High-Order Model-Free Adaptive Iterative Learning Control of Pneumatic Artificial Muscle With Enhanced Convergence

Abstract

Pneumatic artificial muscles (PAMs) have been widely used in actuation of medical devices due to their intrinsic compliance and high power to weight ratio features. However, the nonlinearity and time-varying nature of PAMs makes it challenging to maintain highperformance tracking control. In this paper, a High-Order Pseudo-Partial Derivative based Model-Free Adaptive Iterative Learning Controller (HOPPD-MFAILC) is proposed to achieve fast convergence speed. The dynamics of PAM is converted into a dynamic linearization model during iterations, meanwhile, a high-order estimation algorithm is designed to estimate the pseudo-partial derivative component of the linearization model by only utilizing the input and output data in previous iterations. The stability and convergence performance of the controller is verified through theoretical analysis. Simulation and experimental results on PAM demonstrate that the proposed HOPPD-MFA ILC can track the desired trajectory with improved convergence and tracking performance. Index Terms-Pneumatic artificial muscle, model-free adaptive control, iterative learning control, convergence. I. INTRODUCTION NEUMATIC artificial muscle (PAM) is a tube-like actuator that largely mimics biological human muscle functions [1]. Compared to traditional electrical motors and hydraulic actuators, the lightweight, high compliance and high power-to-weight ratio of PAMs [2] have fueled their popularity among assistive exoskeletons and rehabilitation robots, such as the upper limb exoskeleton series RUPERT [3] and the lower limb orthotics KAFO [4]. However, unlike the conventional actuators adopted in Lokomat [5] and ArmeoPower [6], the nonlinear and time-varying nature of PAMs may cause Manuscript

Bridging Reinforcement Learning and Iterative Learning Control: Autonomous Reference Tracking for Unknown, Nonlinear Dynamics

“…To relax requirements with respect to available model information, recent research has focused on so called datadriven ILC (DD-ILC) [43], which does not require a model of the plant. In the case of nonlinear, unknown dynamics, DD-ILC methods typically employ dynamic linearization of the plant dynamics and estimate the gradient of said linearization [44]- [46]. Alternatively, neural networks (NN) have been employed in DD-ILC to model the unknown dynamics [47], [48].…”

MISO Model Free Adaptive Control of Single Joint Rehabilitation Robot Driven by Pneumatic Artificial Muscles

“…In the case that the above two assumptions are satisfied, the MISO nonlinear discrete-time system is equivalent to (5) where represents the pseudo gradient (PG) of the system, and , . It should be noted that is bounded for any .…”

“…On the other hand, the similarity of human joints makes the robot suitable for the rehabilitation of elbow, knee and any other limb joint, which can greatly improve the utilization rate of equipment. However, the special structure of PAMs determine that they have nonlinear and time-varying characteristics, which makes it very difficult to control the robot driven by PAMs [5]. A large number of people have been involved in related research, using different control methods to control PAM and the robot driven by PAMs.…”

Data‐driven urban traffic model‐free adaptive iterative learning control with traffic data dropout compensation

“…Therefore, the MFAILC scheme has both data-driven and iterative learning features, which motivated the application of MFAILC in urban traffic control field [38,39]. Other practical applications of MFAILC includes multi-agent systems [40,41] and pneumatic artificial muscle [42].…”