Repetitive Learning Research Articles

Anesthetic ketamine counteracts repetitive mechanical stress-induced learning and memory impairment in developing mice

The aim of this study is to investigate whether ketamine, a noncompetitive N-methyl-D: -aspartate receptor (NMDAR) antagonist, had an influence on learning and memory in developing mice. Fifty Kunming mice aged 21 days were randomly divided into 5 subgroups (n = 10 for each) to receive intraperitoneal injection of equal volume of saline (S group) or ketamine (25, 50 or 100 mg/kg of body weight/day) for 7 consecutive days, or to be left untreated (C group). A step-down passive avoidance test was performed to evaluate learning and memory in these mice on days 8 and 9. Additionally, the expression of brain-derived neurotrophic factor (BDNF) in the hippocampus was determined. Rats receiving saline or sub-anesthetic dose of ketamine (25 mg/kg) showed significantly decreased abilities of learning and memory and reduced expression of BDNF, compared to the normal controls (P < 0.05). In contrast, comparable abilities of learning and memory and expression of BDNF were found for anesthetic doses of ketamine (50 or 100 mg/kg)-treated rats and controls (P > 0.05). Repetitive mechanical stress impairs learning and memory performance in developing mice, which may be associated with decreased BDNF expression. The stress-induced learning and memory impairment can be prevented by anesthetic doses of ketamine.

Repetitive Learning Control of a Flexible Whisker in Tip Contact with an Unknown Surface

Repetitive contact imaging uses a flexible whisker attached to a two-axis robot through a load cell. Assuming small deformations and rotations, the pitch axis decouples from the yaw. The yaw axis, under proportional—integral—derivative control, sweeps periodically back and forth across the object while the pitch axis, under repetitive learning (RL) control, maintains a uniform contact force. Once the RL controller converges, the three-dimensional contact points can be determined using an elastica algorithm. The RL controller is proven stable based on a distributed parameter beam model and is experimentally shown to provide a stable performance with improved moment regulation when compared with under proportional-derivative control.

Universal Repetitive Learning Control for Nonparametric Uncertainty and Unknown State-Dependent Control Direction Matrix

We propose a continuous universal repetitive learning control to track periodic trajectory for a class of nonlinear dynamical systems with nonparametric uncertainty and unknown state-dependent control direction matrix. The proposed controller is an integration of high-gain feedback, repetitive learning and Nussbaum gain matrix selector. The control signal is always continuous, thus it avoids the potential chattering effect caused by discontinuity. Asymptotic convergence of the tracking error is achieved by the controller, and the control performance is illustrated by simulation. Although the proposed method is derived for input-state systems, it can be readily extended to multi-input-multi-output systems under appropriate assumption.

A structured matrix approach to efficient calculation of LQG repetitive learning controllers in the lifted setting

This article discusses the use of repetitive control for output reference tracking in linear time-varying discrete time systems with both repetitive and non-repetitive noise components. The design of such controllers is formulated as a lifted linear stochastic output feedback problem on which the mature techniques of discrete linear control may be applied. In many modern applications, the large size of the system matrices in such a control problem inhibits the application of standard solvers and optimisation techniques. For linear quadratic Gaussian (LQG) problems, the matrices of the lifted feedback problem can be fitted into the recently developed sequentially semi-separable structure. Innovative numerical solutions are developed that have 𝒪(N) computational complexity (where N is the trial length) in both controller synthesis and implementation, comparable to that of many non-lifted and Fourier transform based learning control methods. Moreover, within this formulation, the system is allowed to vary over the learning cycle, closed-loop stability is guaranteed, and stochastic noise and disturbances are handled in an LQG sense.

