Teleoperation of remotely located space/underwater vehicles requires the human operator to interact with time delayed vehicle responses to issued commands. This often results in the adoption of a ``move and wait'' strategy whereby the vehicle operator waits to view the results of the previous command before issuing the next command. This work investigates combining a command display (CD) located at the operator control station with a model following controller residing on the remote vehicle to allow the teleoperator to interact with the vehicle in a more seamless manner in time delayed environments.

Command displays differ from more traditional predictive displays in two major ways. First, in a CD system, a trajectory tracking controller is located on the remote vehicle; the teleoperator interacts with an ideal kinematic model of the vehicle at the control station, with the controller forcing the vehicle to fly the indicated trajectory. Second, model information resides on board the vehicle controller in a CD. In a predictor display system the prediction model is located at the control station.

The utility of implementing a CD on a full 6 DOF dynamic simulation of an underwater remotely operated vehicle (UROV) is examined. The task involves 18 subjects maneuvering the UROV through an obstacle course. Sensitivity of the CD to model accuracy is addressed, i.e. does implementation of a vehicle controller capable of adaptation in the presence of model uncertainty improve performance. Successful implementation of an adaptive CD is demonstrated.

Results indicate that the CD is instrumental in improving performance for teleoperated systems with signal transmission delays as seen by decreased completion times, improved accuracy and more consistent use of hand controllers. In addition, the CD proves surprisingly robust to model inaccuracy when time delay is present. On the other hand, results indicate that implementation of a CD may be contraindicated in the absence of time delay. There is evidence that discrepancies between the actual and desired vehicle due to controller accuracy may have confused the test subjects. Task completion times are higher and subjects are less accurate when the CD is implemented with no time delay.