Year: 2024
Type: conference-paper
Abstract: In the field of stroke rehabilitation, lower limb exoskeletons serve as remarkable assistive tools, and the optimization of their control algorithms is crucial. Traditional intent recognition techniques, however, fail to meet the precise identification needs for patients’ voluntary movements. To address this challenge, this paper establishes a comprehensive exoskeleton software-hardware control system, which can anticipate and interpret patients’ movement intentions by capturing and processing ultrasound signals. Building upon this system, this paper employs an intent recognition method using Amplitude-mode ultrasound to perceive deep muscle movements in patients who suffer from stroke, measure muscle thickness variations, and predict joint angles effectively. Subsequently, these signals are translated into exoskeleton motion commands through this software–hardware system. Through simulated motion impeding experiments, this paper demonstrates the method's capability to capture voluntary exertion intentions when the contraction and relaxation of the experimenter's muscles cannot lead to the relative limb displacement. In the experiment focused on knee flexion-extension, the achieved average accuracy reaches 87.265%, and the average absolute error consistently remains within 6% of the range of motion. The method proposed in this paper holds substantial potential in the field of rehabilitative exoskeletons.
Authors: Ling Xing, Zhiwei Hu, Yongchao Yin
Institution: University of Science and Technology of China
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Awards: G3680183638, G5036913079
OpenAlex ID: W4404978425