面向工业互联网的无线非绑定感知理论与关键技术

Industrial Internet is bringing a profound technology revolution to global industry. Industrial Internet aims to connect human, devices and products by smartly identifying, interacting, and controlling them, which is the key to lead the future industry. Sensing technology, as a key to Industrial Internet, underpins smart identification, interaction, and control among human, machines and products. The state-of-the-art research mostly employs device-based sensing, which needs to attach specialized devices to the sensing objects. In industrial environments, however, production people usually do not or are not allowed to bring external devices with them. Thus, device-free contactless sensing, i.e., sensing objects without attaching any devices with them, becomes attractive and popular in such scenarios. Facing the challenges of device-free sensing in industrial environments, we study the theory and key techniques of device-free contactless sensing in this project. We intend to study the modeling the denoising of physical layer channel state information, and realize LOS identification for bandwidth-limited channels, and model the relationship between channel state and environment changes. We conduct extensive research on human sensing, propose scenario independent sensing of both moving and stationary human, and form a unified sensing framework, enabling human sensing in industrial environments.

工业互联网正在引发全球工业界的一场深刻技术变革，是世界各国取得未来工业领先优势的关键。工业互联网旨在实现人、机器、产品等工业实体之间的精确识别、有效交互与智能控制，实现智能生产。而感知技术作为工业实体之间高度协作、共享智能的基础，成为工业互联网的核心。现有的感知技术大多数依赖感知对象携带特定设备，通过感知绑定在对象上的设备实现感知，即绑定感知。但在工业生产环境中，生产人员往往没有或不允许携带特殊设备，因此迫切需要在感知对象未绑定任何设备时即可实现感知，即非绑定感知。本课题围绕面向工业互联网的无线非绑定感知理论与关键技术，拟研究物理层信道状态信息的建模与降噪，实现带宽受限信道上的视距非视距实时辨识技术，进而推衍信道状态与环境变化的关系模型，奠定无线非绑定感知理论基础。课题组拟针对人员感知进行深入研究，提出场景无关的人员非绑定感知技术，并形成统一的感知框架，为工业互联网智能感知提供技术支撑。

工业互联网的一个关键特征是联网设备具备广泛的感知功能，能够感知周围环境、人员等物理环境信息。本项目工作通过挖掘普适无线基础设施（如WiFi）的物理层信道状态信息，探索不依赖特定设备的非绑定感知理论和关键技术，在普通商用WiFi设备上，开发高精度的被动式人员感知与定位追踪理论和技术。具体地，本课题实现了以下内容：面向非绑定无线感知的信道状态信息干扰检测与消除算法；基于WiFi的单链路分米级被动式人员追踪方法； 基于无线信号的人及交互方法和零成本跨场景手势识别。课题已完成全部研究目标，并完成了视觉信号与无线信号结合的室内定位技术研究，在国际权威学术刊物上发表15以上高水平研究论文，并公开了当前领域内规模最大的无线收拾识别数据集。