激光SLAM技术进展

Progress and application of LiDAR-based SLAM technology

针对近年来,基于激光雷达(LiDAR)的同时定位与地图构建(SLAM)技术在激光传感器技术的更新迭代、运动估计算法的稳定性和准确性、地图优化算法的精度和一致性以及与其他多种传感器深度融合等方面虽已取得显著进展,然而面对运动场景和复杂场景时,基于点云匹配的激光雷达里程计鲁棒性较差,容易失效,大规模场景下的后端优化算法会消耗大量计算资源,无法满足实时性的要求,以及不同场景下回环检测的成本较高且易误匹配等问题,研究分析激光SLAM技术进展:阐述激光雷达系统和激光SLAM框架;并详细介绍激光SLAM的关键技术和模块,针对激光SLAM与视觉SLAM系统开展对比研究;最后结合近年来的研究成果,指出激光SLAM技术在自动驾驶、机器人导航和工业自动化领域具有巨大的应用潜力,未来的研究趋势将集中于多源传感器融合、优化策略的改进以及与场景识别和环境感知技术的结合。

Aiming at the problems that in recent years,although significant progress has been made for laser radar(LiDAR)-based simultaneous localization and mapping(SLAM)in the advancements in laser sensor technology,the stability and accuracy of motion estimation algorithms,the accuracy and consistency of map optimization algorithms,and the deep integration with many other sensors,when dealing with moving and complex scenes,the LiDAR odometry based on point cloud matching exhibits poor robustness and is prone to failure,the backend optimization algorithms in large-scale scenarios consume a large amount of computing resources and cannot meet real-time requirements,and the loop back detection in different scenarios is easy to bring high cost and mismatch,and so on,the paper studied and analyzed the progress and application of LiDAR-based SLAM technology:the LiDAR system and the LiDAR-based SLAM framework were elaborated;and the key technologies and modules of LiDARbased SLAM were introduced in detail,then the laser SLAM and visual SLAM systems were comparatively analyzed;finally,combined with recent research advancements,it was pointed out that laser SLAM technology could have significant application potential in the domains of autonomous driving,robot navigation and industrial automation,furthermore,the future research trend would primarily focus on the fusion of multi-source sensors,the enhancement of optimization strategies,and the integration with scene recognition and environmental awareness technologies.