Bio-robotics research for non-invasive myoelectric neural interfaces for upper-limb prosthetic control: a 10-year perspective review

Abstract

A decade ago, a group of researchers from academia and industry identified a dichotomy between the industrial and academic state-of-the-art in upper-limb prosthesis control, a widely used bio-robotics application. They proposed four key technical challenges, if addressed, could bridge this gap and translate academic research into clinically and commercially viable products. These challenges are unintuitive control schemes, lack of sensory feedback, poor robustness, and single sensor modality. Here, we provide a perspective review on the research effort that occurred in the last decade, aiming at addressing these challenges. In addition, we discussed three research areas essential to the recent development in upper-limb prosthetic control research but were not envisioned in the review ten years ago: deep learning methods, sEMG decomposition and open-source databases. To conclude the review, we provided an outlook into the near future of the research and development in upper-limb prosthetic control and beyond.

Snake Robot with Motion Based on Shape Memory Alloy Spring-Shaped Actuators

“…The development of nonclassical robotic systems has been boosted by the increase in application areas where the use of rigid robotic structures is not feasible [ 1 ]. There exist several problems of this kind, such as endoscopic applications, the manipulation of objects in restricted spaces, the implementation of prosthesis devices, the exploration of closed spaces, and so on [ 2 , 3 ].…”

Decoding Multi-DoF Movements Using a CST-Based Force Generation Model With Single-DoF Training

“…However, the practical applications of prosthetic hands are hindered due to lacking of a robust humanmachine interface with intuitive and flexible control systems [3]. In the past decades, surface electromyogram (sEMG) played an essential role in establishing control interfaces because of its convenience and non-invasiveness [4]. The sEMG signals are regarded as the summation of motor unit action potentials (MUAPs) containing the neural control information from the central nervous system (CNS) [5].…”

“…However, these methods require a large training set that includes all possible task patterns [9]- [12]. Consequently, the training phase becomes extensive and cumbersome, posing a significant hurdle to the practical application of prosthetics [4], [13]. This arduous and time-consuming training process significantly increases user's burden.…”

From Forearm to Wrist: Deep Learning for Surface Electromyography-Based Gesture Recognition

“…In the last decade, deep learning methods have gained increased attention and become an effective tool for processing and decoding sEMG signals for gesture recognition [13], [14] [16]. To reduce the number of parameters and training data, Tsinganos et al proposed a temporal convolutional network structure and showed higher classification accuracy than that of CNN on a public sEMG dataset [17].…”

“…They were all representative structures over the development of deep learning and employed for forearm sEMG-based gesture recognition in previous studies [21]. CNN was widely used in biosignal processing for extracting spatial information [14]. GRU, TCN, and Transformer were all proposed in the last decade, introducing different mechanisms to improve the accuracy and efficiency of sequential data [21].…”