Surface EMG-Based Hand Gesture Recognition via Hybrid and Dilated Deep Neural Network Architectures for Neurorobotic Prostheses

Abstract

Motivated by the potentials of deep learning models in significantly improving myoelectric control of neuroprosthetic robotic limbs, this paper proposes two novel deep learning architectures, namely the [Formula: see text] ([Formula: see text]) and the [Formula: see text] ([Formula: see text]), for performing Hand Gesture Recognition (HGR) via multi-channel surface Electromyography (sEMG) signals. The work is aimed at enhancing the accuracy of myoelectric systems, which can be used for realizing an accurate and resilient man–machine interface for myocontrol of neurorobotic systems. The HRM is developed based on an innovative, unconventional, and particular hybridization of two parallel paths (one convolutional and one recurrent) coupled via a fully-connected multilayer network acting as the fusion center providing robustness across different scenarios. The hybrid design is specifically proposed to treat temporal and spatial features in two parallel processing pipelines and to augment the discriminative power of the model to reduce the required computational complexity and construct a compact HGR model. We designed a second architecture, the [Formula: see text], as a compact architecture. It is worth mentioning that efficiency of a designed deep model, especially its memory usage and number of parameters, is as important as its achievable accuracy in practice. The [Formula: see text] has significantly less memory requirement in training when compared to the HRM due to implementation of novel dilated causal convolutions that gradually increase the receptive field of the network and utilize shared filter parameters. The NinaPro DB2 dataset is utilized for evaluation purposes. The proposed [Formula: see text] significantly outperforms its counterparts achieving an exceptionally-high HGR performance of [Formula: see text]%. The TCNM with the accuracy of [Formula: see text]% also outperforms existing solutions while maintaining low computational requirements.

FS-HGR: Few-Shot Learning for Hand Gesture Recognition via Electromyography

“…A common strategy used for hand gesture recognition in recent works is applying DNN with the focus on improving hand gestures classification performance on "never-seenbefore repetitions". Along this line of research, several stateof-the-art works [10]- [12], [14], [16]- [22] mainly used the Ninapro database [23]- [25], which is a public dataset providing kinematic and sEMG signals from 52 finger, hand, and wrist movements. The Ninapro database is similar to data obtained in real-world conditions, and as such it allows development of advanced DNN-based recognition frameworks.…”

“…The common approach in recent studies [10]- [12], [14], [16]- [22], following the recommendations provided by the Ninapro database, is to train DNN-based models on a training set consisting of approximately 2/3 of the gesture trials of each subject. The evaluation is then performed on the remaining trials constituting the test set.…”

“…The first FC layer in the FC Embedding Module is used to increase the input dimensional to (W × 128). Subsequently, the second (which is followed by ReLU activation function) and third FC layers with output size of 100 and 1, respectively, are adopted to re- 2) The TemporalBlock Module: Inspired by [10], [12], [26], [35], [36], the proposed FS-HGR few-shot learning architecture utilizes Dilated Causal 1D-Convolutions over the temporal dimension. The proposed architecture, therefore, provides several advantages over Recurrent Neural Networks (RNNs) such as low memory requirement and faster training.…”

A Multimodal Intention Detection Sensor Suite for Shared Autonomy of Upper-Limb Robotic Prostheses

“…The captured image is enhanced using contrast correction, then converted to grey scale and subsequently processed for edge detection purposes. While more complex techniques are available [34,35], the computational intensity and challenges in implementing them into embedded code systems on board of the Bebionic hand dictated the implementation of leaner algorithms. In this paper, we utilized light image processing approaches to simplify the computation and maximize practicality and translation to the clinic.…”

Few-Shot Learning for Decoding Surface Electromyography for Hand Gesture Recognition

“…Recent evolution in Machine Learning (ML) and Deep Neural Networks (DNNs) coupled with advancements of neuro-rehabilitation technologies, have paved the way for the development of new control systems for myoelectric prostheses. In this regard, surface Electromyogram (sEMG) signals [1][2][3][4][5] is, typically, considered as the input for the control system of prostheses. In particular, the use of machine intelligence for sEMG-based Hand Gesture Recognition (HGR) has been the focus of literature due to its unique potentials to improve the functionality of control and consequently increase the quality of life of individuals with the lack of a biological limb.…”