Repetitive learning control of nonlinear systems over finite intervals

Iterative learning control requires initial repositioning, while the time functions to be learned should be of periodicity in repetitive control. However, there are cases in practice where the time-varying unknowns are not periodic but repetitive, and repetitive learning control is applicable with avoidance of initial repositioning. In this paper, repetitive learning control designs are presented for a broader class of nonlinear systems over finite intervals. The Freeman formula is modified and used for stabilization of the nominal nonlinear time-varying system undertaken. The global stability of the learning system and asymptotic convergence of the tracking error are established through analysis of both partially and fully saturated learning algorithms, respectively. The repetitive learning control method is theoretically shown to be effective in dealing with time-varying parametric uncertainties.

Implementation of frequency-shaped tip reference in conjunction with learning controller for improved tip-tracking control

In the flexible manipulator control, tip-tracking control of flexible manipulator results in non-colocated control problem, which has a non-minimum phase dynamic characteristic. The level of tip-tracking performance in the non-colocated control system depends on the characteristics of the tip reference trajectory to be followed, as well as on the characteristics of the flexible manipulator system itself. In a previous research the use of a tip reference trajectory, filtered by a properly designed time-delay command shaping filter, has been proposed and a multirate repetitive learning control (MRLC) has been used as the tip-tracking controller. The practical implementation of this approach, however, requires estimation of the tip position, which is not easy to obtain. In this paper, a practical implementation of the approach is considered and the tip position is estimated with a fourth-order Kalman filter. The experimental results show that, with the use of Kalman filter, the proposed scheme results in a drastic reduction in residual tip vibrations and the required actuation effort.

The Role of Therapeutic Instrumental Music Performance in Hemiparetic Arm Rehabilitation

ABSTRACT: This study describes the clinical application of Therapeutic Instrumental Music Performance (TIMP) on functional upper extremity movement in stroke patients. Participants for this study included 3 post-stroke outpatients showing pronounced hemiparesis in their upper extremities. Participants underwent six, 35-minute music therapy sessions in which they received training in flexion and extension exercises of their paretic arm using a Roland TD-5 Percussion Sound Module Musical Instrument Digital Interface (MIDI) drum set for TIMP exercises. Functional arm movements were assessed in the pretest and posttest using the Barthel Index, the Fugl-Meyer Test, and the Modified Ashworth Scale. MIDI data on total movement time, variability of movement time, and force were also collected during the pretest and posttest as well as at the halfway point of the treatment program. Results indicated significant improvements on the wrist and hand portion of the Fugl-Meyer Test (t = 5, p = .0377), and MIDI data showed observable improvements throughout the treatment period. These findings suggest that the addition of Neurologic Music Therapy to current rehabilitation training programs is possible and may prove beneficial in rehabilitation practices addressing functional arm movements in stroke patients. Stroke is a major leading cause of disabilities and death in the United States and it is estimated that about 700,000 new or recurrent strokes occur each year (American Heart Association, 2005; Truelsen et al., 2003; U.S. Department of Health and Human Resources, 1996). Forty-three percent of stroke patients are left with moderate to severe neurological functional impairments and 50% with hemiparesis (Hurst, 1994; Kelly-Hayes et al., 2003). About 75% of stroke patients have upper extremity weakness accompanied by muscle tone (spasticity), decreased muscle strength, decreased range of motion (ROM), and decreased ability to maintain posture and control of movements (Charness, 1995; Dewald, Sheshadri, Dawson, & Beer, 2001; Lawrence et al., 2001). A classic study by Twitchell (1951) pointed out the severity of upper extremity weakness at the onset of a stroke as an important predictor in upper extremity motor recovery following a stroke. In typical motor recovery, the upper extremity is more involved than the lower extremity at the onset and eventually this reverses. Even though most recovery takes place in the first three months, recovery may continue over a longer period of time (eMedicine, 2004). In a study conducted by Broeks, Lankhorst, Rumping, and Prevo (1999), patients showed good recovery of arm functions 16 weeks post-stroke. They also found that even patients four years post-stroke still showed good arm function recovery; therefore, continued outpatient therapy is very beneficial for recovering stroke patients. Because there are parallel neural processes in motor and auditory temporal processing which allows them to be coupled, the motor system is very responsive to the auditory system (Thaut, Kenyon, Schauer, & Mcintosh, 1999). Since temporal information of rhythms and timing information of motor movements have similar neural pathways, they also have parallel neural impulses; thus, motor responses can be entrained (synchronized) to auditory rhythms, making the movement more stable and steady (Thaut, 2003). Because the ability to respond to a rhythm and to discriminate between rhythms is innate (Cassidy & Standley, 1995; Standley, 2003; Trehub, 1 993; Trehub & Thorpe, 1989), it has been speculated that internal auditory rhythm can be a good temporal reference used to produce the complex timing and skilled movement of a motor response (Shaffer, 1982). This innate ability to entrain motor movement to an external rhythm can be effectively used for specific therapeutic goals (Thaut, 2003). Repetitive motor learning and structured training have been generally known to improve motor deficits in stroke rehabilitation (Butefisch, Hummelsheim, Denzler, & Mauritz, 1995; Platz, Bock, & Prass, 2001) through cortical rearrangement (brain plasticity). …

