Model-Based Mid-Level Regulation for Assist-As-Needed Hierarchical Control of Wearable Robots: A Computational Study of Human-Robot Adaptation

Abstract

The closed-loop human–robot system requires developing an effective robotic controller that considers models of both the human and the robot, as well as human adaptation to the robot. This paper develops a mid-level controller providing assist-as-needed (AAN) policies in a hierarchical control setting using two novel methods: model-based and fuzzy logic rule. The goal of AAN is to provide the required extra torque because of the robot’s dynamics and external load compared to the human limb free movement. The human–robot adaptation is simulated using a nonlinear model predictive controller (NMPC) as the human central nervous system (CNS) for three conditions of initial (the initial session of wearing the robot, without any previous experience), short-term (the entire first session, e.g., 45 min), and long-term experiences. The results showed that the two methods (model-based and fuzzy logic) outperform the traditional proportional method in providing AAN by considering distinctive human and robot models. Additionally, the CNS actuator model has difficulty in the initial experience and activates both antagonist and agonist muscles to reduce movement oscillations. In the long-term experience, the simulation shows no oscillation when the CNS NMPC learns the robot model and modifies its weights to simulate realistic human behavior. We found that the desired strength of the robot should be increased gradually to ignore unexpected human–robot interactions (e.g., robot vibration, human spasticity). The proposed mid-level controllers can be used for wearable assistive devices, exoskeletons, and rehabilitation robots.

Evaluation of a machine-learning-driven active–passive upper-limb exoskeleton robot: Experimental human-in-the-loop study

“…Here, we are using a machine-learning model from Nasr et al (Nasr et al, 2021a) to estimate the user intent from the IMU and sEMG sensors. The mid-level unit transforms the high-level estimation and prediction into mode-specific reference trajectories, or the desired wrench, using the biomechatronic system’s dynamic equations. In the past, this level of control was achieved using a conventional control method: finite state machines/prerecorded motion (Long et al, 2017), master–slave (Lee et al, 2012), proportional (Tang et al, 2014), CTM (Nasr et al, 2022b), fuzzy-logic (Han et al, 2021; Nasr et al, 2022b), impedance control (Brahmi et al, 2021), haptic/admittance control (Menga and Ghirardi, 2019), or adaptive control (Nasiri et al, 2021). Recently, positive results for the mid-level controller have been demonstrated using the AAN-CTM approach, which augments rather than replaces muscular activity (Nasr et al, 2022b).…”

“…In the past, this level of control was achieved using a conventional control method: finite state machines/prerecorded motion (Long et al, 2017 ), master–slave (Lee et al, 2012 ), proportional (Tang et al, 2014 ), CTM (Nasr et al, 2022b ), fuzzy-logic (Han et al, 2021 ; Nasr et al, 2022b ), impedance control (Brahmi et al, 2021 ), haptic/admittance control (Menga and Ghirardi, 2019 ), or adaptive control (Nasiri et al, 2021 ). Recently, positive results for the mid-level controller have been demonstrated using the AAN-CTM approach, which augments rather than replaces muscular activity (Nasr et al, 2022b ). The low-level agent computes the error between the measured states (from the kinematic and kinetic sensors) and the desired device states (from the mid-level controller) and uses common control algorithms like proportional-integral-derivative (PID) control.…”

“…Modern exoskeleton, prosthetic, assistive, and rehabilitation robot controls follow a hierarchical structure with control laws at the high, middle, and low levels (Young and Ferris, 2017 ; du Plessis et al, 2021 ; Nasr et al, 2022b ). The high-level controller deciphers the user’s motion intent or user wrench (torque/force).…”

“…The mid-level controller converts the high-level controller’s intended motion or wrench into the low-level controller’s desired trajectory of motion or wrench. The assist-as-needed (AAN)-computed-torque method (CTM approach for the mid-level controller, which relies on augmenting rather than substituting motor effort, has shown encouraging outcomes (Gull et al, 2020 ; Yihun et al, 2020 ; Asl et al, 2021 ; Wang et al, 2021 ; Nasr et al, 2022b ). So far, this mid-level controller has been tested in simulation studies.…”

“…So far, this mid-level controller has been tested in simulation studies. Although a comprehensive and verified model may be used for controller design (Nasr et al, 2022b ), it is essential to test the controller in HITL experiments.…”