Gesture-based control and EMG decomposition

This paper presents two probabilistic developments for the use with electromyograms (EMGs). First described is a neuroelectric interface for virtual device control based on gesture recognition. The second development is a Bayesian method for decomposing EMGs into individual motor unit action potentials (MUAPs). This Bayesian decomposition method allows for distinguishing individual muscle groups with the goal of enhancing gesture recognition. All examples presented rely upon sampling EMG data from a subject's forearm. The gesture-based recognition uses pattern recognition software that has been trained to identify gestures from among a given set of gestures. The pattern recognition software consists of hidden Markov models, which are used to recognize the gestures as they are being performed in real time from moving averages of EMGs. Two experiments were conducted to examine the feasibility of this interface technology. The first replicated a virtual joystick interface, and the second replicated a keyboard. Moving averages of EMGs do not provide an easy distinction between fine muscle groups. To better distinguish between different fine motor skill muscle groups, we present a Bayesian algorithm to separate surface EMGs into representative MUAPs. The algorithm is based on differential variable component analysis, which was originally developed for electroencephalograms. The algorithm uses a simple forward model representing a mixture of MUAPs as seen across multiple channels. The parameters of this model are iteratively optimized for each component. Results are presented on both synthetic and experimental EMG data. The synthetic case has additive white noise and is compared with known components. The experimental EMG data were obtained using a custom linear electrode array designed for this study.     

A model of the magnetic fields created by single motor unitcompound action potentials in skeletal muscle

We have developed a computationally simple model for calculating the magnetic-field strength at a point due to a single motor unit compound action potential (SMUCAP). The motor unit is defined only in terms of its anatomical features, and the SMUCAP is approximated using the tripole model. The distributed current density J is calculated within the volume defined by the motor unit. The law of Biot and Savart can then be cast in a form necessitating that J be integrated only over the region containing current sources or conductivity boundaries. The magnetic-field strength is defined as the summation of the contributions to the field made by every muscle fiber in the motor unit. Applying this model to SMUCAP measurements obtained using a high-resolution SUper Conducting Quantum Interference Device (SQUID) magnetometer may yield information regarding the distribution of action currents (AC's) and the anatomical properties of single motor units within a muscle bundle     

A technique to track individual motor unit action potentials in surface EMG by monitoring their conduction velocities and amplitudes

The speed of propagation of an action potential along a muscle fiber, its conduction velocity (CV), can be used as an indication of the physiological or pathological state of the muscle fiber membrane. The motor unit action potential (MUAP), the waveform resulting from the spatial and temporal summation of the individual muscle fiber action potentials of that motor unit (MU), propagates with a speed referred to as the motor unit conduction velocity (MUCV). This paper introduces a new algorithm, the MU tracking algorithm, which estimates MUCVs and MUAP amplitudes for individual MUs in a localized MU population using SEMG signals. By tracking these values across time, the electrical activity of the localized MU pool can be monitored. An assessment of the performance of the algorithm has been achieved using simulated SEMG signals. It is concluded that this analysis technique enhances the suitability of SEMG for clinical applications and points toward a future of noninvasive diagnosis and assessment of neuromuscular disorders.     

Analysis and simulation of changes in EMG amplitude during high-level fatiguing contractions

Changes in surface electromyographic (EMG) amplitude during sustained, fatiguing contractions are commonly attributed to variations in muscle fiber conduction velocity (MFCV), motor unit firing rates, transmembrane action potentials and the synchronization or recruitment of motor units. However, the relative contribution of each factor remains unclear. Analytical relationships relating changes in MFCV and mean motor unit firing rates to the root mean square (RMS) and average rectified (AR) value of the surface EMG signal are derived. The relationships are then confirmed using model simulation. The simulations and analysis illustrate the different behaviors of the surface EMG RMS and AR value with changing MFCV and firing rate, as the level of motor unit superposition varies. Levels of firing rate modulation and short-term synchronization that, combined with variations in MFCV, could cause changes in EMG amplitude similar to those observed during sustained isometric contraction of the brachioradialis at 80% of maximum voluntary contraction were estimated. While it is not possible to draw conclusions about changes in neural control without further information about the underlying motor unit activation patterns, the examples presented illustrate how a combined analytical and simulation approach may provide insight into the manner in which different factors affect EMG amplitude during sustained isometric contractions.     

