Fuzzy vs. non-fuzzy rule base for gait event detection

A method for the detection of gait events in the electrically-stimulated walking of paraplegic subjects was developed using fuzzy system identification methods. The efficacy of various amounts of “fuzziness” (i.e., overlap of sensor measurement membership functions) of the system mas explored. The detection was tested on data from three subjects. Comparison of the results found using fuzzy logic with those found using non-fuzzy rules, in the form of a lookup table, demonstrated that a moderate amount of fuzziness resulted in better detection of the phases for all subjects. This finding was confirmed by both qualitative and quantitative analysis. The performance of the detection varied among the subjects and appeared to be related to the quality of the subjects´ gait