By Elang Parikesit
Year 2023
Abstract
In this research, a low-cost lower limb exoskeleton has been developed for those who struggle with their lower legs, the exoskeleton works as a gait trainer. It is crucial for patients’ lower extremity rehabilitation since it provides aid in their physical recovery. Although, there are robotic assisted gait trainers available in the market, the cost is high. Hence the objective of the study is to create a low-cost lower-limb exoskeleton for gait training with good performance. This research proposes a controller for a robot-assisted gait trainer’s lower limb exoskeleton. The study focuses on 1) development of mechanism of two joints (hip and knee) of a lower-limb exoskeleton for gait training, 2) development of an electric motor driver for two joints (hip and knee) of a lower-limb exoskeleton for gait training, 3) development of control algorithm for two joints (hip and knee) of a lower-limb exoskeleton for gait training, and 4) an evaluation of the control algorithm’s performance on the lower-limb exoskeleton of the robot-assisted gait training.
The human gait movement is not linear. Proportional-Integral Derivative (PID) appeared as a popular controller does not work on nonlinear system either. As a consequence, the PID could not be used in this system. Two experiments on controlling the gait trainer by using two different controllers were carried out: the one using hybrid PID-ILC and the other using Computed Torque Controller (CTC). The former, PID-ILC was an unmodelled control algorithm while the latter, CTC was a modelled controlled algorithm. The computed torque controller was a vital nonlinear controller when the system’s dynamic parameters were merely partially understood. A mathematical model was the foundation of this kind of controller. This proposed control was evaluated employing a scaled-down model. The methodological instruments used to achieve these goals were 1) an investigation and data collection using updated references, 2) a design and development of hardware and software of the system, and 3) a testing and a performance evaluation of the system.
An inexpensive robot-assisted gait trainer has been produced successfully. In the first experiment the robot-assisted gait trainer could follow gait trajectories with the support of the proposed hybrid PID-ILC controller even when there were unmodelled dynamics, uncertainty, and distractions. Real studies utilizing a specific controller load and gain showed that the PID-controlled system had possessed stability, but with an error of up to 10 degrees at steady state. Stability was demonstrated by the recommended hybrid PID-ILC controller. Initially the PID-ILC had steady state errors, notwithstanding after more than ten iterations, less than 1 degree of steady state error was attained. Up to 50% of the steady state error was greatly reduced. The performance of CTC was also evaluated and it was contrasted with that of the PID controller. CTC had better responses than PID for both joint 1 and joint 2. Nonetheless, for the downward direction, the response of joint 2 using CTC was not as good as PID.