Novel ideas for fast muscle action potential simulations using the line source model

Using a signal processing approach, we analyze the line source model for muscle action potential (AP) modeling. We show that the original model presents a tradeoff between violating the Nyquist criterion on one hand and using a discretization frequency that is unnecessarily high with respect to the bandwidth of the generated AP on the other. Here, we present an improved line source model that, compared to the original, allows a lower discretization frequency while retaining the accuracy by simply introducing a continuous-time anti-aliasing filter. Moreover, a transfer function form of the transmembrane current is presented that promote the use of sophisticated signal processing methods on these type of signals. Both continuous-time and discrete-time models are presented. We also address and analyze the implications of the finite length of the muscle fibers. Including this in the model is straightforward, owing to the convolutional form of the line source model, and is manifested by a simple transformation of the associated weighting function. AP modeling is discussed for the three different electrode models: the concentric needle electrode, the single fiber electrode, and the macro electrode. The presented model is suitable for modeling large motor units, where both accuracy and computational efficiency are important factors. To simplify the selection of the discretization interval, we derive what we call the cumulative cutoff frequency that provides an estimate of the required Nyquist frequency.     

Simulation Techniques in Electromyography

A motor unit action potential (MUAP) recorded in clinical electromyography (EMG) is the spatial and temporal summation of the action potentials (AP's) from all muscle fibers in a motor unit (MU). An important determinant of MUAP waveform characteristics is the size of the recording electrode. In this paper, we have described the use of a modified line source model of single muscle fiber action potentials to simulate MUAP's as recorded by single fiber (SF) EMG, concentric needle (CN) EMG, and macro-EMG electrodes. Results indicate that SFEMG recordings from a normal MU contain mainly the AP's of the closest one to three muscle fibers of the MU. The amplitude, area, and duration of the simulated CNEMG MUAP's are determined mainly by the number and size of muscle fibers within a semicircular territory of 0.5, 1.5, and 2.5 mm, respectively, around the tip of the electrode. The amplitude and area of simulated macro-EMG MUAP's increase with the number of muscle fibers in the MU.     

Topographical map of innervation zones within single motor unitsmeasured with a grid surface electrode

A grid electrode with closely spaced contacts was developed to measure the spatial distribution of motor unit action potentials on a skin surface. This electrode was used to estimate the configuration of innervation zones. The derived action potentials showed waveforms originating from the middle length of the muscle belly and propagating bidirectionally along the muscle fibers. The position of the innervation zones was estimated from the source of the propagation. By isolating action potentials from single motor units, the innervation zones for individual motor units were defined. Studies were performed in the biceps brachii of three normal subjects. The innervation zones comprised one to four separate regions and spread over up to 20 mm along the length of muscle fibers. The number and the spreading area of innervation zones varied with the subject and with the motor unit     

Modeling of muscle motor unit innervation process correlation and common drive

Concurrently active motor units (MUs) of a given muscle can exhibit a certain degree of synchronous firings, and a certain degree of common variation in their firing rates. The former property is referred to as motor unit synchrony in the literature, which is termed motor unit innervation process (MUIP) correlation in this study. The latter is referred to as motor unit common drive and can be quantified by the common drive coefficient, which is the correlation coefficient between the smoothed firing rates of the two MUs. Both properties have important roles and implications in the generation and resulting characteristics of the myoelectric signal and for the development of signal processing algorithms in myoelectric signal (MES) applications. In order to study these implications and characteristics, in this paper estimation procedures are developed to quantify the degree of MUIP correlation and common drive as functions of physiological parameters. Also, the interaction between MUIP correlation and motor unit common drive is studied in a physiologically realistic simulation model. Neurons modeled by Hodgkin-Huxley systems form the framework of the simulation model in which excitation and synaptic characteristics can be modified. MUIP correlation and common drive degree and interaction are studied through a number of simulations. To support the simulation results, experimental in vivo motor unit trains were collected at low levels of contraction from 11 subjects, and decomposed into the constituent unit trains giving 50 concurrently active motor unit pairs. The simulation demonstrated that the innervation process correlation coefficient is controlled primarily by the postsynaptic conductance, gsyn, and was less than 0.05 mS/cm2 for realistic values of gsyn. The common drive was found to be controlled by the exciting neuron input with no statistically significant interaction between it and the MUIP correlation. The experimental data gave results in close agreement with those of the simulation.     

Modeling of surface myoelectric signals. II. Model-based signalinterpretation

Experimental electromyogram (EMG) data from the human biceps brachii were simulated using the model described in [10] of this work. A multichannel linear electrode array, spanning the length of the biceps, was used to detect monopolar and bipolar signals, from which double differential signals were computed, during either voluntary or electrically elicited isometric contractions. For relatively low-level voluntary contractions (10%-30% of maximum force) individual firings of three to four-different motor units were identified and their waveforms were closely approximated by the model. Motor unit parameters such as depth, size, fiber orientation and length, location of innervation and tendonous zones, propagation velocity, and source width were estimated using the model. Two applications of the model are described. The first analyzes the effects of electrode rotation with respect to the muscle fiber direction and shows the possibility of conduction velocity (CV) over- and under-estimation. The second focuses on the myoelectric manifestations of fatigue during a sustained electrically elicited contraction and the interrelationship between muscle fiber CV, spectral and amplitude variables, and the length of the depolarization zone. It is concluded that a) surface EMG detection using an electrode array, when combined with a model of signal propagation, provides a useful method for understanding the physiological and anatomical determinants of EMG waveform characteristics and b) the model provides a way for the interpretation of fatigue plots.     

