一种融合自适应点云特征提取的激光SLAM方法

Laser SLAM method with adaptive feature extraction fusion

针对传统SLAM系统前端里程计基于固定阈值提取特征点易导致位姿估计精度降低的问题,该文提出一种自适应点云特征提取方法,通过引入非线性衰减系数,根据点云到激光雷达中心的距离自适应调整特征提取阈值,同时附加非线性权重,提高系统位姿估计的鲁棒性。该文基于FAST-LIO2算法,利用M2DGR数据集验证了所提方法的可靠性,实验结果表明,本文方法的定位和建图精度优于原算法和同类型算法,在提高精度的同时保证了算法的运行效率,有效提升了算法的适用性。

Aiming at the problem that the front-end odometer of the traditional SLAM(Simultaneous Localization and Mapping)system extracts feature points based on a fixed threshold,which easily leads to a decrease in the accuracy of pose estimation,this paper proposes an adaptive point cloud feature extraction method.By introducing a nonlinear attenuation coefficient,the feature extraction threshold is adaptively adjusted according to the distance from the point cloud to the center of the laser radar,and a nonlinear weight is added to improve the robustness of the system pose estimation.Based on the FAST-LIO2 algorithm,this paper verifies the reliability of the proposed method using the M2DGR dataset.The experimental results show that the positioning and mapping accuracy of the proposed method is better than the original algorithm and the same type of algorithm.While improving the accuracy,it ensures the operation efficiency of the algorithm and effectively improves the applicability of the algorithm.