Abstract
Conventional myoelectric controllers provide a mapping between electromyographic signals and prosthetic functions. However, due to a number of instabilities continuously challenging this process, an initial mapping may require an extended calibration phase with long periods of user-training in order to ensure satisfactory performance. Recently, studies on co-adaptation have highlighted the benefits of concurrent user learning and machine adaptation where systems can cope with deficiencies in the initial model by learning from newly acquired data. However, the success remains highly dependent on careful weighting of these new data. In this study, we proposed a function driven directional forgetting approach to the recursive least-squares algorithm as opposed to the classic exponential forgetting scheme. By only discounting past information in the same direction of the new data, local corrections to the mapping would induce less distortion to other regions. To validate the approach, subjects performed a set of real-time myoelectric tasks over a range of forgetting factors. Results show that directional forgetting with a forgetting factor of 0.995 outperformed exponential forgetting as well as unassisted user learning. Moreover, myoelectric control remained stable after adaptation with directional forgetting over a range of forgetting factors. These results indicate that a directional approach to discounting past training data can improve performance and alleviate sensitivities to parameter selection in recursive adaptation algorithms.
Highlights
Surface electromyography (EMG) offers a non-invasive window to the peripheral nervous system (PNS) and has been used as the control input for powered prostheses since the 1950s [1,2].In particular, myoelectric devices have been marketed towards upper-limb amputees with the appeal of providing partial functional restoration of the affected limb whilst retaining anthropomorphic aesthetics
As Shapiro–Wilk tests indicated non-normality of some results, the validity of interaction differences were confirmed by conducting the same testing on square-root transformations of the data which resulted in normality
Significant differences spread across all performance metrics and all subjects except Subject 1 were detected
Summary
Surface electromyography (EMG) offers a non-invasive window to the peripheral nervous system (PNS) and has been used as the control input for powered prostheses since the 1950s [1,2].In particular, myoelectric devices have been marketed towards upper-limb amputees with the appeal of providing partial functional restoration of the affected limb whilst retaining anthropomorphic aesthetics. Most clinically available devices still employ a simplified control scheme which restricts operation to highly unintuitive sequential activation of degrees-of-freedom (DoF). Contractions from an agonist–antagonist muscle pair drives device operation along one DoF while mode-switching to other DoFs is toggled via co-contraction or pulsing [3]. More sophisticated interpretations of residual muscle activity based on pattern-recognition (PR). Using features (in time or frequency domain, or a combination of both) extracted from multi-channel EMG data, a repertoire of analytically distinguishable contraction patterns can be learned by the system. This allows amputees to access different prosthetic functions without switching modes while controlling actuation speeds based on contraction intensities
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have