Model-based, Distributed, and Cooperative Control of Planar Serial-link Manipulators
- PDF / 1,380,791 Bytes
- 14 Pages / 594.77 x 793.026 pts Page_size
- 34 Downloads / 210 Views
ISSN:1598-6446 eISSN:2005-4092 http://www.springer.com/12555
Model-based, Distributed, and Cooperative Control of Planar Serial-link Manipulators S. Soumya and K. R. Guruprasad* Abstract: In this paper, we propose a novel distributed control scheme for a planar serial-link manipulator with revolute joints. The control scheme is based on the conventional model-based nonlinear control scheme that achieves linearization by feedback. A dedicated controller controls each joint of the manipulator, as in the case of the decentralized manipulator control scheme. However, in the proposed control scheme, the joint-level controllers communicate and cooperate to account for the nonlinear dynamic coupling between the links. The proposed control scheme can achieve the performance level of that of the model-based nonlinear control scheme, and at the same time, reduce the computational lead-time by distributing the computational load associated with the control law among the joint-level controllers. We design a distributed cooperative control law for a three-link planar manipulator and demonstrate its trajectory tracking performance using simulation experiments. Keywords: Distributed cooperative control, joint-level controller, manipulator dynamics, multi-agent systems.
1.
INTRODUCTION
Motion control is one of the fundamental problems addressed in robotics. Though there has been a substantial advance in the field of control system design in general, and manipulator control in particular, owing to its coupled nonlinear nature of dynamics, and high computation cost involved in the computation of the dynamic equations, manipulator control is an active and challenging problem even to date. 1.1. Control architectures for networked systems Control of networked systems such as cooperative control for multiple manipulators [1, 2], or control of networked systems in an industrial manufacturing setup [3, 4], considers the problem of cooperative and distributed control of a network of robotic manipulators. In [5], a single manipulator, being a multi-input-multioutput (MIMO) system, was perceived as a networked or a multi-agent system. In the context of multi-agent or networked systems, the control architectures may be classified as centralized, decentralized, or distributed control schemes. These control architectures are illustrated in Fig. 1. In the centralized control architecture, a single central controller controls all the agents. In the decentralized architecture, each agent (or subsystem) is controlled by an individual controller. There is no interaction
between these controllers, though the agents may interact with each other. One of the major advantages of this architecture over the centralized architecture is reduced computational load as the multiple agent-level controllers share this load. However, as the individual controllers do not communicate among themselves, the coupling between the agents is not considered. In a distributed architecture, the agent-level controllers cooperate by communicating among themselves, thus ac
Data Loading...
