Recently, it has been proven that targeting motor impairments as early as possible while using wearable mechatronic devices for assisted therapy can improve rehabilitation outcomes. However, despite the advanced progress on control methods for wearable mechatronic devices, the need for a more natural interface that allows for better control remains. To address this issue, electromyography (EMG)-based gesture recognition systems have been studied as a potential solution for human–machine interface applications. Recent studies have focused on developing user-independent gesture recognition interfaces to reduce calibration times for new users. Unfortunately, given the stochastic nature of EMG signals, the performance of these interfaces is negatively impacted. To address this issue, this work presents a user-independent gesture classification method based on a sensor fusion technique that combines EMG data and inertial measurement unit (IMU) data. The Myo Armband was used to measure muscle activity and motion data from healthy subjects. Participants were asked to perform seven types of gestures in four different arm positions while using the Myo on their dominant limb. Data obtained from 22 participants were used to classify the gestures using three different classification methods. Overall, average classification accuracies in the range of 67.5–84.6% were obtained, with the Adaptive Least-Squares Support Vector Machine model obtaining accuracies as high as 92.9%. These results suggest that by using the proposed sensor fusion approach, it is possible to achieve a more natural interface that allows better control of wearable mechatronic devices during robot assisted therapies.

Smart textile sensors have been gaining popularity as alternative methods for the continuous monitoring of human motion. Multiple methods of fabrication for these textile sensors have been proposed, but the simpler ones include stitching or embroidering the conductive thread onto an elastic fabric to create a strain sensor. Although multiple studies have demonstrated the efficacy of textile sensors using the stitching technique, there is almost little to no information regarding the fabrication of textile strain sensors using the embroidery method. In this paper, a design guide for the fabrication of an embroidered resistive textile strain sensor is presented. All of the required design steps are explained, as well as the different embroidery design parameters and their optimal values. Finally, three embroidered textile strain sensors were created using these design steps. These sensors are based on the principle of superposition and were fabricated using a stainless-steel conductive thread embroidered onto a polyester–rubber elastic knit structure. The three sensors demonstrated an average gauge factor of $1.88\pm 0.51$ over a 26% working range, low hysteresis ($8.54\pm 2.66\%$), and good repeatability after being pre-stretched over a certain number of stretching cycles.

The rise in popularity of wearable robotic devices has brought many opportunities in the area of assistive devices for rehabilitation. However, despite their numerous advantages, these wearable mechatronic devices are not widely adopted due to a poor interaction between the device and the wearer. To solve this issue, several studies have proposed the use of sensor fusion of electroencephalography (EEG) and electromyography (EMG) signals using Convolutional Neural Networks (CNN) to detect the motion intention of the user and control the wearable device. Although normalization techniques can be applied during data pre-processing in order to improve performance, the correct normalization methods to apply for combined EEG and EMG data are unknown, hindering the potential improvement of the resulting CNN models. Therefore, in this study, no normalization, subject-wise, speed-wise, image-wise, and channel-wise normalization methods were compared. The effectiveness of these methods was tested on the overall and speed specific accuracies of CNN models trained using a database of combined EEG–EMG data collected during elbow flexion–extension motions at different speeds. The results of this study showed that no normalization improved the performance of the overall accuracy of CNN models, with grouped spectrogram-based models achieving an accuracy of 81.57 ± 7.11%. A similar trend was found for speed specific accuracies, in which no normalization showed to be a slightly better option during slow motions. Overall, these results provide information on the different tools that can be used to improve the performance of CNN-based models used for the control of wearable robotic devices.

The real-time use of electromyography (EMG)- based mechatronic rehabilitation devices aiming to detect stroke patients' hand grasping intention is hindered by a significant concern: the lack of robustness against variations in EMG signal patterns due to arm posture changes. This problem results in degraded EMG signal measurements and inaccurate recognition of muscle patterns. Several studies have aimed at tracking changes in EMG patterns by placing multiple EMG sensors around the forearm and developing a classifier using data collected from various arm postures recorded by all sensors. Although these methods show promise, the significant computational resources required for real-time data processing become notable concerns when using multiple EMG sensors. To address these challenges, this study introduces a novel approach that aims to reduce the number of EMG channels that need to be processed. The study proposes a new optimal-channel-selection technique, coupled with a convolutional neural network (CNN), which selects two out of eight EMG channels within an armband based on the arm posture and individual demographics. As a result of using only two channels rather than the entire array (eight channels), the user's grasping intention prediction time took only 2.3 seconds with a classification accuracy of around 81%. In comparison, the commonly used eightchannel method took 8.6 seconds for grasping intention detection with an accuracy level of 79%. These findings show potential in tackling the challenge of EMG measurement degradation caused by arm motion, offering a path towards enhanced accuracy and quicker responsiveness in EMG-based mechatronic rehabilitation devices

It is estimated that at least 66% of stroke survivors will present some condition that impair their ability to perform well during their activities of daily living. This is the reason why stroke survivors engage in rehabilitation therapies to improve their quality of life. To aid in their recovery process, robot-assisted technologies in the form of soft wearable devices have been proposed as a complementary method to traditional rehabilitation sessions. Within these soft devices, great attention has been placed on their sensing mechanisms. Among the different sensing modalities used to provide feedback to the soft wearable mechatronic device, force sensing stands out due to them being easier to integrate within the wearable system. However, one flaw with these sensors is that most of the times they are used as pressure sensors due to their non-stretchable characteristics. Therefore, in this study a novel stretchable Kirigami-based force sensor is presented. This force sensor was created by embroidering a silver-plated conductive onto an elastic band. To test the performance of the sensor in terms of sensitivity, linearity, hysteresis, and repeatability, three sensor samples were fabricated and stretched at a slow, medium and high speed while force data were being collected. After these tests, the sensor showed high repeatability, an average R2 of 0.9782, an average 10.27% hysteresis, and an average 0.03 V/N gauge factor. These results show that embroidered force sensors have the potential to be used to detect interaction forces within soft wearable robot-assisted therapies.

The rise of soft wearable sensors has opened the door for less obtrusive sensing during upper limb rehabilitation. Many studies have proposed different methods of fabrication for these sensors, but the simplest ones include those made using knitting, stitching, or embroidering to create resistive strain sensors. However, the reliability of these sensors is influenced by the amount of contact points of the conductive thread used at any given time. These contact points can suffer from deformations due to forces applied during each stretching cycle, which can affect the sensor response and produce erroneous measurements. These issues can be avoided by creating embroidered sensors with patterns that do not affect the contact points of the stitches. Still, forces applied directly to the conductive thread can cause irreparable damage to the sensor. Therefore, in this study a novel embroidered strain sensor is created using a honeycomb pattern. This pattern has two main purposes: a distribution of the axial forces across the walls of the pattern to protect the conductive thread; and the addition of stretchiness to the embroidered sensor. Sensors created using this pattern were embroidered onto an elastic band and then attached to a strain divider system to increase the stretchability of the sensor further. After 50 stretching cycles, sensors showed good linearity, an average gauge factor of 0.24, an average hysteresis of 36.85% and a 55.56% working range. These results show that the proposed sensor is robust to thread damages, thus making it a viable alternative for strain sensing applications.