Recognition of hand movements in a trans-radial amputated subject by sEMG

Abstract

Trans-radially amputated persons who own a myoelectric prosthesis have currently some control via surface electromyography (sEMG). However, the control systems are still limited (as they include very few movements) and not always natural (as the subject has to learn to associate movements of the muscles with the movements of the prosthesis). The Ninapro project tries helping the scientific community to overcome these limits through the creation of electromyography data sources to test machine learning algorithms. In this paper the results gained from first tests made on an amputated subject with the Ninapro acquisition protocol are detailed. In agreement with neurological studies on cortical plasticity and on the anatomy of the forearm, the amputee produced stable signals for each movement in the test. Using a k-NN classification algorithm, we obtain an average classification rate of 61.5% on all 53 movements. Successively, we simplify the task reducing the number of movements to 13, resulting in no misclassified movements. This shows that for fewer movements a very high classification accuracy is possible without the subject having to learn the movements specifically.

Fully embedded myoelectric control for a wearable robotic hand orthosis

“…Compared to conventional EMG data acquisition systems (as used in [15]- [17]) the Myo armband is highly convenient to use. The dry electrodes do not require exact placement and can be placed all at once, such that stroke survivors themselves can don the Myo armband.…”

“…Once the Myo armband is donned, the user is guided through a training routine, where the classifier is trained with desired gestures. Such training is necessary for every state of the art EMG system [15]- [18], as EMG is not an absolute measure and changes for different users and different sessions (electrode placement, fatigue of the user, etc.) [21].…”

“…In order to cover the large variety in human grasp types, several gestures have to be recognized [14]. Surface electromyography (sEMG) based algorithms have been used to classify up to 53 grasp types at 61.5% overall accuracy [15]. However, EMG systems typically use multiple electrodes [15]- [17] or electrode arrays [18], which require cumbersome cabling, conductive gels and experience in electrode placement, making them hardly applicable in a home rehabilitation scenario.…”

“…Surface electromyography (sEMG) based algorithms have been used to classify up to 53 grasp types at 61.5% overall accuracy [15]. However, EMG systems typically use multiple electrodes [15]- [17] or electrode arrays [18], which require cumbersome cabling, conductive gels and experience in electrode placement, making them hardly applicable in a home rehabilitation scenario. The commercially available Myo armband (Thalmic Labs, Kitchener, Ontario, Canada) is a lightweight and easy to use alternative to state of the art EMG data acquisition systems.…”

Classifying sEMG-based hand movements by means of principal component analysis

“…This database is a result of NINAPRO project [13] and can be downloaded from NINAWEB -the web interface of the NINAPRO project [14]. Two versions of the database were presented, as well as the various techniques used for data preprocessing, feature extraction, and hand posture classification [2], [15]- [16]. One of the techniques offered by NINAPRO project is the use of Principal Component Analysis (PCA) for feature extraction.…”