Performance Analysis of Universal Robot Control System Using Networked Predictive Control

TL;DR

This paper evaluates a networked predictive control system for robot arms, demonstrating stability and performance improvements despite high network delays and packet loss, through predictor-based delay compensation and redundant control packets.

Contribution

It implements and tests a robot control system using networked predictive control, enhancing stability and performance under network constraints compared to previous methods.

Findings

System remains stable with high delay and packet loss

Predictor effectively compensates for network delay

Redundant control packets mitigate packet loss

Abstract

Networked control systems are feedback control systems with system components distributed at different locations connected through a communication network. Since the communication network is carried out through the internet and there are bandwidth and packet size limitations, network constraints appear. Some of these constraints are time delay and packet loss. These network limitations can degrade the performance and even destabilize the system. To overcome the adverse effect of these communication constraints, various approaches have been developed, among which a representative one is networked predictive control. This approach proposes a controller, which compensates for the network time delay and packet loss actively. This paper aims at implementing a networked predictive control system for controlling a robot arm through a computer network. The network delay is accounted for by a predictor, while the potential of packet loss is mitigated using redundant control packets. The results will show the stability of the system despite a high delay and a considerable packet loss. Additionally, improvements to previous networked predictive control systems will be suggested and an increase in performance can be shown. Lastly, the effects of different system and environment parameters on the control loop will be investigated.