How event-driven intermittent control with unstable open & closed-loop dynamics lead to bounded response in human postural control

Abstract

Event-driven intermittent feedback control is a form of feedback control in which the variables of interest are observed continuously, but corrective control action is only initiated intermittently based on certain threshold criteria. According to the literature, the human central nervous system (CNS) adopts event-driven intermittent control strategy while performing various tasks such as stabilize upright posture, control of saccadic eye movement or while driving. In this paper, we examine whether event-driven intermittent control when applied to a system wherein both open-loop dynamics and continuous closed-loop dynamics are unstable can yield bounded overall response (as has been observed in some situations of human postural control). With the help of both illustrative example simulations and experimental results from a table-top experiment inspired from human postural control, we show that this is indeed possible. We further provide insights using passivity theory and phase space analysis. Our analysis shows that bounded response is possible even if the phase trajectories of unstable open-loop and unstable closed-loop systems do not diverge in opposite directions or when Hurwitz convex combinations are not feasible. This study does not necessitate any assumption regarding the stabilizability or passivity of any of the constituent subsystems.