Muscle-like contraction control of tendon-sheath artificial muscle

The development of artificial muscles has focused on a high energy–weight ratio and soft structures, however, little work has been done towards muscle-like contraction behaviors. This unfortunately leads to a lack of comfort and safety, which is especially important for robotic applications like orthotics and exoskeletons. In this paper, we propose a contraction control method for a tendon-sheath artificial muscle to contract and relax like biological muscles. In view of the nonlinear transmission characteristics of the tendon-sheath artificial muscle, a transmission model is established and its accuracy is verified through experiments. A muscle-like contraction control method is then proposed based on the transmission model and the Hill-type muscle model. Through this method, the contraction force of the tendon-sheath artificial muscle can be adjusted according to the output displacement and velocity of the tendon-sheath mechanism estimated by the transmission model. Isometric contraction and quick-release experiments are then conducted. The experimental results demonstrate that this control method allows the tendon-sheath artificial muscle to contract with specific muscle-like force–length and force–velocity properties.     

A robust adaptive controller for PM motor drives in roboticapplications

The paper deals with theoretical development and practical implementation of an adaptive speed and position regulator suitable for robotic applications. The proposed adaptive control scheme is characterized by a reduced amount of computation and is based on the model reference adaptive control approach to compensate the variations of the system parameters, such as inertia and torque constant. A disturbance torque observer is employed to balance the required load torque and reduce the complexity of the adaptive algorithm. Simulation tests of a robotic drive, including an interior type permanent magnet synchronous (IPMS) motor, are reported in order to compare the proposed control scheme with standard speed and position regulators. Experimental results, obtained from a prototype based on a commercial PC board, are also reported in order to practically evaluate the feasibility and the features of the proposed adaptive control scheme     

An algorithm for sequential signal estimation and system identification for EMG signals

This paper presents a new algorithm for optimal adaptation of the signal templates of a matched filter bank used in the detection of the motor unit action potential waveforms (abbreviated as MUAP's) in an electromyogram (EMG). It is of interest, for clinical diagnosis and therapy, to detect as many MUAP's as possible in a single measurement, and to determine for each motor unit the repetition. rate of its respective MUAP. For this purpose, we have developed a computer program which, in addition to other subprograms, contains the adaptive filter bank mentioned above. The templates in this fllter bank have to be adapted to nonpredetermined changes in measurement conditions such as the movement of the needle electrode inserted in the muscle. In the present paper, the above templates are estimated by means of a "tumbling algorithm," so called because the successive MUAP's from a given motor unit are used as noisy data vectors in a time-varying Kalman filter-predictor framework, which alternately estinates their evolving shapes and identifies the time-varying parameters of the model generating them. The algorithm has been applied with success to synthetic and real EMG data.     

Impact dynamics of a differential gears based underactuated robotic arm for moving target capturing

This paper presents impact dynamic analysis of a differential gears based underactuated robotic arm colliding with a moving target. The robotic arm employs a set of one-input-dual-output planetary gears to distribute driving torque of the underactuated joints, which takes advantage of the underactuation to cope with collision. Based on an analysis of the movement relation of three joints, the driving torque is resolved into generalized force corresponding to coordinates of two underactuated joints for establishing a dynamic model of the underactuated robotic arm. Along with the dynamic equation, impulse-momentum equations involving the restitution coefficient of the system (arm and target) are used to establish the impact dynamic model which can extrapolate the impact momentum on the drive component. With the impact dynamic model, the impact-absorbing capability of the underactuated robotic arm is illustrated that the impact momentum on the drive component is reduced by a factor of three compared with a 2-DOF fully actuated rigid robotic arm. Impact experiments were carried out to validate the inference results presented in this paper by using of a prototype of the proposed arm.     

An experimentally validated analytical model for gate line-edgeroughness (LER) effects on technology scaling

This letter introduces an analytical model to represent line-edge roughness (LER) effects on both off-state leakage and drive current for sub-100-nm devices. The model partitions a given device into small unit cells along its width, each unit cell assumes a constant gate length (i.e., cell's width is small compared to LER spatial frequency). An analytical model is used to represent saturated threshold voltage dependency on the unit cell's gate length. Using this technique, an efficient and accurate model for LER effects (through V_ts variations) on off-state leakage and drive current is proposed and experimentally validated using 193 and 248 nm lithography for devices with 80-nm nominal gate lengths. Assuming that the deviation from the ideal 0-LER case remains constant from generation to generation, the model predicts that 3 nm or less LER is required for 50-60-nm state-of-the-art devices in the 0.1-μm technology node. Based on data presented, we suggest that the LER requirement for this technology node is attainable with an alternated phase-shift type of patterning process     

MIMO非线性系统的多模型建模方法

针对实际工业过程中多变量系统存在着非线性、工况范围广的特点,本文提出了一种新的多模型建模方法.首先对系统调度变量进行满意模糊c均值聚类,在此基础上采用基于加权性能指标的多模型辨识算法辨识多模型系统,得到的模型在全局拟合与局部特性之间取得良好的权衡,同时能得到每个局部模型的适用域.以典型pH中和过程为对象,采用上述建模方法建立其系统多模型,仿真结果验证了该建模方法的有效性.     

