Predictive Input Delay Compensation with Grey Predictor for Networked Control System

摘要

The performance of networked control systems is affected strictly by time delay. Most of the literature in the area handle the problem from a stability perspective. However, stability optimized algorithms alone are not sufficient to reduce synchronization problems caused by time delay between the action and reaction in geographically distant places, and the effect and performance of other system components should also be taken into account. In teleoperation applications the reference is often provided by a human, known as the operator, and due to the nature of the human system, references provided by the human operator are of a much lower bandwidth when compared to common control reference inputs. This paper focuses on the operator, and proposes an approach to predict the manipulator's motion (created by the operator) ahead of time with an aim to reduce the time delay between the master and slave manipulator trajectories. To highlight the improvement offered by the developed approach, hereby called Predictive Input Delay Compensator (PIDC), we compare the performance with the only other study in the literature that handles this problem using the Taylor Series approach. The performance of these two approaches is evaluated experimentally for the forward (control) path on a PUMA robot, manipulated by a human operator and it has been demonstrated that the efficient latency in the forward path is decreased by 100ms, on average, reducing the forward latency from 350ms to 250ms.