Myoelectric co-adaptation: stochastic and chaotic, yet describable
Loading...
Files
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of New Brunswick
Abstract
Myoelectric control systems have advanced over decades, but their performance remains a challenge. Adaptive algorithms have emerged as a possible solution; however, the user is also always learning, causing the resulting system's performance to sometimes be subpar and even unstable. Most studies approximate these co-adaptive systems as combinations of deterministic and stochastic linear systems. Instead, we propose that chaos, a subtype of nonlinear dynamical systems, is required to explain the intermittent performance of myoelectric control systems. Slight, unknown, changes to a chaotic system can shift its dynamics between stable and unstable. In this work, a 10-day co-adaptive myoelectric control experiment was designed to explore this chaotic behaviour. Poincaré mapping, a technique used to visualize chaos, was used to observe how changes to experimental parameters influence the stability of co-adaptive systems and user learning. A better understanding of the dynamics of co-adaptive human-machine interfaces may enable the design and control of more robust systems in the future.