基于低成本商用设备无线信号的目标识别、成像与定位研究

Over the past decades, the target "sensing" technology based on wireless signals is a promising candidate in extensive scenarios such as gesture recognition, security detection, medical imaging, etc. The key advantages of wireless signals over traditional optical imaging and recognition approaches are strong penetrability, wide coverage, non-sensitivity to lighting and simple data-processing, etc. However, the pioneer wireless signal based “sensing” schemes encounter the bottlenecks of high equipment overhead and poor universality since they rely on expensive dedicated devices. Therefore, this project explores an effective scheme for target material identification, shape imaging and localization building on WiFi or other cheap commercial wireless devices. By studying the correlations among target material, shape and transmitted signal fluctuation characteristics, we can achieve the goal of target material identification and shape imaging; by combining the radar “spatial spectrum estimation” and “spatio-temporal joint estimation of angle and time delay”, we can address the challenge of multipath. The project will establish the prototype system for typical application scenarios, test and optimize the performance of proposed scheme. The scientific essence lies in exploring the inner relationship among the target material, shape, location and physical properties of wireless signals such as amplitude and phase. The research results will provide valuable theoretical foundations and technical reference for wireless sensing applications, such as robot manipulation and interaction, and will be of great significance for promoting the development of IoT (Internet of Things) applications.

近年来，基于无线信号的目标“感知”技术在诸如手势识别，安防安检，医疗影像等领域展现了巨大的应用潜力。与传统光学成像和识别相比，无线信号具有穿透性好、覆盖面广、不受光线条件影响和数据处理简单等的优势。但是现有基于无线信号的“感知”方法由于依赖于价格不菲的专用设备，因此面临价格昂贵难以普适的困境。对此，本项目探索利用WiFi等低代价商用无线设备实现目标材质识别、形状成像和定位的有效方法。通过研究目标材质、形态、位置与信号波动特征间关系，实现目标材质识别、成像与定位；将雷达“空间谱估计”和“基于角度和时延的时空联合估计”等理论相结合解决多径挑战。项目将针对典型应用场景建立原型系统，测试并优化所提方案。其科学实质是探索感知目标材质、形态和位置与无线信号振幅、相位等物理属性间内在关系，成果将为机器人操控交互等无线感知和物联网应用提供有价值的理论依据与技术参考，对推动物联网和无线感知应用有重要意义。

本项目针对现有基于无线信号的感知方法依赖于价格昂贵的专用设备导致难以普适的问题，研究了在WiFi等低代价商用无线设备下目标材质识别、成像以及室内定位等问题。针对目标材质感知，本项目在商用Wi-Fi、声音等商用设备上，通过分析无线电波受到环境和目标对于信号振幅、相位和频率，对不同的液体和固体进行材质识别。针对目标成像感知，本项目探索了在商用毫米波雷达设备上结合关联成像技术对静态目标进行成像。针对定位与感知探索，本项目利用LoRa和Wi-Fi等商用廉价设备进行非接触、大范围的目标定位以及跨目标感知。针对感知数据距离、吞吐受限问题，本项目利用超材料，通过设计可调超平面RFlens控制环境中的无线信号的波束，使波束汇聚或者转向到期望方向上，解决了物联网感知设备的感知距离以及感知盲区等问题。其次，利用backscatter技术提高感知通信距离和数据吞吐等问题。该项目的多个成果先后发表在Mobicom, Sigcomm, Ubicomp, TON, TMC等高水平国际会议和期刊上。本项目的研究成果将为机器人操控交互等无线感知和物联网应用提供有价值的理论依据与技术参考，对推动物联网和无线感知应用有重要意义。

内容获取失败，请点击重试