环境感知与地下空间导航是煤矿智能化信息领域的重要研究方向,对实现无人化、全自动化、智能化的煤矿生产作业至关重要。 

随着第五代移动通信技术(5th generation mobile networks,5G)和毫米波成像雷达软硬件日益紧密结合与成熟,毫米波探测与通讯应用到更多领域。 

5G 通讯技术依托高速率、低延时、高带宽的特点给现有的无线电通讯技术带来巨大的变革；同时,毫米波雷达相比激光雷达,低成本、抗干扰、三维点云(3 dimension point cloud,3D)数量相对激光点云数量少1 ~ 2 个数量级的特点,使得其在地下环境 3D 成像及同步定位与地图构建( Simultaneous Localization and Mapping,SLAM)领域得到越来越多的关注。 基于 5G 通讯的 V2X(Vehicle to Everything)技术结合毫米波 SLAM 导航,为煤矿机器人的自主导航提供新的解决方案。 

系统综述了当下煤矿机器人自主导航以及实现煤矿智能化所面临的问题；近期国内外毫米波成像最新进展；地下环境毫米波雷达模块组通讯与信号获取方法；高分辨率成像遇到的稀疏特征提取问题；稀疏点云的处理策略与算法评估；深度学习在毫米波稀疏点云处理中的研究现状与发展方向；SLAM 算法应用于不同环境的研究现状及 SLAM 导航算法。 

最后归纳了煤矿地下环境中应用 SLAM 地图构建、路径规划及避障的困难和挑战,并对未来煤矿复杂环境下毫米波通讯与导航兼容并蓄的新应用提出了展望。

Environmental detection and underground space navigation is an important research direction in the field of intelligent information for coal mines,which is very important for the realization of unmanned,fully automatic and intel- ligent coal mine production. With the development of the Fifth Generation Mobile Networks (5G) and mmWave ima- ging technology,the integration of hardware and software design of millimeter wave in the detection and communication has been a considerable growth. The 5G communication technology relies on the characteristics of high speed,low de- lay and high bandwidth,which brings great changes to the existing radio communication technology. Compared with laser,mmWave radar has the characteristics of low cost,anti-jamming,and the point cloud pixels for each frame of image are 1-2 orders less than that of laser,which makes its more popular in the 3D imaging of underground environment and simultaneous localization and mapping (SLAM). The V2X (Vehicle to Everything) technology based on 5G com- munication combined with millimeter wave SLAM navigation provides a new solution for the autonomous navigation of coal mine robots. This paper systematically reviews the problems faced by the autonomous navigation of coal mine ro- bots and the realization of intelligent coal mine,and the research progress of mmWave imaging recently. The sparse feature extraction method for high-resolution imaging,and also the schematic diagram of communication and signal ac- quisition for multiple module groups are introduced. The processing strategy and algorithm evaluation of sparse point cloud and the research status and development direction of deep learning in sparse point cloud processing of mmWave imaging are summarized. Finally,the problems and challenges of SLAM map construction,path planning and obstacle avoidance in the underground mining environment,the research status of SLAM algorithm applied to different environ- ments and SLAM navigation algorithm are categorized and elaborated. In addition,the problems that need to be solved in the further study of millimeter wave communication and navigation and possible future development directions are proposed.