基于移动感知计算的智能活动识别及其应用

Precise sensing on context information is the essential foundation for intelligent sensing applications, such as smart home, smart healthcare etc. Activity Recognition provides an effective way to sense the context information through inference on the intension according to the actions. With the prevalance of smart mobile devices, the mobile sensing based activity recognition becomes a key research problem. This project conducts the research on intelligent activity recognition and its applications based on mobile sensing methods..The project will focus on scientific problems such as continuous sensing with mobile device,uncertain information based plan recognition, activity pattern representation, inference and mining, intelligent distributed control meachanism aiming to intelligent assistance etc.Based on the scientific problems and key steps in activity recognition, the project will mainly focus on four research issues: mobile middle ware for effective and long-term continous sensing; markov logic network and case based planning based plan recognition method; stochastic context free grammaer based activity presentation, recognition and mining methods; Activity Recognition based intelligent applications.Through the research, the project will propose an efficient method for continous sensing; provide an accurate approach for activity recognition; build a system for intelligent assistence. Overall, the project will provide creative and proactive research achievements for intelligent sensing applications and internet of things.

对上下文信息的准确感知是智能家居、智能医疗等智能感知应用的必要基础。活动识别通过从动作到动作执行意图的推断,实现了对于上下文信息的有效感知。随着智能手机等具有感知功能的移动计算设备的不断普及，基于移动感知计算的活动识别成为一个重要的研究问题。本项目针对移动感知计算环境下的智能活动识别及其应用展开研究工作。其中，拟重点研究移动感知设备的连续感知、面向非确定性信息的规划识别、活动模式的表示、推理与挖掘以及针对智能协助的分布式智能控制等科学问题。根据拟研究的科学问题以及活动识别过程中的核心步骤，项目拟重点研究四方面内容：连续感知中间件技术；基于马尔科夫逻辑网的规划识别方法；基于随机上下文无关文法的活动模式表示、识别与挖掘方法;基于多智能体的分布式控制机制。通过研究，项目将实现连续的行为感知、准确的活动识别，构建基于活动识别的智能协助系统，为智能感知应用、物联网应用提供具有前瞻性的创新研成果。