A model for time-delays for teleoperation over the Internet

The stochastic nature of the communication delay in teleoperation systems makes it hard to predict the behavior of such systems. Although achieving a general-purpose model for the Internet is not easy, if possible at all, stochastic models for specific cases enable us design predictors, which can help in stabilizing the teleoperation systems. In this paper we have developed an initial notion to create an autoregressive model for delays over the Internet. By probing the system behavior at different times of the day for a period of time, the general behavior of the delay can be comprehended and so partially predicted. Introducing a predictor in the teleoperation system can improve the system behavior through for example gain scheduling at the master controller.