Partial camera automation in a simulated unmanned air vehicle

With the rapid development of automatic control techniques a central question is how the division of labor between the human operator and the automaton should be optimally distributed. In this connection, the present study focussed on an intelligent, semi-autonomous, interface for a camera operator of a simulated Unmanned Air Vehicle (UAV). This interface used inherent system 'knowledge' concerning UAV motion in order to assist a camera operator in tracking an object moving through the landscape below. This landscape was sensoreddeo camera attached to the UAV-platform and presented to the operator on a monitor display. The semi-automated system compensated for the translations of the UAV relative to the earth. This compensation was accompanied by the appropriate joystick movements ensuring tactile (haptic) feedback of these system interven-tions. The operator had to superimpose camera movements over these system actions required to track the motion of a target (a driving truck) relative to the terrain. Consequently, the operator remained in the loop; he still had total control of the camera-motion system. In order to investigate the effects of this semi-automation over a broad range of task situations, the tracking task was carried out under two conditions of update frequency of the monitor image and control mode difficulty.
The data showed that subjects performed substantially better with an active system. Apparently, the subjects had no difficulty in maintaining control; i.e., 'following' the active stick wve stick while superimposing self-initiated control movements over the system-interventions. Furthermore, tracking performance with an update frequency of 5 Hz was clearly superior relative to 2 Hz. The magnitude of the active-interface effect was equal to the update-frequency effect. On the basis of effects of difficulty in steering dynamics, it was also concluded that the benefits of update frequency enhancement and semi-automated tracking will be the greatest under difficult tracking conditions. Mental workload scores indicated that for the difficult tracking-dynamics condition, both semi-automation and frequency increase resulted in less experienced mental effort in task perform-ance. For the easier dynamics this effect was only seen for update frequency.