Sliding-mode repetitive learning control with integral sliding-mode perturbation compensation

This paper proposes a sliding-mode repetitive learning control (SMRLC) scheme with an integral sliding-mode perturbation observer (ISMPO) for repetitive tracking control tasks. The three control strategies which are synthesized to yield excellent tracking performance are: (1) the pole-placement feedback control to specify the desired error dynamics; (2) ISMPO-based feedback compensation as the robust part; and (3) a feedforward learning component that refines the control to improve system performance through repetitive trials. The ISMPO-based feedback compensation ensures that there is only small tracking error during initial learning trials and enhances system insensitivity to exceptional and aperiodic disturbances. The feedforward learning compensation is updated according to a certain switching signal that is equivalent to the compensation error of the feedforward control, yielding fast convergence of the learning process from trial to trial. Experimental results demonstrate the feasibility of the proposed scheme.

Structure adaptation of hierarchical knowledge-based classifiers

This paper introduces a new method to identify the qualified rule-relevant nodes to construct hierarchical neuro-fuzzy systems (HNFSs). After learning, the proposed method analyzes the entire history of activities and behaviors of all rule nodes, which reflects their levels of involvement or contribution during the process. The less qualified rule-relevant nodes can then be identified and removed, reducing the size and complexity of the HNFS. Upon the repetitive learning process, the method may be repetitively applied until a satisfactory result is obtained, simultaneously improving the performance and reducing the size and complexity. Incorporated with the method is a new HNFS architecture which addresses both the scalability problem experienced in rule based systems and the restriction of the “overcrowded defuzzification” problem found in hierarchical designs. In order to verify the performance, the proposed method has been successfully tested against five well-known classification problems whose results are provided and then discussed in the concluding remarks.

DC Link Voltage and Supply-Side Current HarmonicsMinimization of Three Phase PWM BoostRectifiers Using Frequency Domain BasedRepetitive Current Controllers

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> This paper presents a digital plug-in frequency domain based repetitive control scheme for minimizing the odd order harmonics in the supply line side currents of the three phase pulsewidth modulation (PWM) boost rectifier under the distorted and unbalanced supply voltage conditions. Based on the mathematical model of the three-phase PWM boost rectifier under the generalized supply voltage conditions, the control task is divided into: (a) dc-link voltage harmonics control and (b)supply line side current harmonics control. The proposed plug-in repetitive controller together with the conventional PI controller is designed to achieve supply line side currents with low total harmonic distortion (THD) for the three phase PWM boost rectifier. The repetitive control learning algorithm is implemented in the frequency domain by means of Fourier series approximation, instead of commonly used time domain based scheme, and provides the flexibility of choosing different learning gains and phase angle delay compensations individually for each harmonic component. The experimental test results obtained from a 1.6 kVA laboratory based PWM rectifier confirm that the THD of the supply line side currents can be reduced from 21.09% to 4.12% with the plug-in frequency domain based repetitive current controllers. </para>

