A learning control scheme based on neural networks for repeatable robot trajectory tracking

Published on Jan 1, 1999
· DOI :10.1109/ISIC.1999.796638
Jizhong Xiao1
Estimated H-index: 1
Qing Song17
Estimated H-index: 17
Danwei Wang65
Estimated H-index: 65
This paper presents an iterative learning controller using neural network (NN) for the robot trajectory tracking control. The basic control configuration is briefly presented and a new weight-tuning algorithm of NN is proposed with a dead-zone technique. Theoretical proof is given which shows that our modified algorithm guarantees the convergence of NN estimation error in the presence of disturbance. The simulation study demonstrates that the proposed weight-tuning algorithm is robust and less sensitive to noise compared to the standard backpropagation algorithm in identifying the robot inverse dynamics. Moreover, the simulation results also shows that the proposed NN learning control scheme can greatly reduce tracking errors as the iteration number increases.
📖 Papers frequently viewed together
3 Authors (A. Alessandri, ..., M. Maggiore)
3 Citations
2,391 CitationsSource
A family of two-layer discrete-time neural net (NN) controllers is presented for the control of a class of mnth-order MIMO dynamical system. No initial learning phase is needed so that the control action is immediate; in other words, the neural network (NN) controller exhibits a learning-while-functioning-feature instead of a learning-then-control feature. A two-layer NN is used which is linear in the tunable weights. The structure of the neural net controller is derived using a filtered error a...
117 CitationsSource
The incorporation of dead zones in the error signal of basis function networks avoids the networks' overtraining and guarantees the convergence of the normalized least mean square (LMS) algorithm and related algorithms. A new so-called error-minimizing dead zone is presented providing the least a posteriori error out of the set of all convergence assuring dead zones. A general convergence proof is developed for LMS algorithms with dead zones, and the error-minimizing dead zone is derived from th...
24 CitationsSource
#1Frank L. LewisH-Index: 123
#2Aydin YesildirekH-Index: 12
Last. Kai LiuH-Index: 1
view all 3 authors...
A multilayer neural-net (NN) controller for a general serial-link rigid robot arm is developed. The structure of the NN controller is derived using a filtered error/passivity approach. No off-line learning phase is needed for the proposed NN controller and the weights are easily initialized. The nonlinear nature of the NN, plus NN functional reconstruction inaccuracies and robot disturbances, mean that the standard delta rule using backpropagation tuning does not suffice for closed-loop dynamic ...
957 CitationsSource
#1Seul Jung (UC Davis: University of California, Davis)H-Index: 23
#2T.C. Hsia (UC Davis: University of California, Davis)H-Index: 24
The basic robot control technique is the model based computer-torque control which is known to suffer performance degradation due to model uncertainties. Adding a neural network (NN) controller in the control system is one effective way to compensate for the ill effects of these uncertainties. In this paper a systematic study of NN controller for a robot manipulator under a unified computed-torque control framework is presented. Both feedforward and feedback NN control schemes are studied and co...
27 CitationsSource
#1Fu-Chuang Chen (NCTU: National Chiao Tung University)H-Index: 15
#2Hassan K. KhalilH-Index: 63
Layered neural networks are used in a nonlinear self-tuning adaptive control problem. The plant is an unknown feedback-linearizable discrete-time system, represented by an input-output model. To derive the linearizing-stabilizing feedback control, a (possibly nonminimal) state-space model of the plant is obtained. This model is used to define the zero dynamics, which are assumed to be stable, i.e., the system is assumed to be minimum phase. A linearizing feedback control is derived in terms of s...
501 CitationsSource
#1Wei Wei (ZJU: Zhejiang University)H-Index: 1
#2J.P. Jiang (ZJU: Zhejiang University)H-Index: 1
This paper presents a new learning control method for robot trajectory tracking of unknown model system based on both the inverse dynamics neural network identifier and feedforward compensation network controller. Basic control configurations are briefly presented and new online training schemes are proposed. Some measurements to accelerate the learning and to improve the stability and convergence are discussed. It is also shown that the method can follow and arbitrarily prescribe trajectories. ...
