Prosthesis Control Research Articles

Temporal and spatial goal-directed reaching in upper limb prosthesis users.

Understanding the fundamental characteristics of prosthetic movement control is imperative in improving prosthesis design and training. This study quantified how using an upper limb prosthesis affected performance during goal-directed reaching tasks. Nine prosthesis users with unilateral transradial limb absence and nine healthy controls completed a series of goal-directed reaching movements with different goals: one spatial and three temporal with different goal frequencies. We quantified end-point accuracy, smoothness, and peak speed for the spatial task and temporal accuracy, horizontal distance, and speed for the temporal task. For the temporal task, we also used a goal-equivalent manifold (GEM) approach to decompose variability in movement distance and speed into those perpendicular and tangential to the GEM. Detrended fluctuation analysis (DFA) quantified the temporal persistence of each time series. For the spatial task, movements made with prostheses were less smooth, had larger end-point errors, and had slower peak speed compared to those with control limbs (p < 0.041). For the temporal task, movements made with prostheses and intact limbs of prosthesis users and control limbs were similar in distance and speed and had similar timing errors (p > 0.138). Timing errors, distance, speed, and GEM deviations were corrected similarly between prosthetic limbs and control limbs (p > 0.091). The mean and variability of distance, speed, and perpendicular deviations decreased with increased goal frequency (p < 0.001). Our results suggest that prosthesis users have a sufficient internal model to successfully complete ballistic movements but are unable to accurately complete movements requiring substantial feedback.

An All-in-One, Bioderived, Air-Permeable, and Sweat-Stable MXene Epidermal Electrode for Muscle Theranostics.

Muscle fatigue is a common symptom experienced by many people and associated with less maximal force production of fatigued muscle. It is highly desirable to simultaneously and imperceivably diagnose muscle fatigue and restore muscle function using one skin electrode, yet no such electrode has been developed so far. Herein, we report an all-in-one, bioderived, air-permeable, and sweat-stable MXene electrode that can concurrently and comfortably record electromyographic (EMG) signals and achieve electrostimulation and electrothermal therapy for muscle theranostics. Leveraging the structural arrangement of perennial herbs and ion cross-linking of MXene in sweat, MXene-based electrodes (MBE) exhibit high breathability, are ultralightweight (∼0.25 mg/cm3), and have low and stable electrode-skin interfacial impedance at a variety of environments, facilitating the long-term reliable monitoring of electrophysiology. Taken together with electrostimulation and electrothermal therapy at the skin surface, MBE can diagnose muscle fatigue and restore muscle function by stimulating blood circulation. In addition, it can also be used for muscle rehabilitation training and prosthesis control via human-computer interaction. Our all-in-one, bioderived, air-permeable, and sweat-stable MXene electrode has a great potential for daily wearable healthcare of muscle fatigue.

LSTM Recurrent Neural Network for Hand Gesture Recognition Using EMG Signals

Currently, research on gesture recognition systems has been on the rise due to the capabilities these systems provide to the field of human–machine interaction, however, gesture recognition in prosthesis and orthesis has been carried out through the use of an extensive amount of channels and electrodes to acquire the EMG (Electromyography) signals, increasing the cost and complexity of these systems. The scientific literature shows different approaches related to gesture recognition based on the analysis of EMG signals using deep learning models, highlighting the recurrent neural networks with deep learning structures. This paper presents the implementation of a Recurrent Neural Network (RNN) model using Long-short Term Memory (LSTM) units and dense layers to develop a gesture classifier for hand prosthesis control, aiming to decrease the number of EMG channels and the overall model complexity, in order to increase its scalability for embedded systems. The proposed model requires the use of only four EMG channels to recognize five hand gestures, greatly reducing the number of electrodes compared to other approaches found in the literature. The proposed model was trained using a dataset for each gesture EMG signals, which were recorded for 20 s using a custom EMG armband. The model reached an accuracy of to 99% for the training and validation stages, and an accuracy of 87 ± 7% during real-time testing. The results obtained by the proposed model establish a general methodology for the reduction of complexity in the recognition of gestures intended for human.machine interaction for different computational devices.

EMG feedback outperforms force feedback in the presence of prosthesis control disturbance.

