원문정보
초록
영어
Robot-assisted self-rehabilitation for patients with stroke is significant for their motor recovery. Meanwhile, the surface EMG can reflect human neuromuscular activity and can be used for rehabilitation robot control. In this paper, we propose a Two-Step SVM classification method based on One-versus-One SVM muti-class classification method in order to improve the time efficiency of upper limb motion classification by sEMG, and then promote the real- time control of upper limb rehabilitation robot. A control experiment is done between the Two-Step SVM classification method and the One-versus-One SVM method. Four muscles in human upper limb are chosen to train and six motions are to recognize according to the aim of rehabilitation and the characteristic of people’s daily life. The classifier training and motion recognize times between these two methods are compared. The result shows, the Two-Step SVM classification method proposed is improved in time efficiency, which is meaningful to improve the real time control of the robot during the process of rehabilitation.
목차
1. Introduction
2. Basic Research and Concepts
2.1. Upper Limb Rehabilitation Robot System
2.2. Surface Electromyography
2.3. Feature Extraction
2.4. Support Vector Machine
3. Multi-classification Methods of SVM
3.1. OVO-SVM Classification Method
3.2. Two-Step SVM Classification Method
4. Experiment and Result
4.1. Data Acquisition and Feature Extraction
4.2. Motion Classification
4.3 Result and Discussion
5. Conclusion
Acknowledgements
References