In the present paper, we suggest an algorithm for a sensor agent robot with a laser range finder to recognize the flows of residents in the living spaces in order to achieve flow recognition in the living spaces, recognition of the number of people in spaces, and the classification of the flows. House reform is or will be demanded to prolong the lifetime of the home. Adaption for the individuals is needed for our aging society which is growing at a rapid pace. Home autonomous mobile robots will become popular in the future for aged people to assist them in various situations. Therefore we have to collect various type of information of human and living spaces. However, a penetration in personal privacy must be avoided. It is essential to recognize flows in everyday life in order to assist house reforms and aging societies in terms of adaption for the individuals. With background subtraction, extra noise removal, and the clustering based k-means method, we got an average accuracy of more than 90% from the behavior from 1 to 3 persons, and also confirmed the reliability of our system no matter the position of the sensor. Our system can take advantages from autonomous mobile robots and protect the personal privacy. It hints at a generalization of flow recognition methods in the living spaces.