Closing the prosthesis control loop by providing artificial somatosensory feedback can improve utility and user experience. Additionally, closed-loop control should be more robust with respect to disturbance, but this might depend on the type of feedback provided. Thus, the present study investigates and compares the performance of EMG and force feedback in the presence of control disturbances. Twenty able-bodied subjects and one transradial amputee performed delicate and power grasps with a prosthesis in a functional task, while the control signal gain was temporarily increased (high-gain disturbance) or decreased (low-gain disturbance) without their knowledge. Three outcome measures were considered: the percentage of trials successful in the first attempt (reaction to disturbance), the average number of attempts in trials where the wrong force was initially applied (adaptation to disturbance), and the average completion time of the last attempt in every trial. EMG feedback was shown to offer significantly better performance compared to force feedback during power grasping in terms of reaction to disturbance and completion time. During power grasping with high-gain disturbance, the median first-attempt success rate was significantly higher with EMG feedback (73.3%) compared to that achieved with force feedback (60%). Moreover, the median completion time for power grasps with low-gain disturbance was significantly longer with force feedback than with EMG feedback (3.64 against 2.48 s, an increase of 32%). Contrary to our expectations, there was no significant difference between feedback types with regards to adaptation to disturbances and the two feedback types performed similarly in delicate grasps. The results indicated that EMG feedback displayed better performance than force feedback in the presence of control disturbances, further demonstrating the potential of this approach to provide a reliable prosthesis-user interaction.

Effectiveness evaluation of three-dimensional printed titanium-alloy prosthesis reconstruction after distal tibia tumor segment resection

To evaluate the feasibility and short-term effectiveness of three-dimensional (3D) printed titanium-alloy prosthesis reconstruction after the distal tibia tumor segment resection. The clinical data of 6 patients with bone defect after distal tibia tumor segment resection treated with 3D printed titanium-alloy prosthesis reconstruction and tibiotalar joint fusion between January 2020 and December 2021 were retrospectively analyzed. There were 2 males and 4 females; the age ranged from 12 to 35 years, with an average of 18.5 years. Among them, 4 cases were osteosarcoma, 1 case was Ewing sarcoma, and 1 case was giant cell tumor of bone. The Enneking staging was stage ⅡA in 3 cases, stage ⅡB in 2 cases, and stage Ⅲ in 1 case. The disease duration was 2-5 months (mean, 3.2 months). All patients received preoperative neoadjuvant therapy, and patients with osteosarcoma and Ewing sarcoma started chemotherapy at3 weeks after operation. The systemic and local tumor conditions and prosthesis conditions were evaluated regularly after operation. The Musculoskeletal Tumor Society (MSTS) score and the American Orthopaedic Foot and Ankle Society (AOFAS) score were used to evaluate the lower extremity and ankle function. All patients were followed up 8-26 months, with an average of 15.6 months. There was no local recurrence and distant metastasis during the follow-up. The ankle joints of 5 cases were all in 90° functional position at last follow-up, and there was no complication such as prosthesis loosening and fracture; the ankle joint fusion was stable, the local bone ingrowth was good, and the daily activities could be completed, but the ankle range of motion was limited and the ankle joint was stiff. The MSTS score ranged from 22 to 26, with an average of 24, and 3 cases were evaluated as excellent and 2 cases were good; the AOFAS score ranged from 71 to 86, with an average of 80.6, and 4 cases were evaluated as good and 1 case was fair. One patient had severe periprosthetic infection at 2 months after operation, resulting in failure of prosthesis implantation, pain in limb movement, and poor ankle function; MSTS score was 12, AOFAS score was 50, and both were evaluated as poor; distraction osteogenesis was performed after removal of prosthesis and infection control, at present, it was still in the process of distraction osteogenesis, and local osteogenesis was acceptable. Using 3D printed titanium-alloy prosthesis and tibiotalar joint fusion to reconstruct the bone defect after distal tibia tumor segment resection has satisfactory mechanical stability and function, and is one of the effective distal tibial limb salvage methods.

