November 01 (Wednesday)
바이오공학부문 포스터
Poster,
3층 로비,
15:50~16:40
  • Chair :
  •  신현정(KAIST), 조영삼(원광대), 김성진(건국대), 양성욱(KIST)
We16D-9
15:50~16:40
평지에서 계단전환 보행을 구분하는 검출 알고리즘 개발

김판권(서강대학교), 이진규(서강대학교 기계공학과), 신충수(서강대학교)
The purpose of this study was to develop a detection algorithm for classification of level walking (LW) and transition walking from level-to-stair (TW-LS) based on plantar pressure. Thirteen subjects performed LW, transition walking from level to stair ascent (TW-LSA) and transition walking from level to stair descent (TW-LSD). The stance time, vertical ground reaction force (vGRF), anteroposterior (AP) and mediolateral (ML) center of pressure (COP) at initial contact (IC) and AP/ ML range of COP were calculated based on plantar pressure during stance phase. The data were evaluated statistically by one-way analysis of variance (ANOVA) and post-hoc comparison was performed at significance level of 0.05. The stance time, AP COP at IC and AP range of COP were significantly different in the comparison among three walking conditions (all, p<0.001). The ML COP at IC was significantly different when TW-LSA and TW-LSD were compared by LW (both, p<0.05). The peak vGRF was significantly different when LW and TW-LSD were compared by TW-LSA (both, p<0.05). The multinomial logistic regression models were developed with various combinations of parameters based on comparison results. The accuracy of the model was the highest when the combination of stance time, AP COP at IC and AP range of COP were used. As the result, the probability of classifying LW and TW-LS was 0.952. In conclusion, the multinomial logistic regression model based on plantar pressure can be used for classifying LW and TW-LS with high accuracy.
Keywords : detection algorithm(검출 알고리즘), transition walking(전환 보행), multinomial logistic regression model(다항 로지스틱 회귀모델), stair walking(계단보행), level walking(평지보행)
Paper : We16D-9.pdf

(사)대한기계학회, 서울시 강남구 테헤란로 7길 22 한국과학기술회관 신관 702호, Tel: (02)501-3646~3648, E-mail: ksme@ksme.or.kr