Publicação
Feedback-error learning for gait rehabilitation using a powered knee orthosis: first advances
| Resumo: | Powered assistive devices have been playing a major role in gait rehabilitation. Hereby, the development of time-effective control strategies to manage such devices is a key concern to rehabilitation engineering. This paper presents a real-time Feedback-Error Learning control strategy, by means of an Artificial Neural Network as a feedforward controller to acquire the inverse model of the plant, and a Proportional-Integral-Derivative feedback controller to guarantee stability and handle with disturbances. A Powered Knee Orthosis was used as the assistive device and a trajectory generator assistive strategy, previously acquired through an inertial system, was applied. A validation with one subject walking in a treadmill at 1 km/h with the Powered Knee Orthosis controlled by the Feedback-Error Learning control was performed. Evidences on the control behavior presented good performances, with the Artificial Neural Network taking 90 seconds to learn the inverse model, which enabled a decrease in the angular position error by 75% and eliminated the phase delay, when compared to solo Proportional-Integral-Derivative feedback controller. Robust reactions to external disturbances were also achieved. The implemented Feedback-Error Learning strategy proves to be a time-effective asset to control assistive powered devices. |
|---|---|
| Autores principais: | Fernandes, Pedro Nuno |
| Outros Autores: | Carvalho, Simao; Figueiredo, Joana; Moreno, Juan C.; Santos, Cristina |
| Assunto: | Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática |
| Ano: | 2019 |
| País: | Portugal |
| Tipo de documento: | comunicação em conferência |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | Powered assistive devices have been playing a major role in gait rehabilitation. Hereby, the development of time-effective control strategies to manage such devices is a key concern to rehabilitation engineering. This paper presents a real-time Feedback-Error Learning control strategy, by means of an Artificial Neural Network as a feedforward controller to acquire the inverse model of the plant, and a Proportional-Integral-Derivative feedback controller to guarantee stability and handle with disturbances. A Powered Knee Orthosis was used as the assistive device and a trajectory generator assistive strategy, previously acquired through an inertial system, was applied. A validation with one subject walking in a treadmill at 1 km/h with the Powered Knee Orthosis controlled by the Feedback-Error Learning control was performed. Evidences on the control behavior presented good performances, with the Artificial Neural Network taking 90 seconds to learn the inverse model, which enabled a decrease in the angular position error by 75% and eliminated the phase delay, when compared to solo Proportional-Integral-Derivative feedback controller. Robust reactions to external disturbances were also achieved. The implemented Feedback-Error Learning strategy proves to be a time-effective asset to control assistive powered devices. |
|---|