基于传感器数据的用户行为识别方法综述

Overview of Human Behavior Recognition Methods Based on Sensor Data

中文关键词: 行为识别  行为认知  传感数据理解  服务机器人  智能家居

英文关键词: behavior recognition  behavior cognition  insight of sensor data  service robot  smart home

中文摘要:

      目的 综述了运用传感器数据感知用户行为的最新进展和研究成果，为多模态数据驱动的服务机器人用户行为认知应用提供技术支撑。方法 主要从4个方面总结了用户行为识别方法的最新进展，具体包括:基于可穿戴传感器的数据感知与行为识别;基于机器视觉的行为识别;基于环境传感器数据的行为识别;异常行为识别与预测。结论 服务机器人在行为感知到认知方面的研究仍处于初步阶段，运用多模态数据全面建模用户行为有望提升基于机器人的养老产品智能化水平。

英文摘要:

      This paper reviews the latest developments and research achievements of human behavior perception using sensor data, which provides technical support for human behavior recognition application of multi-modal data-driven service robot. The paper summarizes the latest developments of user behavior recognition methods from four aspects. The data perception and behavior recognition technologies based on wearable sensors are introduced, the behavior recognition method based on computer vision or environmental sensor data is summarized, and the identification and prediction method of abnormal behavior is elaborated. Conclusion could be drawn that the research on behavioral perception up to recognition of service robots is still in the preliminary stage. It is also pointed out that the intelligence level of robot-based elderly care products is expected to be improved by comprehensively modeling human behavior using multi-modal data.