반복 학습과 퍼지 로직에 의한 n-링크 최적 제어

In this paper, we propose a motion control method for a robot by iterative learning of its practical motion. The preexistence motion control methods usually use algorithms based on mathematics and dynamics of models. However, it is almost impossible for a model designer to apply specific algorithms derived after numerous simulations of a robot"s motion considering in many situations. Therefore, we propose a motion control method similar to animals" method which are motivated by their movement on the earth. It is a modeling of a method about controlling their muscles based on the memory acquired from birth about how to control their body well. That is, we apply various models of muscles controled in many situations of gravity. Remembering their experiences, we apply part of process of making many different motions based on the stored memory to basic n-link control that is coupled with robot"s body. Motion control that dose not exist in the memory is replaced with the method of fuzzy reasoning from several combinations of other situations in the memory.