A nonlinear model of the phasic dynamics of muscle activation

This paper presents a phasic excitation-activation (PEXA) model of the process of motoneuron excitation and the resultant activation and force development of a motor unit. The model input is an amount of depolarizing current (as when injected with an intracellular electrode) and the model output is muscle force. The model includes dynamics and nonlinearities similar to phenomena discovered experimentally by others: the firing rate response of motoneurons to steps of depolarizing current and the "catch-like enhancement" of force produced by overlapping motor neuron action potentials. The parameter values used in this model are derived from experimentally measured data and expressed in physical units, and model predictions extend to published data beyond those used in generating the model parameter values.     

Learning techniques to train neural networks as a state selectorfor inverter-fed induction machines using direct torque control

Neural networks are receiving attention as controllers for many industrial applications. Although these networks eliminate the need for mathematical models, they require a lot of training to understand the model of a plant or a process. Issues such as learning speed, stability, and weight convergence remain as areas of research and comparison of many training algorithms. This paper discusses the application of neural networks to control induction machines using direct torque control (DTC). A neural network is used to emulate the state selector of the DTC. The training algorithms used in this paper are the backpropagation, adaptive neuron model, extended Kalman filter, and the parallel recursive prediction error. Computer simulations of the motor and neural-network system using the four approaches are presented and compared. Discussions about the parallel recursive prediction error and the extended Kalman filter algorithms as the most promising training techniques is presented, giving their advantages and disadvantages     

Contact control concepts in manipulation robotics-an overview

This paper presents the state of the art in the control of robotic manipulators in constrained motion tasks. Contact control concepts are classified using different criteria, and their main characteristics are analyzed. For each of the presented contact control concepts, the essential characteristics are stated. The paper covers some early ideas and their later improvements, as well as trends in this field. The advantages and drawbacks of the various control schemes are outlined, and they are compared from the standpoint of their implementation issues. In the paper, all characteristic results in the stability analysis of robotic manipulators in the constrained motion tasks are briefly reported. A new approach to the correct solution of contact tasks control is mentioned as well     

A New Type of Motor: Pneumatic Step Motor

This paper presents a new type of pneumatic motor, a pneumatic step motor (PneuStep). Directional rotary motion of discrete displacement is achieved by sequentially pressurizing the three ports of the motor. Pulsed pressure waves are generated by a remote pneumatic distributor. The motor assembly includes a motor, gearhead, and incremental position encoder in a compact, central bore construction. A special electronic driver is used to control the new motor with electric stepper indexers and standard motion control cards. The motor accepts open-loop step operation as well as closed-loop control with position feedback from the enclosed sensor. A special control feature is implemented to adapt classic control algorithms to the new motor, and is experimentally validated. The speed performance of the motor degrades with the length of the pneumatic hoses between the distributor and motor. Experimental results are presented to reveal this behavior and set the expectation level. Nevertheless, the stepper achieves easily controllable precise motion unlike other pneumatic motors. The motor was designed to be compatible with magnetic resonance medical imaging equipment, for actuating an image-guided intervention robot, for medical applications. For this reason, the motors were entirely made of nonmagnetic and dielectric materials such as plastics, ceramics, and rubbers. Encoding was performed with fiber optics, so that the motors are electricity free, exclusively using pressure and light. PneuStep is readily applicable to other pneumatic or hydraulic precision-motion applications     

Modeling and control strategies for a variable reluctancedirect-drive motor

A high-performance ripple-free dynamic torque controller for a variable-reluctance (VR) motor intended for trajectory tracking in robotic applications is designed. A modeling approach that simplifies the design of the controller is investigated. Model structure and parameter estimation techniques are presented. Different approaches to the overall torque controller design problem are discussed, and the solution adopted is illustrated. A cascade controller structure consisting of a feedforward nonlinear torque compensator, cascaded to a nonlinear flux or current closed-loop controller is considered, and optimization techniques are used for its design. Although developed for a specific commercial motor, the proposed modeling and optimization strategies can be used for other VR motors with magnetically decoupled phases, both rotating and linear. Laboratory experiments for model validation and preliminary simulation results of the overall torque control system are presented