Echo State Networks as Novel Approach for Low-Cost Myoelectric Control

Abstract

Myoelectric signals (EMG) provide an intuitive and rapid interface for controlling technical devices, in particular bionic arm prostheses. However, inferring the intended movement from a surface EMG recording is a non-trivial pattern recognition task, especially if the data stems from lowcost sensors. At the same time, overly complex models are prohibited by strict speed, data parsimony and robustness requirements. As a compromise between high accuracy and strict requirements we propose to apply Echo State Networks (ESNs), which extend standard linear regression with 1) a memory and 2) nonlinearity. Results show that both features, memory and nonlinearity, independently as well as in conjunction, improve the prediction accuracy on simultaneous movements in two degrees of freedom (hand opening/closing and pronation/supination) recorded from four able-bodied participants using a low-cost 8-electrode-array. However, it was also shown that the model is still not sufficiently resistant to external disturbances such as electrode shift.

Counteracting Electrode Shifts in Upper-Limb Prosthesis Control via Transfer Learning

“…In recent years, machine learning control for upper limb prostheses has made significant progress, driven by new control algorithms (Janne M. Hahne, Biebmann, et al 2014;Ning Jiang et al 2014;Muceli, I. Vujaklija, et al 2017;Prahm, Schulz, Paaßen, et al 2017; Aidan D. Roche et al 2014), new training paradigms, such as co-adaptive training, virtual reality, and games (J. M. Hahne et al 2015;Prahm, Ivan Vujaklija, et al 2017; Aidan D. Roche et al 2014), new surgical techniques, such as targeted muscle reinnervation (Todd et al 2009; Aidan D. Roche et al 2014), new prosthetic devices (Belter et al 2013;Controzzi et al 2017), and new electrodes to record user's control signal, such as highdensity electrode grids (Daley et al 2012;Muceli, N. Jiang, and D. Farina 2014) or implantable sensors (Janne M. Hahne, Dario Farina, et al 2016;Ortiz-Catalan et al 2012;Pasquina et al 2015). However, translating many promising results from the lab to an amputee's everyday life remains a challenge due to various sources of disturbance, such as posture changes, sweating, weight of grasped objects, long term changes, or electrode shifts (D. Farina et al 2014;L.…”