【摘要】 针对纯激光雷达里程计在几何特征稀疏场景下特征点不足和定位精度低的问题，设计了一种基于通用特征的紧耦合激光雷达里程计。首先，在数据预处理部分利用IMU(Inertial Measurement Unit)测量数据通过线性插值对雷达点云数据进行运动畸变剔除；其次，在特征提取部分，两个线程同时工作：一个线程提取环境中的线、面特征，另一个线程对点云数据进行鸟瞰图投影并提取ORB特征；然后，在状态估计部分，使用帧间线面特征和通用特征匹配建立观测方程，并通过误差卡尔曼滤波器将IMU数据与激光雷达数据融合，输出位姿。最后，为验证上述方法的有效性，使用KITTI数据集和仿真环境进行实验验证。实验结果表明，与现有方法相比，上述方法在精度上有一定程度的提高。更多还原

【Abstract】 Aiming at the problem of insufficient feature points and low positioning accuracy of pure LiDAR odometry in geometric sparse scenes, a compact coupled LiDAR odometry based on universal features was designed. Firstly, in the data preprocessing part, the Inertial Measurement Unit(IMU)measurement data were used to remove motion distortion of laser point cloud data through linear interpolation. Secondly, in the feature extraction part, two threads worked simultaneously: one thread extracted line and plane features in the environment, while the other thread projected the point cloud data onto a bird’s-eye view and extracts ORB features. Then, in the state estimation part, the observation equation was established by matching inter-frame line-plane features and generic features, and the IMU data were fused with laser data through error Kalman filter to output pose. Finally, in order to verify the effectiveness of the proposed method, some experiments were conducted using the KITTI dataset and simulation environment. The experimental results show that the proposed method has a certain degree of improvement in accuracy compared with existing methods.更多还原