Estimation of a Cubital Joint Torque Using Surface Electromyogram Signals

This work focuses on the development of a wearable power-assisted orthosis for nursing care as a health care device to reduce the physical burden on carers. The most important aspect of developing a power-assisted orthosis is that it reflects the orthosis wearer’s intentions, coordinating with the wearer’s voluntary movements. To realize this intention, our work focuses on a torque around a cubital joint that a wearer outputs as a control signal. Generally, it is difficult to directly measure the joint torque being worn. Therefore, this paper suggests a model that estimates the cubital joint torque that the wearer outputs from surface electromyogram (SEMG) signals. As torque estimation models, we suggest and examine two models: 1) A torque estimation model using a linear sum without considering the joint angle and that assumes that the relationship between measured SEMG and relevant joint torque is linear; 2) A torque estimation model that considers the joint angle and that assumes the joint angle is a nonlinear element and builds it to the model. The results of this study clearly demonstrate that the torque estimation model that considers the joint angle can estimate the cubital joint torque with a high degree of accuracy.