An adaptive learning control approach is proposed which combines a mechanism to improve the control input sequence as well as to improve the learning control scheme based on the knowledge learned about the unknown sys...
详细信息
An adaptive learning control approach is proposed which combines a mechanism to improve the control input sequence as well as to improve the learning control scheme based on the knowledge learned about the unknown system and environment. The iterative learning control problem is treated from the 2D system point of view. A 2D model for a class of iterative learning control system is formulated. A learning gain estimator algorithm based on the 2D model is presented. The overall learning control system structure is given. The proposed learning control scheme does not require prior knowledge of the controlled system and has the ability to generalize the knowledge learned from one task operation to other tasks. This scheme can be applied to nonlinear system control problems. To demonstrate the feasibility of the proposed learning algorithm, simulation results on learning control for a three-water-tank system are given. The results show an excellent learning performance, even for nonrepetitive tasks.< >
An adaptive learning control approach is proposed that combines a mechanism to improve control input sequence as well as to improve the learning control scheme based on the knowledge learned about unknown system and e...
详细信息
An adaptive learning control approach is proposed that combines a mechanism to improve control input sequence as well as to improve the learning control scheme based on the knowledge learned about unknown system and environment. First, the iterative learning control problem is treated from the 2-D system point of view. A 2-D model for a class of iterative learning control systems is formulated. Then a learning gain estimator algorithm based on the 2-D model is presented. The overall learning control system structure is given.< >
All linear multirate controllers for a given multirate sampled-data system plant are parameterized in terms of a single parameter Q(z) that is allowed to be any stable transfer matrix provided that certain causality c...
详细信息
All linear multirate controllers for a given multirate sampled-data system plant are parameterized in terms of a single parameter Q(z) that is allowed to be any stable transfer matrix provided that certain causality conditions are met. The result parallels a well-known parameterization result for single-rate sampled-data plants. The parameterization suggests architectures for multirate controllers.< >
A new robust technique for robot control is presented which is based on a combination of the feedback computed torque (FCT) method with the model reference adaptive control (MRAC) algorithm. It is shown that perfect t...
详细信息
A new robust technique for robot control is presented which is based on a combination of the feedback computed torque (FCT) method with the model reference adaptive control (MRAC) algorithm. It is shown that perfect tracking is feasible despite the errors induced by a practical FCT controller based on an “imprecise” or “simplified” available robot model. The MRAC part of the controller is simpler than the ones customarily used for robot control, and only a minimum amount of information is required. The results are illustrated by a double link manipulator computational example.
An approach to intelligent PID (proportional integral derivative) control of industrial systems which is based on the application of fuzzy logic is presented. This approach assumes that one has available nominal contr...
详细信息
An approach to intelligent PID (proportional integral derivative) control of industrial systems which is based on the application of fuzzy logic is presented. This approach assumes that one has available nominal controller parameter settings through some classical tuning technique (Ziegler-Nichols, Kalman, etc.). By using an appropriate fuzzy matrix (similar to Macvicar-Whelan matrix), it is possible to determine small changes on these values during the system operation, and these lead to improved performance of the transient and steady behavior of the closed-loop system. This is achieved at the expense of some small extra computational effort, which can be very easily undertaken by a microprocessor. Several experimental results illustrate the improvements achieved.< >
Some important dynamic properties of manipulators with flexible joints and links are presented. Partial feedback linearization, the zero dynamics of flexible-link manipulators, and a class of multibody flexible struct...
Some important dynamic properties of manipulators with flexible joints and links are presented. Partial feedback linearization, the zero dynamics of flexible-link manipulators, and a class of multibody flexible structures are considered. It is shown that the nonlinear systems in this class are non-minimum-phase when sensing and actuation are noncollocated. Only link flexibility is included. However, analogous results can be obtained if the effects of joint flexibility are considered
It is shown that on the set of w-input, p-output asymptotically stable, minimal, nth order systems that the H∞p·mand the RNquoiient induced topologies coincide. Implications of this in frequency domain system id...
详细信息
It is shown that on the set of w-input, p-output asymptotically stable, minimal, nth order systems that the H∞p·mand the RNquoiient induced topologies coincide. Implications of this in frequency domain system identification are discussed
The paper considers the problem of using reduced-order models for robot control. A feedback linearizing and decoupling transformation (FLDT) is first developed and applied to the full-order robot model. Then it is sho...
详细信息
This paper describes the realization of an expert system prototype of a robot task planning system used for automatic assembly of simple structures. The system involves a module which learns and identifies physical de...
详细信息
This paper describes the realization of an expert system prototype of a robot task planning system used for automatic assembly of simple structures. The system involves a module which learns and identifies physical descriptions from functional definitions in order to help the user to specify the tasks s(he) wants to achieve. The output of the system is a sequence of main goals which are decomposed in a set of elementary manipulation actions, and ensure the satisfaction of the assembly task specifications. Two other supplementary modules are also included; the first for describing the items to be assembled only by their functional characteristics, and the second for dealing with space problems, such as the problem of finding free place for putting an item out of the way. The system is written in C-Prolog of a VAX 11/750, and has been tested in several cases
暂无评论