Time delay prediction for space teleoperation based on Non-Gaussian Auto-Regressive Model

It is well-known that random large time delay (TD) between the human operator and the telerobot is the main difficulty for a stable teleoperation, so the characteristics and prediction of the TD become rather significant. Though good solutions have been presented under network situation, few works have been done for TD prediction under space situation. Since this lack of knowledge may introduce an obstacle into a successful space teleoperation, this paper conducts a detailed research on TD measurement, model and prediction for space teleoperation. Measure systems are established especially, and probability density distribution is estimated and tested. Then from Non-Gaussian Auto-Regressive Model (NGAR) and measure results, we derive the state space for uplink time delay (UTD) and conduct a successful UTD prediction with absolute error less than 80ms and relative error below 7%. Finally, computer simulation results show that motion precision simulated by the virtual model of the telerobot will be improved with predicted value of the UTD compensated, and future work is outlined in the end of this paper.