Learning how to restructure: absorptive capacity and improvisational views of restructuring actions and performance

AbstractThis paper examines the role of learning in corporate restructuring. Drawing from two viewpoints of organizational learning, absorptive capacity and organizational improvisation, we examine whether experience with corporate restructuring modes (sell‐offs, spin‐offs) influences subsequent restructuring and financial performance. Consistent with an absorptive capacity view, cumulative and repetitive experience with sell‐offs was related to the adoption of an ensuing sell‐off and to higher performance. Conversely, and consistent with an organizational improvisation view, short‐term and contemporaneous experience with spin‐offs was related to the subsequent use of spin‐offs and to increases in financial performance. The findings contribute to a dynamic explanation of corporate restructuring and its influence on financial performance, illustrate differences between learning in a repetitive situation and learning when repetition is rare, and indicate when absorptive capacity and organizational improvisational views are most profitable. Overall, these findings show that different kinds of restructuring experiences were associated with different modes of restructuring and performance records. Considered collectively, the organizational learning perspective offers insights into why some corporate restructuring strategies appear as intentional and deliberate actions while others resemble more spontaneous and simultaneous responses. Copyright © 2008 John Wiley &amp; Sons, Ltd.

Repetitive and Iterative Learning Controllers Designed by Duality with Experimental Verification

A duality theory existing between iterative learning and repetitive control for linear time-invariant systems has been reported. This paper considers the application of this duality in the design of such controllers for a non-minimum phase experimental facility and a three-axis gantry robot, where the task performed can be configured in either mode of operation. The models used in the design work have been obtained using frequency domain tests conducted on the physical plants. The control design has been performed for both systems, and verified using simulation studies in the case of the gantry robot and experimentally obtained test results in the case of the non-minimum phase plant.

Frequency Domain Based Repetitive Control of Three Phase Boost PWM Rectifier Under Distorted Supply Voltage Conditions

This paper presents a digital plug-in frequency domain based repetitive control scheme for minimizing the even order harmonics at the output dc link voltage and odd order harmonics in the line currents under the distorted and unbalanced supply voltage conditions. The proposed current controller consists of a conventional PI controller and a frequency domain based plug-in repetitive controller. Based on the mathematical model of the three-phase PWM boost rectifier under the generalized supply voltage conditions, the control task is divided into: (a)dc-link voltage harmonics control and (b)line side current harmonics control. The plug-in repetitive controller is designed to achieve line side currents with low THD for the three phase PWM boost rectifier. The repetitive control learning algorithm is designed in frequency domain instead of commonly used time domain and provides the flexibility of choosing different learning gains and phase delay compensations individually for each harmonic component. The experimental test results obtained from a 1.6 kVA laboratory based PWM rectifier confirm that the THD of the line side current can be reduced from 21.09% to 4.12% with the insertion of the plug-in frequency domain based repetitive controller.

Incremental Nonnegative Matrix Factorization for Face Recognition

Nonnegative matrix factorization (NMF) is a promising approach for local feature extraction in face recognition tasks. However, there are two major drawbacks in almost all existing NMF‐based methods. One shortcoming is that the computational cost is expensive for large matrix decomposition. The other is that it must conduct repetitive learning, when the training samples or classes are updated. To overcome these two limitations, this paper proposes a novel incremental nonnegative matrix factorization (INMF) for face representation and recognition. The proposed INMF approach is based on a novel constraint criterion and our previous block strategy. It thus has some good properties, such as low computational complexity, sparse coefficient matrix. Also, the coefficient column vectors between different classes are orthogonal. In particular, it can be applied to incremental learning. Two face databases, namely FERET and CMU PIE face databases, are selected for evaluation. Compared with PCA and some state‐of‐the‐art NMF‐based methods, our INMF approach gives the best performance.

