Biofied building is a building system that provides safe, secure and comfortable space for residents using a robot. The robot acquires the motions and positions of residents. However, it is difficult to control the robot not to disturb the residents’ activities. The purpose of this research is to incorporate motion prediction in addition to motion recognition. Many of the conventional methods use RNN (Recurrent Neural Networks), which is difficult to implement due to the necessity of big data and large calculation cost. Thus, we aim to propose a simple motion prediction method using a human body model based on dynamics.