Control of Networked Robotic Systems

Thumbnail Image


Publication or External Link






With the infrastructure of ubiquitous networks around the world, the study of robotic systems over communication networks has attracted widespread attention. This area is denominated as networked robotic systems. By exploiting the fruitful technological developments in networking and computing, networked robotic systems are endowed with potential and capabilities for several applications. Robots within a network are capable of connecting with control stations, human operators, sensors, and other robots via digital communication over possibly noisy channels/media. The issues of time delays in communication and data losses have emerged as a pivotal issue that have stymied practical deployment. The aim of this dissertation is to develop control algorithms and architectures for networked robotic systems that guarantee stability with improved overall performance in the presence of time delays in communication.

The first topic addressed in this dissertation is controlled synchronization that is utilized for networked robotic systems to achieve collective behaviors. Exploiting passivity property of individual robotic systems, the proposed control schemes and interconnections are shown to ensure stability and convergence of synchronizing errors. The robustness of the control algorithms to constant and time-varying communication delays is also studied. In addition to time delays, the number of communication links, which prevents scalability of networked robotic systems, is another challenging issue. Thus, a synchronizing control with practically feasible constraints of network topology is developed.

The problem of networked robotic systems interacting with human operators is then studied subsequently. This research investigates a teleoperation system with heterogeneous robots under asymmetric and unknown communication delays. Sub-task controllers are proposed for redundant slave robot to autonomously achieve additional tasks, such as singularity avoidance, joint angle limits, and collision avoidance. The developed control algorithms can enhance the efficiency of teleoperation systems, thereby ameliorating the performance degradation due to cognitive limitations of human operator and incomplete information about the environment.

Compared to traditional robotic systems, control of robotic manipulators over networks has significant advantages; for example, increased flexibility and ease of maintenance. With the utilization of scattering variables, this research demonstrates that transmitting scattering variables over delayed communications can stabilize an otherwise unstable system. An architecture utilizing delayed position feedback in conjunction with scattering variables is developed for the case of time-varying communication delays. The proposed control architecture improves tracking performance and stabilizes robotic manipulators with input-output communication delays. The aforementioned control algorithms and architectures for networked robotic systems are validated via numerical examples and experiments.