In teleoperation, time delays in the communication between host and slave site can significantly degrade task performance. There are two delays, control delay and feedback delay. The sum of both delays is known as roundtrip delay. These time delays are commonly random and time varying, especially when using the Internet as the communication channel.

Delay problems can be improved with predictive displays using a scene model to simulate immediate visual feedback to the human operator. However, their performance depends on a priori knowledge of the roundtrip delay. This assumption is not valid in many applications where the time delay changes over time. For this, a new technique that uses Artificial Neural Network (ANN) predictors in combination with Model Predictive Display (MPD) and a Smith predictor is proposed. This new tech- nique is able to cope with time varying roundtrip delays by providing time-advanced visual feedback from a virtual robot model.

The performance of the proposed technique is investigated by performing experiments on a simulated two-link robot manipulator controlled by a human subject. It is shown that the proposed control method diminishes the effect of feedback delays and control delays up to 500 ms, whereas the Smith predictor without the ANN compensates only for the feedback delay. Additional experiments demonstrate the benefits of including an ANN adaptive control to compensate for model uncertainty.