Discrete control of transfemoral prostheses for human walking with magnetorheological compensation

This abstract presents the modeling and control of a transfemoral prosthesis with magnetorheological compensation. Initially, leg modeling is shown by the formulation of the Euler-Lagrange motion equations and the inclusion of the modified Bouc-Wen model for the magnetorheological damper. Subsequently, the linealization of these models is presented using the Taylor series approximation. For the design of the controller is used the analysis of the geometric place of the roots. The simulation results of the control system indicate that the prosthesis is capable of accurately following established joint references and effectively rejects external disturbances.

The Usage of Constrained Independent Component Analysis to Reduce Electrode Displacement Effects in Real-Time Surface Electromyography-Based Hand Gesture Classifications

Aim: In real-time control of prosthesis, orthosis, and human–computer interface applications, the displacement of surface electrodes may cause a total disruption or a decline in the classification rates. In this study, a constrained independent component analysis (cICA) was used as an alternative method for addressing the displacement problem of surface electrodes. Materials and Methods: The study was tested by classifying six-hand gestures offline and in real-time to control a robotic arm. The robotic arm has five degrees of freedom, and it was controlled using surface electromyography (sEMG) signals. The classification of sEMG signals is realized using artificial neural networks. cICA algorithm was utilized to improve the performance of classifiers due to the negative effect of electrode displacement issues. Results: In the study, the classification results of the cICA applied and unapplied sEMG signals were compared. The results showed that the proposed method has provided an increase between 4% and 13% in classifications. The average classification rates for six different hand gestures were calculated as 96.66%. Conclusions: The study showed that the cICA method enhances classification rates while minimizing the impact of electrode displacement. The other advantage of the cICA algorithm is dimension reduction, which is important in real time applications. To observe the performance of the cICA in the real-time application, a robotic arm was controlled using sEMG signals.

SEMG-Based Gesture Recognition Using Deep Learning From Noisy Labels.

Gesture recognition for myoelectric prosthesis control utilizing sparse multichannel surface Electromyography (sEMG) is a challenging task, and from a Muscle-Computer Interface (MCI) standpoint, the performance is still far from optimal. However, the design of a well-performed sEMG recognition system depends on the flexibility of the input-output function and the dataset's quality. To improve the performance of MCI, we proposed a novel gesture recognition framework that (i) Enrich the spectral information of the sparse sEMG signals by constructing a fused map image (denoted as sEMG-Map) that integrates a multiresolution decomposition (by means of orthogonal wavelets) through the raw signals then rely upon the Convolutional Neural Network (CNN) capacity to exploit the composite hierarchies in the constructed sEMG-Map input. (ii) Deals with the label noise by proposing a data-centric method (denoted as ALR-CNN) that synchronously refines the falsely labeled samples and optimizes the CNN model based on two basic assumptions. First, the deep model accuracy improves as the training progress. Second, a set of successive learnable max-activated outputs of a well-performed deep model is a reliable estimator for motion detection in the muscle activation pattern. Our proposed framework is evaluated on three large-scale public databases. The average classification accuracy is 95.50%, 95.85%, and 85.58% for NinaPro DB2, NinaPro DB7, and NinaPro DB3, respectively. The experimental results verify the effectuality of the proposed method and show high accuracy.

Simultaneous estimation of grip force and wrist angles by surface electromyography and acceleration signals

In myoelectric control, simultaneous and proportional (SP) control of multiple degrees of freedom (DOFs) can realize a high level of dexterity. This study proposed a new control scheme that simultaneously estimates continuous grip force and wrist angles using a combination of surface electromyography (sEMG) and acceleration (ACC) signals. Four popular EMG features were utilized to study the regression accuracy. Nine intact subjects were recruited. They performed six force-position varying movements which involved simultaneous activation of four DOFs (wrist flexion/extension, ulnar/radial deviation, pronation/supination, and grip). The estimation performance was evaluated by the Pearson Correlation Coefficient (cc). The highest cc value was 92.12 ± 1.51%, 96.68 ± 1.16%, and 97.18 ± 0.69% for the sEMG-only, ACC-only, and multi-modal methods, respectively. Results indicate that the multi-modal approach can significantly improve regression accuracy of concurrent grip force and wrist angles, especially in pronation/supination. Besides, the multi-modal method showed higher robustness when the available number of sensors decreased and better generalization ability between movements. In conclusion, the proposed multi-modal approach has the potential for SP control of future hand prostheses.

