Young-Kiu Choi

An adaptive neurocontroller using RBFN for robot manipulators

In recent years, neural networks have fulfilled the promise of providing model-free learning controllers for nonlinear systems; however, it is very difficult to guarantee the stability and robustness of neural network control systems. This paper proposes an adaptive neurocontroller for robot manipulators based on the radial basis function network (RBFN). The RBFN is a branch of neural networks and is mathematically tractable. Therefore, we adopt the RBFN to approximate nonlinear robot dynamics. The RBFN generates control input signals based on the Lyapunov stability that is often used in the conventional control schemes. A saturation function is also chosen as an auxiliary controller to guarantee the stability and robustness of the control system under the external disturbances and modeling uncertainties.

Design and implementation of an adaptive neural-network compensator for control systems

Recently, many studies have been made for intelligent controls using the neural-network (NN). These NN approaches for control strategies are based on the concept of replacing the conventional controller with a new NN controller. However, it is usually difficult and unreliable to replace the factory-installed controller with another controller in the workplace. In this case, it is desirable to install an additional outer control loop around the conventional control system to compensate for the control error of the preinstalled conventional control system. This paper presents an adaptive NN compensator for the outer loop to compensate for the control errors of conventional control systems. The proposed adaptive NN compensator generates a new command signal to the conventional control system using the control error that is the difference between the desired reference input and the actual system response. The proposed NN-compensated control system is adaptable to the environment changes and is more robust than the conventional control systems. Experimental results for a SCARA-type manipulator show that the proposed adaptive NN compensator enables the conventional control system to have precise control performance.

Similarity measure construction using fuzzy entropy and distance measure

The similarity measure is derived using fuzzy entropy and distance measure. By the relations of fuzzy entropy, distance measure, and similarity measure, we first obtain the fuzzy entropy. And with both fuzzy entropy and distance measure, similarity measure is obtained. We verify that the proposed measure become the similarity measure.

An on-line PID control scheme for unknown nonlinear dynamic systems using evolution strategy

The paper presents an on-line PID control scheme with varying gains for unknown nonlinear dynamic systems. For the on-line control, it is necessary to have an on-line identifier of the system, so that an identifier is constructed in the form of an autoregressive moving average (ARMA) model. In order to tune the parameters of the identifier and the gains of the PID controller efficiently, we propose a modified evolution strategy. Experimental studies show that the proposed on-line control scheme has robust control performance under unknown disturbance and noise.

Optimal trajectory planning and sliding mode control for robots using evolution strategy

Although robots have some kinematic and dynamic constraints such as the limits of the position, velocity, acceleration, jerk, and torque, they should move as fast as possible to increase the productivity. Researches on the minimum-time trajectory planning and control based on the dynamic constraints assume the availability of full dynamics of robots. However, the dynamic equation of robot may not often be exactly known. In this case, the kinematic approach for the minimum-time trajectory planning is more meaningful. We also have to construct a controller to track precisely the minimum-time trajectory. But, finding a proper controller is also difficult if we do not know the explicit dynamic equations of a robot.This paper describes an optimization of trajectory planning based on a kinematic approach using the evolution strategy (ES), as well as an optimization of a sliding mode tracking controller using ES for a robot without dynamic equations.

An adaptive control method for robot manipulators using radial basis function networks

The neural network known as a sort of intelligent control strategy is used as a powerful tool of control systems since it has learning ability. But it is difficult for neural network controllers to guarantee the stability of control systems. In this paper we try connecting a radial basis function network to an adaptive control strategy. Radial basis function networks are simpler and easier to handle than multilayer perceptrons. We use the radial basis function network to generate control input signals that are similar to the control inputs of adaptive control using liner reparameterization of the robot manipulator. We adopt the signum function as an auxiliary controller. This paper also proves mathematically the stability of the control system under the existence of disturbances and modeling errors.

Decentralized H ∞ Control of Maglev Systems

This paper presents a decentralized H_∞ controller design for controlled-permanent-magnet systems with multiple magnets. The structure of the decentralized H_∞ controller is simple and the computational burden is less than that of the centralized H_∞ controller. The decentralized H_∞ controller uses two Riccati equations and has an iterative form. The controlled-permanent-magnet system with two magnets is mathematically modelled and an experimental vehicle has been built to study the performance of a multi-magnet system. Some control problems in magnetic levitation systems such as the reference input tracking and the disturbance rejection are solved by adopting the H_∞ controller. Also, the possibility to apply the presented decentralized H_∞ controller to full-scale magnetic levitation systems is verified by real experiments

On development of stroke sensing cylinder for automatic excavator

We developed a part of a stroke sensing cylinder and its measurement system for an automatic excavator. In this paper, for the stroke sensing cylinder, we developed a 2-axis control instrument system with a magnetic sensor. The performance of a cylinder rod of an instrument system is achieved by a sliding mode control which is a new method diminishing the chattering in that control by setting 2-boundary layer along the switching line. The unknown parameters for sliding mode control are estimated by the signal compression method.

Design of a pitch controller using disturbance accommodating control for wind turbines under stochastic environments

This paper describes a design of a wind turbine pitch controller based on the disturbance accommodating control (DAC) theory. Wind turbine systems generally operate under stochastic environments, such as random wind inputs and noise corrupted sensor signals. Especially wind inputs can be treated as persistent disturbances and DAC can play a role in reducing effects of wind disturbances. By doing this, wind turbines suffer less from mechanical fatigue loads and their lifespan can be increased. Because wind disturbances we must accommodate are stochastic, we design DAC for stochastic plants and compare with other controllers not considering stochastic conditions to verify the performances of our proposed controller.

Trajectory optimization and control for robot manipulator using evolution strategy and fuzzy logic

Robot manipulators have some physical constraints such as the limits of the position, velocity, acceleration and jerk. In order to increase the productivity, it is desirable to drive the manipulators as fast as possible. The problem can be formulated as the time optimal control problem under the given physical constraints. But it is very difficult to obtain the exact solution of the time optimal control problem. The paper solves this problem in two steps. In the first step, we find the minimum time trajectories by optimizing cubic polynomial joint trajectories under the physical constraints using the modified evolution strategy. In the second step, the fuzzy controller is optimized for robot manipulator to track precisely the optimized trajectory found in the previous step. Experimental results for SCARA type manipulator show that the proposed method is very useful.