NEUROFUZZY CONTROL STRUCTURE FOR  A ROBOT MANIPULATOR

Most nonlinear control concepts used in robotics are based on a more or less accurate inverse model of the robot. In contrast to this, the design and properties of a general $n$-loop control structure based on a divided forward model of the robot, the so-called multi-loop Model Following Control Structure ($n$-MFC), is presented in this paper. Its theoretical basics and its concept are explained. The stability and robustness of the proposed control structure is analyzed. The theoretical assumptions are verified in many experiments with a two-joint robot manipulator. Qualitative as well as quantitative results of the experiments are presented and discussed.

Download Full-text

Error attenuation in the control of a parallel robot manipulator using a dual-model-based structure

Proceedings of the Institution of Mechanical Engineers Part C Journal of Mechanical Engineering Science ◽

10.1243/095440603762826486 ◽

2003 ◽

Vol 217 (2) ◽

pp. 161-171 ◽

Cited By ~ 4

Author(s):

Q Li

Keyword(s):

Control Structure ◽

Robot Manipulator ◽

Parallel Robot ◽

Parallel Robots ◽

Internal Model Control ◽

Sensitivity Analyses ◽

Parallel Structure ◽

Dual Model ◽

Model Based ◽

Model Control

Parallel structure robots have been receiving growing attention from both academia and industry in recent years. This is due to their advantages over serial structure robots, such as high stiffness, high motion accuracy and a high load-structure ratio. Control of parallel robots, however, produces difficulties to control engineers due to the modelling errors arising from the highly non-linear and complex structures. This paper proposes a dual-model-based structure for error attenuation in the trajectory-tracking control of a parallel robot manipulator. In this design, a conventional model-based control algorithm employing an estimated robot dynamic model is first implemented in the inner loop of the control structure. Then, in order to reduce the unwanted effects caused by modelling erros, another model-based structure, developed based on the concept of the internal model control, is appended in the outer loop of the control structure as a compensator. A combination of these two model-based components results in a novel dual-model-based structure for parallel robot control. Sensitivity analyses show that the effects due to modelling errors and external disturbances can be significantly reduced by applying this new control structure without relying on a high-gain control solution. The effectiveness of this control design is successfully demonstrated by numerical studies on a planar parallel robot with 2 degrees of freedom.

Download Full-text

Adaptive Nonlinear PD Learning Control for Robot Manipulators

Volume 2: 28th Biennial Mechanisms and Robotics Conference, Parts A and B ◽

10.1115/detc2004-57131 ◽

2004 ◽

Cited By ~ 2

Author(s):

P. R. Ouyang ◽

W. J. Zhang ◽

M. M. Gupta

Keyword(s):

Control Structure ◽

Control Method ◽

Robot Manipulator ◽

Robot Manipulators ◽

Learning Control ◽

Asymptotic Convergence ◽

Simulation Studies ◽

Pd Control ◽

On Line ◽

On Line Learning

A new control method, called adaptive nonlinear PD learning control (NPD-LC), is proposed for robot manipulator applications in this paper. The proposed control structure is a combination of a nonlinear PD control structure and a directly learning structure. Consequently, this new control method possesses both adaptive and on-line learning properties. One of the unique features of the NPD-LC algorithm is that the learning is based on the previous torque profile of the repetitive task. It is proved that the NPD-LC enjoys the asymptotic convergence for both tracking positions and tracking velocities. Simulation studies were conducted by comparing the proposed method with many other existing methods. As a result, it was demonstrated that the NPD-LC method can achieve a faster convergence speed. The proposed NPD-LC is robust and can be implemented for the control of robot manipulators.

Download Full-text