基于因子图优化的激光惯性SLAM方法研究

Research on method of laser inertial SLAM based on factor graph optimization

融合激光雷达和惯性测量单元的SLAM方法是拒止环境下自动驾驶定位建图的重要技术手段。该技术包含前端和后端2个数据处理模块,在后端数据处理方面,现有方法存在长时间运行时累积误差较高、回环检测计算负载较大以及复杂拒止环境下鲁棒性不理想等问题。针对上述需求,提出一种适配前端激光惯性里程计的新型后端数据处理方法。该方法采用因子图优化算法架构,建立激光连续关键帧间的惯性单元预积分模型,将该模型作为因子图架构中表征惯性单元数据的算法因子,降低数据处理的计算负载。构建基于Scan-Context描述符的高效回环检测方法,将点云数据三维空间结构特征转化为二维特征图,在保证回环检测精度的前提下进一步提高计算效率。结合前端里程计信息,构建包含里程计因子、惯性单元预积分因子和回环检测因子误差项的目标函数,通过非线性优化算法求解最优位姿状态,形成完整的SLAM算法结构。对所述方法及FAST-LIO2、LIO-SAM和SC-LeGO-LOAM等现有主流激光惯性SLAM方法基于开源数据集进行对比验证,并开展实车试验。结果表明:相较于现有方法,所述DSC-Algo方法在公开数据集测试中的计算性能和全局定位精度实现了显著提升,在现实拒止环境实车测试中的定位精度和算法鲁棒性也具有明显优势。

SLAM with the integration of LiDAR and Inertial Measurement Unit is an important means of localization and mapping for autonomous driving in denial environments.It comprises front-and back-end two data processing modules.In current back-end data processing,problems including high accumulative error after long-time running,high calculated load and unsatisfactory robustness in complicated denial environments still exist.To address these issues,this paper proposes a new back-end data processing algorithm adapting to laser inertial odometer based front-end algorithm.Factor graph optimization is applied to build the framework.First,an IMUpre-integration model between laser consecutive key frames is built and employed as the optimization factor for describing the IMUdata in factor graph framework,helping reduce the calculated load of data processing.Second,an efficient loop-closing detection method based on the Scan-Context descriptor is introduced,where three-dimensional spatial structure features of the point cloud are converted into a two-dimensional feature map.The calculation efficiency,therefore,is improved in the premise of keeping an eligible loop-closing detection precision.Finally,with the integration of the front-end odometer information,the whole SLAM model is built using odometer factor,IMUpre-integration factor and loop-closing detection factor,and the optimized position-orientation is solved by nonlinear optimization algorithm.In open-source-dataset and real-vehicle-based tests,the proposed DSC-Algo method exhibits markedly higher global localization precision,better calculating performance and robustness in complicated denial environments compared with other mainstream methods including FAST-LIO2,LIO-SAM and SC-LeGO-LOAM.