Our study aims to construct a novel neurorehabilitation system to extract motor intention more easily and quickly by only focusing the amplitude increase/decrease by means of Fuzzy Template Matching (FTM). This way, we are free to frequency and oscillatory factors such as ERS/ERD. Here we report the development of such novel rehabilitation system for the lower limb movement, including a method of learning FTM, the mechanism of device, control design, the system setup. The consequence of the operation and verification of the rehabilitation system by measuring real human brain activity.
2. Learning Fuzzy Template Matching(L-FTM)
2.1 About L-FTM
For this study, Learning-type-Fuzzy Template Matching (L-FTM)
[15] [16] was used to classify the EEG features. In conventional fuzzy template matching [17] [18], templates are constructed using the position of the peak of the membership functions. In our developed BCI, a template was constructed with the fuzzy labels of “high” and “low” used as input values in an antecedent clause of a fuzzy rule. Because of the peculiarity of fuzzy reasoning, rules (templates) can consist of inputs of various types. In this case, 216 rules are constructed when the inputs are 16 and the number of fuzzy labels is 2, as shown in Figure 1.