Repetitive Learning Control for Time-varying Robotic Systems: A Hybrid Learning Scheme

Repetitive learning control is presented for finite-time-trajectory tracking of uncertain time-varying robotic systems. A hybrid learning scheme is given to cope with the constant and time-varying unknowns in system dynamics, where the time functions are learned in an iterative learning way, without the aid of Taylor expression, while the conventional differential learning method is suggested for estimating the constant ones. It is distinct that the presented repetitive learning control avoids the requirement for initial repositioning at the beginning of each cycle, and the time-varying unknowns are not necessary to be periodic. It is shown that with the adoption of hybrid learning, the boundedness of state variables of the closed-loop system is guaranteed and the tracking error is ensured to converge to zero as iteration increases. The effectiveness of the proposed scheme is demonstrated through numerical simulation.

Repetitive Learning Control of Nonlinear Continuous-Time Systems Using Quasi-Sliding Mode

In this brief, a quasi-sliding mode (QSM)-based repetitive learning control (RLC) method is proposed for tackling multi-input multi-output nonlinear continuous-time systems with matching perturbations. The proposed RLC method is able to perform rejection of periodic exogenous disturbances as well as tracking of periodic reference trajectories. It ensures a robust system stability when it is subject to nonperiodic uncertainties and disturbances. In this brief, an application to a robotic manipulator is used to illustrate the performance of the proposed QSM-based RLC method. A comparative study with the conventional variable structure control (VSC) technique is also included

Predictor‐based repetitive learning control for a class of remote control nonlinear systems

AbstractIn this paper, a repetitive learning control (RLC) approach is proposed for a class of remote control nonlinear systems satisfying the global Lipschitz condition. The proposed approach is to deal with the remote tracking control problem when the environment is periodic or repeatable over infinite time domain. Since there exist time delays in the two transmission channels: from the controller to the actuator and from the sensor to the controller, tracking a desired trajectory through a remote controller is not an easy task. In order to solve the problem caused by time delays, a predictor is designed on the controller side to predict the future state of the nonlinear system based on the delayed measurements from the sensor. The convergence of the estimation error of the predictor is ensured. The gain design of the predictor applies linear matrix inequality (LMI) techniques developed by Lyapunov Kravoskii method for time delay systems. The RLC law is constructed based on the feedback error from the predicted state. The overall tracking error tends to zero asymptotically over iterations. The proof of the stability is based on a constructed Lyapunov function related to the Lyapunov Kravoskii functional used for the proof of the predictor's convergence. By well incorporating the predictor and the RLC controller, the system state tracks the desired trajectory independent of the influence of time delays. A numerical simulation example is shown to verify the effectiveness of the proposed approach. Copyright © 2007 John Wiley &amp; Sons, Ltd.

The repetitive learning method under time pressure in e-Learning system

The repetitive learning method under time pressure in e-Learning system

幾何学的拘束を受けるマニピュレータの学習適応制御

When an endpoint of a manipulator is moving in touch with a rigid smooth surface, the manipulator should be controlled to achieve both a desired position trajectory of the endpoint and a desired contact force at the contact point. In this paper, we deal with a hybrid controller of repetitive learning control and model-based adaptive control for geometrically constrained robot manipulator. The convergence of joint angular errors and contact force errors is proved under an appropriate initial condition and the smoothness of the constraint surface. Furthermore, by the simulation experiment, we show that the proposed hybrid controller makes 2-link constrained manipulator track a desired position and a desired constraint force trajectory

On Repetitive Learning Control for Periodic Tracking Tasks

In this note, a repetitive learning control (RLC) approach is proposed to deal with periodic tracking tasks for nonlinear dynamical systems with nonparametric uncertainties. We address two fundamental issues associated with the learning control methodology: The existence of the solution, and learning convergence property. Applying the existence theorem of the neutral differential difference equation, and using Lyapunov-Krasovskii functional, the existence of the solution and learning convergence can be proven rigorously. A further extension of the RLC to cascade systems is also explored