Motion intention estimation of lower limbs based on sEMG supplement with acceleration signal

Lower extremity exoskeleton robot can assist the person standing and walking which are important functions for the disabled or old people who can not make move by themselves. The priority task for exoskeleton robot is to get the movement intentions of wearer. This paper proposes an intention estimation method of lower limbs motion based on multi-types signals including surface electromyography (sEMG) and 3-axis acceleration data. 5 channels sEMG and 3-axis acceleration were collected at the 5 same points from able-bodied and the disabled people respectively. After preprocessed and normalized, different features were extracted from the obtained signals. Support vector machine (SVM) was utilized for motion classification, where features of sEMG signals and acceleration signals were taken as input respectively. We also tested the fusion features of the both signals. Furthermore, compared experiments were carried for the disabled and normal people. Results demonstrated that the proposed method was effective for able-bodied people, while the accuracy of the method for disabled people need to be further improved.