Abstract: Accurate and efficient Simultaneous Localization and Mapping (SLAM) algorithms are crucial in mobile robotics and autonomous driving technologies. Existing SLAM methods often face challenges in precision and robustness when dealing with complex environments and dynamic scenes. This paper proposes a SLAM algorithm based on point-line features and multi-IMU (Inertial Measurement Unit) fusion to address these challenges. This study utilizes multi-sensor fusion technology, including LiDAR and cameras, to extract point and line features from the environment. These geometric features provide rich environmental information, aiding in the construction of more detailed and accurate maps. During the point-line feature extraction process, an optimization-based feature matching algorithm is employed to ensure the accuracy and stability of feature extraction. Multi-IMU fusion technology enhances the system's motion estimation capabilities. Multi-IMU fusion not only improves the robustness of a single IMU in high-dynamic environments but also, through optimized sensor data fusion algorithms, provides more precise pose estimation. The experimental section verifies the proposed SLAM algorithm in various typical indoor and outdoor environments, including static and dynamic scenes. Results indicate that the proposed SLAM algorithm significantly improves positioning accuracy, mapping quality, and real-time performance. Compared to traditional methods, this algorithm performs superiorly in complex environments, effectively handling environmental changes and noise disturbances.

Key words: slam, point-line feature, multi-IMU fusion, autonomous navigation, Graph Optimization, sensor fusion, Light Detection And Ranging （LiDAR）, vision sensor

摘要： 在移动机器人和无人驾驶技术中，准确且高效的同时定位与地图构建（SLAM）算法至关重要。现有的SLAM方法在处理复杂环境和动态场景时，往往面临着精度和鲁棒性不足的问题。本论文提出了一种基于点线特征与多IMU（惯性测量单元）融合的SLAM算法，以解决这些挑战。本研究利用激光雷达和摄像头等多传感器融合技术，从环境中提取点和线特征。这些几何特征提供了丰富的环境信息，有助于构建更加详细和准确的地图。在点线特征提取过程中，采用了基于优化的特征匹配算法，确保特征提取的准确性和稳定性。通过多IMU融合技术，增强了系统的运动估计能力。多IMU融合不仅提高了单一IMU在高动态环境下的鲁棒性，还通过优化的传感器数据融合算法，提供了更为精确的位姿估计。实验部分在多种典型的室内和室外环境中进行了验证，包括静态和动态场景。结果表明，本文提出的SLAM算法在定位精度、建图质量以及实时性方面均有显著提升。与传统方法相比，该算法在复杂环境中的表现更为优越，能够有效应对环境中的变化和噪声干扰。

关键词: slam, 点线特征, 多IMU融合, 自主导航, 图优化, 传感器融合, 激光雷达, 视觉传感器