多四旋翼无人机编队LiDAR系统的城市地形三维重建关键技术研究

The fast and precise 3D reconstruction of urban landscape has great application potentials in remote sensing and mapping, earthquake disaster rescue, military reconnaissance and urban planning. Due to measurement accuracy and range properties, the aerial Light Detection And Ranging (LiDAR) system is generally believed as the most effective technology in 3D urban landscape reconstruction. However, most LiDAR systems are built upon expensive equipment, which further complicates the data acquisition process and thus limits the application of LiDAR systems. In this proposal, we present a 3D urban landscape reconstruction method based on multiple 4-rotor UAVs mounting aerial LiDAR systems. The UAVs form a highly efficient formation for laser scanning of urban landscape to acquire LiDAR point cloud data. A streaming framework is then incorporated for seamless landscape reconstruction. The proposal focuses on three major research topics: the research on three basic flying motions for 4-rotor UAV control, incluing attitude control, hover flying and path following, and the investigation on synthesizing the three basic motions to obtain an organization approach of a complex flying motion; the research on a decentralized cooperation method utilizing an optimal formation parameter set, for achieving multiple UAVs autonomous formation flight; the research on the classification and segmentation methods of the LiDAR point cloud, and the incorporation of a data streaming framework for fast and precise seamless reconstruction of urban landscape. The proposal synthesizes the most advanced researches in UAV control, formation flight and sensor fusion areas. The research results will provide new approaches and solutions for remote sensing and mapping, earthquake disaster rescue, military reconnaissance and urban planning.

城市地形的快速精确三维重建，在遥感测绘、抗震救灾、军事侦察和城市规划中有巨大应用价值。空载轻质探测测距（LiDAR)系统，因测量精度高、范围广、速度快，是最有效的城市地形三维重建技术方法。然而一般LiDAR 系统设备成本昂贵、数据获取难度大，限制了其应用范围。本课题提出采用多个装载小型LiDAR 系统的四旋翼无人机，以高效巡航编队扫描城市地形，同步获取LiDAR 数据进行无缝拼接以快速精确重建三维城市地形：探索实现四旋翼无人机姿态控制、悬停操控和路径跟踪基本飞控动作，组合实现无人机复杂飞控动作；采用分布式协同控制方法，确定优化协动位姿参数，实现多无人机自主巡航编队；研究LiDAR 数据流的分类分割和无缝拼接方法，快速精确重建三维城市地形。本课题融合了无人机控制、编队飞行、信息融合等前沿研究领域，研究成果将为遥感测绘、抗震救灾和城市规划提供一种新的思路和方法。

面向遥感测绘、抗震救灾、军事侦察和城市规划应用需求，开展基于多四旋翼无人机搭载小型 LiDAR 系统的城市地形三维重建实现方法项目研究。迄今，通过研究复杂环境下四旋翼无人机基本飞控动作实现方法，实现了四旋翼无人机复杂飞行控制；通过研究多无人单元自主编队控制，实现多无人单元系统的自主高效编队；完成了LiDAR 数据获取、拼接和配准、数据滤波，开发快速精确的室外复杂环境三维重建方法；设计并搭建了无人机LiDAR系统实验平台，实现基于 LiDAR 点云数据的城市地形三维重建。.项目主要研究成果如下：（1）提交课题研究报告一份；（2）搭建基于LiDAR系统的无人机系统两套；（3）在国内外期刊、会议发表高水平论文 5 篇； （4）申请发明专利 7 项；（5）培养中国科学技术大学博士研究生1名，硕士研究生3名。.项目研究成果已应用在多传感融合无人车自主定位、光伏商用系统三维环境重建等，给企业带来一定的经济效益；同时，课题组对项目研究成果在市场化应用方面进行了相关调研，预计项目成果在激光检测、安防巡检、城市规划领域有规模化应用前景。