Hardware-in-the-loop study of a hybrid active force control scheme of an upper-limb exoskeleton for passive stroke rehabilitation
The burden of stroke has necessitated the employment of robotics to mitigate the inability of physiotherapists to cope with the increasing demand for rehabilitation by stroke survivors. Continuous passive motion training has been demonstrated to be able to allow stroke patients to regain their mobil...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/24981/ http://umpir.ump.edu.my/id/eprint/24981/ http://umpir.ump.edu.my/id/eprint/24981/1/Hardware-in-the-loop%20study%20of%20a%20hybrid%20active%20force%20control%20scheme.pdf |
| Summary: | The burden of stroke has necessitated the employment of robotics to mitigate the inability of physiotherapists to cope with the increasing demand for rehabilitation by stroke survivors. Continuous passive motion training has been demonstrated to be able to allow stroke patients to regain their mobility. Furthermore, this form of rehabilitation is non-trivial particularly in the acute and sub-acute phase of neurorehabilitation. This thesis aims at evaluating a class of robust control scheme, namely active force control (AFC) on a two degrees of freedom upper-limb exoskeleton that is able to compensate disturbances arising from different upper-limb weights that are unique for different individuals without the need for further re-tuning. In order to evaluate the efficacy of the proposed controller, a simulation investigation was performed. The dynamics of the system are derived based on the Euler-Lagrange formulation by incorporating anthropometric measurements of the human upper limb. The efficacy of the proposed controllers, namely classical Proportional-Derivative AFC (PDAFC) architecture optimised by means of fuzzy logic (FL), artificial neural network (ANN), particle swarm optimisation (PSO) and simulated Kalman filter (SKF) against classical PD control in mitigating different disturbance configurations (no disturbance, constant disturbance of 30 N.m. and harmonic disturbance of 30 N.m. at a frequency of 10 Hz at different speeds, i.e., slow (0.375 rad/s), medium (0.430 rad/s) and fast (0.502 rad/s) of a typical rehabilitation trajectory for the shoulder and elbow joints were evaluated. It is shown from the simulation investigation that the PDSKFAFC scheme is better in comparison to all the evaluated schemes, particularly the classical PD control scheme. A data-driven model is developed based on the exoskeleton prototype built. A hardware-in-the-loop simulation is carried out to evaluate the appropriate gains of both the PD and the AFC inertial parameter gained that is tuned via the SKF algorithm. It is demonstrated through the experimental works, that the PDSKFAFC scheme is able to compensate against the disturbance attributed by the attached mannequin mass of the upper arm (2 kg) and forearm (1.5 kg), respectively to the exoskeleton prototype in comparison the classical PD scheme. |
|---|