Datensatz

Zusammenfassung

Hovering a helicopter is a difficult task. Helicopter dynamics are unstable, comparable to an inverse pendulum. Helicopter pilots need to actively control the position and orientation of the
helicopter at every moment in time. It is assumed that they use not only visual information, but
also body cues (vestibular/somatosensory) for this task.
To test this hypothesis, we measured human stabilisation performance in a helicopter simulator
(similar to a Robinson R-22 helicopter) using a Stewart platform. We compared four
body motion cueing conditions: platform off, platform translations, platform rotations, and
both translations and rotations. Washout-filters were used for platform translations whereas
pitch and roll rotations were directly transferred to the platform. A computer-generated scene
was projected onto a screen (70x54 degrees visual field) 1.15m in front of the participants. The
stimulus consisted of a textured ground plane with a target ball fixed to the ground and a second
ball as position marker (fixed to the helicopter 15 meters in front). The visual scene was always
shown from an observer in the simulated helicopter. The task was to control the helicopter so
that the two balls remained as close as possible to each other, while minimizing heading drift.
Each trial lasted for two minutes. Trials with different motion cueing were interleaved.
Participants controlled three of the four axes of the helicopter: pitch (forward-backward
acceleration), roll (sideways acceleration), and yaw (heading orientation). Height above ground
was automatically controlled by the simulator. Only subjects who after training managed to
stabilize the helicopter with a mean distance of less than 5 meters were included in the study.
For every trial, we measured the mean distance between target and helicopter position marker,
mean velocity of the marker, and mean pitch and roll angles of the helicopter. We then analyzed
the influence of the motion cueing conditions on these measured variables.
We found that participants could better stabilize the helicopter if body motion cues were
available. Body rotations had a larger beneficial effect on stabilization performance than body
translations. The mean distances from the target were smaller in the sideways direction than in
the forward-backward direction for all four conditions.
We conclude that participants can use body motion cues to increase their stabilisation performance
when controlling a simulated helicopter. Possibly humans are not very sensitive
to small visual acceleration, and body motion can help to overcome this limitation. As the
simulated helicopter always accelerates in the direction in which it is tilted, body tilt can provide
immediate feedback to the pilot in which direction he is accelerating, even before he can
identify an acceleration visually. These findings have important implications for the design of
helicopter simulation platforms.