Utilization of Flex Controlled Robotics on Prosthetic Hand instead of Damaged Body Parts

Scientists have researched human anatomy and biology to produce human-like prosthetic hands. In the absence of a hand, prosthetics assist in reestablishing independence and confidence. Prosthetic hands are limited in functionality and cost. A low-cost anthropomorphic prosthetic hand was designed. With five independently operated fingers and powered thumb roll movement, the design is unlike anything else on the market. This research aims to provide a cheaper alternative to the market’s robotic prostheses. Inability to use limbs causes psychological and physical issues for amputees. Arduino, an open-source microprocessor, was used to build a prototype prosthetic hand. An Android app allows users to select a hand motion to circumvent myoelectric prosthesis control mechanisms. The Arduino uses this data to power servos that individually trigger each finger.

Modeling the Transitional Kinematics Between Variable-Incline Walking and Stair Climbing.

Although emerging powered prostheses can enable people with lower-limb amputation to walk and climb stairs over different task conditions (e.g., speeds and inclines), the control architecture typically uses a finite-state machine to switch between activity-specific controllers. Because these controllers focus on steady-state locomotion, powered prostheses abruptly switch between controllers during gait transitions rather than continuously adjusting leg biomechanics in synchrony with the users. This paper introduces a new framework for powered prosthesis control by modeling the lower-limb joint kinematics over a continuum of variable-incline walking and stair climbing, including steady-state and transitional gaits. Steady-state models for walking and stair climbing represent joint kinematics as continuous functions of gait phase, forward speed, and incline. Transition models interpolate kinematics as convex combinations of the two steady-state models, with an additional term to account for kinematics that fall outside their convex hull. The coefficients of this convex combination denote the similarity of the transitional kinematics to each steady-state mode, providing insight into how able-bodied individuals continuously transition between ambulation modes. Cross-validation demonstrates that the model predictions of untrained kinematics have errors within the range of physiological variability for all joints. Simulation results demonstrate the model's robustness to incline estimation and mode classification errors.

Composite Recurrent Convolutional Neural Networks Offer a Position-Aware Prosthesis Control Alternative While Balancing Predictive Accuracy with Training Burden.

To mitigate the "limb position effect" that hinders myoelectric upper limb prosthesis control, pattern recognition-based models must accurately predict user-intended movements across a multitude of limb positions. Such models can use electromyography (EMG) and inertial measurement units to capture necessary multi-position data. However, this data capture solution requires lengthy user-performed model training routines, with movements in many limb positions, plus retraining thereafter due to inherent signal variations over time. While a general-purpose control model (trained with a dataset that represents numerous device users) eliminates the user-training requirement altogether, it yields low movement predictive accuracy. Conversely, a user-specific control model (trained with a smaller dataset from an individual) yields high predictive accuracy, but requires retraining over time. This study capitalizes on the benefits offered by both such control options, and contributes an alternative control solution-a novel recurrent convolutional neural network (RCNN)-based Composite Model that combines the representation portion of a general-purpose model, with the decision portion of a user-specific model. The resulting Composite Model offers moderate movement predictive accuracy across various limb positions and a reduction in user training routine requirements, suggesting a new research direction to help mitigate the limb position effect along with model training burden.

Unsupervised Myocontrol of a Virtual Hand Based on a Coadaptive Abstract Motor Mapping.

Applications of simultaneous and proportional control for upper-limb prostheses typically rely on supervised machine learning to map muscle activations to prosthesis movements. This scheme often poses problems for individuals with limb differences, as they may not be able to reliably reproduce the training activations required to construct a natural motor mapping. We propose an unsupervised myocontrol paradigm that eliminates the need for labeled data by mapping the most salient muscle synergies in arbitrary order to a number of predefined prosthesis actions. The paradigm is coadaptive, in the sense that while the user learns to control the system via interaction, the system continually refines the identification of the user's muscular synergies. Our evaluation consisted of eight subjects without limb-loss performing target achievement control tasks of four actions of the hand and wrist. The subjects achieved comparable performance using the proposed unsupervised myocontrol paradigm and a supervised benchmark method, despite reporting increased mental load with the former.

