who repeatedly perform the same ramp tracking task show a high level of adaptation to characteristics of the control task in their behavior. From the theory of Successive Organization of Perception it is expected that this development progresses towards precognitive behavior. To get to a better understanding of the transition between pursuit behavior and precognitive behavior, a pursuit ramp tracking and disturbance rejection experiment is performed to further study this. For this four conditions with a different ramp steepness are used in blocks of ten runs, with eight similar ramps in each run. Two sessions are performed, one ordered session where the same condition is performed ten times consecutively and a random session, where conditions are presented in random order. After each block of ten ordered runs also a surprise run with a different condition is presented. The behavior is analyzed using performance metrics and by performing a time-domain parametric identification of a combined compensatory and feedforward pilot model. The behavior is analyzed in three ways: first by using the averaged data of the final five runs as steady-state behavior, then by using the data per run to investigate the development over the runs and finally the data per ramp is analyzed to investigate developments within a run. The surprise runs are analyzed over the entire run and per ramp. From performance and behavioral metrics no difference in ramp tracking behavior is shown when conditions are presented consecutively as opposed to randomly. From the model parameters though, it was found that the difference in behavior is predominantly visible in the feedforward pilot gain, and it is shown that the difference in behavior is mostly apparent in the first ramps of each tracking run. After the first two ramps, the pilots are found to already have adjusted their behavior towards their optimal control setting. From the surprise runs strong indications were found that pilots need more time to adjust after being conditioned in the previous ten runs to a different condition. This suggests that pilots use precognitive control strategies when they are more experienced in the task performed and think they know what to expect. Novel research should be performed to extend existing pilot models by modeling precognitive pilot behavior

Control
Simulation
Human Machine Interaction
HMI
Modeling
Skill Development
feedforward
Skill

http://resolver.tudelft.nl/uuid:3a061451-78d2-4a22-922f-0116ca6662ea

2016-06-01

Student theses

master thesis

(c) 2012 Willems, M.E.