Power-assist robots are expected to work in many fields such as industry, military, medicine, etc. A lower-limb power-assist robot for physically weak persons is supposed to be used for self-rehabilitation or daily motion assist. In order to assist daily motion of the physically weak persons, the robot must estimate the motion intention of the user in real-time. Although there are several kinds of method to estimate the motion intention of the user in real-time, Electromyogram (EMG) signals are often used to estimate that since they reflect the users muscle activities. However, EMG-based real-time motion estimation is not very easy because of several reasons. In this chapter, an EMG-based control method is introduced to control the power-assist lower-limb exoskeleton robot in accordance with users motion intention. A neuro-fuzzy modifier is applied to deal with those problems. The problems of EMG-based motion estimation are cleared by applying the neuro-fuzzy modifier. Sometimes there is a problem in the users motion even though the users motion is assisted, if the user misunderstands interaction between the users motion and a surrounding environment. In that case, the users motion should be modified to avoid an accident. In this chapter, a method of perception-assist is also introduced to automatically modify the users motion properly.
All Science Journal Classification (ASJC) codes