Optimizing Template Models to Quantifiably Assess Center of Mass Kinematic Reconstruction.

The center of mass (COM) plays a fundamental role in human ambulation, but the redundant nature of the human body adds complexity to mathematically modeling its dynamics. Template models like the Bipedal Spring Loaded Inverted Pendulum (B-SLIP) and the Virtual Pivot Point (VPP) address this complexity by removing the redundancy while retaining desired salient characteristics, such as the COM evolution. However, template models for the COM during human walking have mostly been used for qualitative analysis due to issues such as overestimation of COM vertical displacement. This paper considers a quantifiable template-based analysis of human walking by using an optimization framework to set the model parameter values for matching both explicitly and implicitly considered gait characteristics. Furthermore, it is shown that allowing the leg stiffness of the B-SLIP and VPP model to vary throughout the gait cycle better matches vertical COM trajectories with 54%-63% error reduction. These optimized template models show promise in retaining ground reaction force (GRF) information, which is not explicitly considered during the optimization process. Future work looks to incorporate these optimized trajectories as a reference for control of a lower-limb knee-ankle prosthesis.

Data-driven Gait-predictive Model for Anticipatory Prosthesis Control.

Physiological movement is pre-planned based on current movement state, proprioception, and environmental cues. This preplanning is necessary to allow efficient use of the viscoelastic properties of musculoskeletal tissues in a 4D-environment. Similarly, efficient use of prosthetic devices needs to compensate for the time it takes to control the system. In this study, we propose a gated recurrent net-based gait predictive model to continuously predict the ankle angles and moments fifty milliseconds in advance based on the past trajectory of the input signals. It was observed that using a single input signal (the shank angle), high accuracy of prediction $(R^{2}\gt 0.91)$ was achieved for both ankle angle and moments on walking trials at a self-selected comfortable speed. The results of our study can be utilised for anticipatory lower-limb prosthesis control where embedded sensor information that reflects a prosthetic user's locomotive intent can be used to predict the required angles and moments in advance for actuating a prosthetic joint.

Toward improving control performance of myoelectric arm prosthesis by adding wrist position feedback.

Wearing a myoelectric prosthesis is a basic way for limb amputees to restore their lost limb functions in the activities of daily living (ADLs). However, it is estimated that around 40% of amputees refuse the prosthesis. One of the primary reasons would be that the current prostheses lack appropriate sensory feedback. Currently, the amputees only depend on their visual feedback (Vis-FB) when using their arm prostheses. It would be difficult for them to accurately control the wrist position, which is vital for flexible manipulation in ADLs. This manuscript designed a myoelectric arm prosthesis with wrist position feedback (WP-FB). To study the effect level of position feedback on prosthetic control, two tests were performed. The vibrotactile perception range test aims to analyze the perception sensitivity of the vibration in humans and obtain the optimal perception range utilized in the sensory feedback test. The sensory feedback test analyzes the effectiveness of the position feedback by comparing three feedback methods of Vis-FB, WP-FB, and a combination of Vis-FB and WP-FB (VP-FB). These tests were conducted by asking six able-bodied subjects to perform 20 movement combinations of five target positions. The WP-FB was transiently activated with five vibrating motors embedded in an armband to stimulate the arm stump when the prosthetic wrist rotates to the target positions. Our experimental results showed that when WP-FB was added to the prosthetic control, the absolute angular error (AAE) of the prosthetic wrist declined from 4.50° to 1.08° while the success rate 3 (SR3) increased from 0.34 to 0.84, respectively. This study demonstrates the importance of WP-FB to the effective control of the arm prosthesis.

A data-driven machine learning approach for brain-computer interfaces targeting lower limb neuroprosthetics.