3 CitationsSource
#1Jianguo Fu (McMaster University)H-Index: 1
#2Naresh K. Sinha (McMaster University)H-Index: 23
In this paper, an iterative learning controller using neural networks has been studied for the motion control of robotic manipulators. Simulations of a two-link robot have demonstrated that the proposed control scheme for robotic manipulators can greatly reduce tracking errors after a few trials. Our modification of the original back-propagation algorithm is employed in the neural network, resulting in a much faster learning rate. The results of simulation have also shown that the proposed itera...
20 CitationsSource
#1Frank L. LewisH-Index: 123
#2Darren M. DawsonH-Index: 76
Last. Chaouki T. AbdallahH-Index: 32
view all 3 authors...
Control of robot manipulators , Control of robot manipulators , مرکز فناوری اطلاعات و اطلاع رسانی کشاورزی
943 Citations
#1Marios M. Polycarpou (SC: University of Southern California)H-Index: 67
#2Petros Ioannou (SC: University of Southern California)H-Index: 74
A class of feedforward neural networks, structured networks, has recently been introduced as a method for solving matrix algebra problems in an inherently parallel formulation. A convergence analysis for the training of structured networks is presented. Since the learning techniques used in structured networks are also employed in the training of neural networks, the issue of convergence is discussed not only from a numerical algebra perspective but also as a means of deriving insight into conne...
109 CitationsSource
Cited By5
#1Hyo-Sung Ahn (Gwangju Institute of Science and Technology)H-Index: 39
#2YangQuan ChenH-Index: 94
Last. Kevin L. MooreH-Index: 53
view all 3 authors...
In this paper, the iterative learning control (ILC) literature published between 1998 and 2004 is categorized and discussed, extending the earlier reviews presented by two of the authors. The papers includes a general introduction to ILC and a technical description of the methodology. The selected results are reviewed, and the ILC literature is categorized into subcategories within the broader division of application-focused and theory-focused results.
1,129 CitationsSource
#1F. Jatta (National Research Council)H-Index: 2
#2Giovanni LegnaniH-Index: 20
Last. Antonio VisioliH-Index: 36
view all 3 authors...
This paper deals with the implementation of a hybrid force/velocity controller for the automatic edge following of two-dimensional unknown planar contours performed by an industrial robot manipulator. In particular, the authors address the problem of compensating the joint friction effects that have to be taken into account in the controller design in order to achieve a reasonable performance with regards to normal force and tangential velocity errors. For that reason, two model-based friction-c...
58 CitationsSource
#1Kok Kiong Tan (NUS: National University of Singapore)H-Index: 43
#2Shao Zhao (NUS: National University of Singapore)H-Index: 10
Last. Sunan Huang (NUS: National University of Singapore)H-Index: 34
view all 3 authors...
A learning control scheme is proposed which is suitable for high-precision and repetitive motion control applications. It comprises of a self-tuning radial basis function (RBF) network operating in parallel with an iterative learning control (ILC) component. Unlike the usual ILC scheme which adapts a feedforward control signal to achieve improved tracking performance over time, the proposed scheme iteratively adjusts the reference signal. The RBF network is employed as a nonlinear function estim...
26 CitationsSource
Aug 19, 2004 in ISNN (International Symposium on Neural Networks)
#1Gaoxiang Ouyang (Yanshan University)H-Index: 23
#2Xiaoli Li (Yanshan University)H-Index: 75
Last. Ruxu Du (CUHK: The Chinese University of Hong Kong)H-Index: 31
view all 6 authors...
In plastic injection molding, the ram velocity plays an important role in production quality. This paper introduces a new method, which is a combination of the current cycle feedback control and neural network (NN) learning, to control the ram velocity in injection process. It consists of two parts: a PD controller (current cycle feedback control) is used to stabilize the system, and the feedforward NN learning is used to compensate for nonlinear/unknown dynamics and disturbances, thereby enhanc...
1 CitationsSource
#1Kok Kiong Tan (NUS: National University of Singapore)H-Index: 43
#2Shao Zhao (NUS: National University of Singapore)H-Index: 10
In this paper, a new iterative learning control (ILC) scheme is proposed which is suitable for high precision and repetitive motion control applications. Unlike the usual ILC scheme which adapts a feedforward control signal to achieve improved tracking performance over time, the proposed scheme iteratively adjusts the reference signal. To achieve a higher convergence rate, a radial basis function neural network is employed to model the tracking error over a cycle, and subsequently used implicitl...
2 CitationsSource