Abstract: Human activity recognition(HAR) technology based on intelligent perception has shown great application potential,especially in the fields of health monitoring,intelligent sports and rehabilitation training.In order to analyze the current level and future development direction of human activity recognition technology,firstly,the intelligent sensing technology based on wearable sensor is described.Secondly,the open data sets of different modes are summarized and compared.Thirdly,the research status of existing intelligent algorithms is summarized,and the application effects of machine learning algorithms,deep learning algorithms and multi-modal algorithms in human activity recognition are analyzed.Finally,the future research direction and main challenges of new sensing technology,integrated sensor-memory and computing architecture,and multi-modal approach are discussed,such as data diversity,algorithm generalization ability and privacy protection.