随着机器人软硬件水平的逐步成熟，智能机器人产业迅速崛起。工业机器人、医疗机器人、农业移动机器人、导游机器人等逐渐普及，为人们生活带来了极大的便利。在这些场景的实际应用时，自主定位导航是移动机器人的核心功能。针对单一传感器定位导航方案的局限性，本文采用激光雷达、IMU（惯性测量单元）、里程计、RGBD相机等多传感器数据融合的方式，针对室内轮式移动机器人自主定位导航需要解决的定位、建图与路径规划三个问题展开研究，并设计实现一套集成主流的同步定位与建图算法、定位算法、路径规划算法的机器人定位导航实验系统。具体工作如下：（1）采用自主定位与建图（SLAM）方法构建环境地图，并基于多传感器融合克服激光SLAM定位的不足。使用Extended Kalman Filter算法融合IMU、里程计数据优化相对位姿，引入激光和地图数据估计全局位姿，之后再通过点云匹配进一步提升位姿精度。（2）对机器人导航的全局和局部路径规划展开研究，实现了Dijkstra、A*、Hybrid A*三种全局路径规划算法，设计了基于激光和深度相机信息融合的三层局部代价地图，并基于此实现了Trajectory Rollout和Time Elastic Band两种局部动态避障方法。（3）基于机器人操作系统ROS进行实验系统的开发，实现了支持局域网内远程操控的实验平台，支持建图、定位、导航、通信、可视化等功能。在实验过程中，首先对系统的可视化、定位、建图、导航以及ROS话题通信功能进行测试，验证了系统的有效性；然后基于系统对不同定位方式和多种路径规划算法进行测试，验证了多传感器融合定位导航方案的可行性和有效性；最后通过定位精度对多传感器融合定位导航系统的性能进行评估与比较。实验表明，本文采用的多种传感器数据融合方案，优于单一的里程计定位与纯激光雷达方案，有效改进传统移动机器人性能。此外，设计的定位导航系统成本较低、运行可靠、里程精度较高、鲁棒性好，具有广阔的应用前景。

With the gradual maturity of robot software and hardware levels, the intelligent robot industry is rapidly emerging. Industrial robots, medical robots, agricultural mobile robots and tour guide robots are gradually becoming popular, bringing great convenience to people‘s lives. In the practical application of these scenarios, autonomous localization and navigation is the core function of mobile robots.To address the limitations of single-sensor positioning and navigation solutions, this paper adopts a multi-sensor data fusion approach, such as LiDAR, IMU (inertial measurement unit), odometer, and RGBD camera, to study the three problems of positioning, map building, and path planning that need to be solved for autonomous positioning and navigation of indoor wheeled mobile robots, and designs and implements a set of robot positioning and navigation algorithms integrating mainstream simultaneous positioning and map building algorithms, positioning algorithms, and path planning algorithms, The experimental system of robot positioning and navigation integrates the mainstream synchronous positioning and mapping algorithms, positioning algorithms, and path planning algorithms. The specific work is as follows: (1) We adopt the autonomous localization and mapping (SLAM) method to construct the environment map, and overcome the shortage of laser SLAM localization based on multi-sensor fusion. The Extended Kalman Filter algorithm is used to fuse IMU and odometer data to optimize the relative positional position, and introduce laser and map data to estimate the global positional position, after which the positional accuracy is further improved by point cloud matching. (2) The global and local path planning for robot navigation is investigated, and three global path planning algorithms, Dijkstra, A*, and Hybrid A*, are implemented, and a three-layer local cost map based on laser and depth camera information fusion is designed. (3) Based on the robot operating system ROS (3) The experimental system is developed based on the robot operating system ROS, and an experimental platform supporting remote control in the local area network is realized, which supports map building, localization, navigation, communication, visualization and other functions.In the experimental process, the visualization, localization, map building, navigation and communication functions of the system and ROS topic were tested to verify the effectiveness of the system; then different localization methods and multiple path planning algorithms were tested based on the system to verify the feasibility and effectiveness of the multi-sensor fusion localization and navigation scheme; finally, the performance of the multi-sensor fusion localization and navigation system was evaluated by localization accuracy Finally, the performance of the multi-sensor fusion positioning navigation system is evaluated and compared by positioning accuracy. The experiments show that the multi-sensor data fusion scheme is better than the single odometer positioning and pure LIDAR scheme, and effectively improves the performance of the traditional mobile robot. In addition, the designed positioning and navigation system has a broad application prospect because of its low cost, reliable operation, high odometry accuracy, and good robustness.