Prosthetic devices that replace a lost limb have become increasingly performant in recent years. Recent advances in both software and hardware allow for the decoding of electroencephalogram (EEG) signals to improve the control of active prostheses with brain-computer interfaces (BCI). Most BCI research is focused on the upper body. Although BCI research for the lower extremities has increased in recent years, there are still gaps in our knowledge of the neural patterns associated with lower limb movement. Therefore, the main objective of this study is to show the feasibility of decoding lower limb movements from EEG data recordings. The second aim is to investigate whether well-known neuroplastic adaptations in individuals with an amputation have an influence on decoding performance. To address this, we collected data from multiple individuals with lower limb amputation and a matched able-bodied control group. Using these data, we trained and evaluated common BCI methods that have already been proven effective for upper limb BCI. With an average test decoding accuracy of 84% for both groups, our results show that it is possible to discriminate different lower extremity movements using EEG data with good accuracy. There are no significant differences (p = 0.99) in the decoding performance of these movements between healthy subjects and subjects with lower extremity amputation. These results show the feasibility of using BCI for lower limb prosthesis control and indicate that decoding performance is not influenced by neuroplasticity-induced differences between the two groups.

Improved Classification Accuracy of Hand Movements Using Softmax Classifier and Kalman Filter.

Accurate identification of the intended hand movement from the surface Electromyography (sEMG) data is desired for effective control of myoelectric lower arm prostheses. This study improves the classification accuracy of hand gestures by using feature arrays, Kalman filter (KF), and a Softmax classifier. We use data from BioPatRec database to classify ten hand movements performed by 17 participants. The proposed classifier achieved 95.3% accuracy without KF, and 99.3% accuracy when KF was used to smooth the training data.

Non-Invasive Stable Sensory Feedback for Closed-Loop Control of Hand Prosthesis.

The absence of somatotopic sensory feedback limits the function of conventional prosthetic hands. In this study, we integrated a non-invasive sensory feedback system into a commercial Bebionic hand with new customized surface stimulation electrodes. Multiple modalities of tactile and hand aperture sensory information were conveyed to the amputee via the technique of evoked tactile sensation (ETS) elicited at projected finger map (PFM) of residual limb and an additional electrotactile stimulation in the ipsilateral upper arm. A previously developed anti-stimulus artifact module was used to remove the stimulus artifact from surface electromyographic (sEMG) signals, and the filtered sEMG envelops controlled the speed of open/close of the Bebionic hand. The Ag/AgCl surface stimulation electrode in 10-mm diameter was specially designed to fit the restricted PFM areas for stable perception. We evaluated the alternating-current (AC) impedance magnitude of this electrode stimulated over 12 hours. The perceptual and upper thresholds in pulse-width over 200 days at PFM areas were recorded to assess the stability of the non-invasive sensory neural interface. The efficacy of multi-modality feedback for identification of physical properties of objects was also assessed. Results showed that the AC impedance of customized surface stimulation electrode was stable over 12 hours of stimulation. The perceptual and upper thresholds were stable over 200 days. This non-invasive sensory feedback enabled a forearm amputee to identify the compliance and length of grasped objects with an accuracy of 100 %. Results illustrated that the multi-modality sensory feedback system provided stable and sufficient sensory information for perceptual discrimination of physical features of grasped objects. Clinical Relevance- This study demonstrated a promising and novel way to restore stable sensory feedback non-invasively for commercial hand prostheses.

An Exploration of the Optimal Feature-Classifier Combinations for Transradial Prosthesis Control.

Within state-of-the-art gesture-based upper-limb myoelectric prosthesis control, gesture recognition commonly relies on the classification of features extracted from electromyographic (EMG) data gathered from the amputee's residual forearm musculature. Despite best efforts in broadly maximizing gesture recognition accuracy, there does not yet exist a feature-classifier combination accepted as best-practice. In turn, this work hypothesizes that no single feature-classifier combination can consistently maximize accuracy across subjects, positing instead that control schemes should be personalized to the individual. To investigate this hypothesis, the study employed the 40-subject, 49-gesture Ninapro Database 2 (DB2) to compare the performance of 7 different historic, more recent, and state-of-the-art feature sets when classified by 5 machine learning algorithms commonly seen within EMG-based pattern recognition literature. The results demonstrate the ability of Linear Discriminant Analysis (LDA) to marginally exceed other more computationally intensive classifiers in terms of mean accuracy, while the feature set which maximized the highest proportion of individuals' accuracies was shown to vary with both classifier choice and gesture count.