Towards Control of a Transhumeral Prosthesis with EEG Signals

Abstract

Robotic prostheses are expected to allow amputees greater freedom and mobility. However, available options to control transhumeral prostheses are reduced with increasing amputation level. In addition, for electromyography-based control of prostheses, the residual muscles alone cannot generate sufficiently different signals for accurate distal arm function. Thus, controlling a multi-degree of freedom (DoF) transhumeral prosthesis is challenging with currently available techniques. In this paper, an electroencephalogram (EEG)-based hierarchical two-stage approach is proposed to achieve multi-DoF control of a transhumeral prosthesis. In the proposed method, the motion intention for arm reaching or hand lifting is identified using classifiers trained with motion-related EEG features. For this purpose, neural network and k-nearest neighbor classifiers are used. Then, elbow motion and hand endpoint motion is estimated using a different set of neural-network-based classifiers, which are trained with motion information recorded using healthy subjects. The predictions from the classifiers are compared with residual limb motion to generate a final prediction of motion intention. This can then be used to realize multi-DoF control of a prosthesis. The experimental results show the feasibility of the proposed method for multi-DoF control of a transhumeral prosthesis. This proof of concept study was performed with healthy subjects.

Real-Time EMG Based Pattern Recognition Control for Hand Prostheses: A Review on Existing Methods, Challenges and Future Implementation

“…Electroencephalogram (EEG), and ECog (Electrocortocogram) measures brain signals, and they could be used to supersede EMG for prostheses control. ECog electrodes are invasive as they are placed directly inside the head, whereas EEG electrodes are non-invasive, as they are positioned on the scalp area [100], where information regarding the targeted body movements are measurable [101]. EEG and ECog have currently found an application as a brain-machine interface [102], and in theory, can control the movement of the prosthesis similar to the EMG.…”

Modeling-Based EMG Signal (MBES) Classifier for Robotic Remote-Control Purposes

“…There are many ways that can be used to classify EMG signals, such as Neural Networks [37], Fuzzy Logic [38], Support Vector Machine [39] and other computational techniques [18,40]. All of these methods calculate some features (mean value, RMS value, etc.)…”

“…A second class of research activities adopts electroencephalograph (EEG) signals, the expression of electrical activity in the brain: in [16], EEG signals were used to control the e-Puck robot to carry the food of a rat; in [17], to control a wheelchair to move through a building whose five floors are connected by an elevator. Other examples of use of biological signals are in different kinds of rehabilitation and medical assistive devices (prostheses and orthoses), humanoid robots and industrial robots [18][19][20][21][22][23]. Electromyographic (EMG) signals, an expression of electrical activity of muscles, were widely adopted in several applications: to control the line tracking of a mobile robot [24]; to control an exoskeleton, powered by pneumatic muscles, that supports the back while performing weightlifting movements [25]; in combination with an inertial measurement unit (IMU) sensor, to control a mobile robot based on gesture recognition [26]; in combination with electro-oculography (EOG), EEG, vision systems and head movements, to control a robotic arm [27] or in combination with a Kinect device [28].…”

Advances in robotic prostheses

“…It nevertheless continues to be the topic of a significant research effort. An example is work at Kyushu University in which an EEG-based, two-stage hierarchical approach has been developed with the potential to control a multi-degrees of freedom (DOF) transhumeral prosthesis (Bandara et al, 2018). The intention for reaching with an arm or lifting with a hand is identified using neural network (NN) and k-nearest neighbour (k-nn) classifiers trained with motion-related EEG features.…”