Affiliation:
1. College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China
Abstract
Simultaneous localization and mapping (SLAM) is considered a challenge in environments with many moving objects. This paper proposes a novel LiDAR inertial odometry framework, LiDAR inertial odometry-based on indexed point and delayed removal strategy (ID-LIO) for dynamic scenes, which builds on LiDAR inertial odometry via smoothing and mapping (LIO-SAM). To detect the point clouds on the moving objects, a dynamic point detection method is integrated, which is based on pseudo occupancy along a spatial dimension. Then, we present a dynamic point propagation and removal algorithm based on indexed points to remove more dynamic points on the local map along the temporal dimension and update the status of the point features in keyframes. In the LiDAR odometry module, a delay removal strategy is proposed for historical keyframes, and the sliding window-based optimization includes the LiDAR measurement with dynamic weights to reduce error from dynamic points in keyframes. We perform the experiments both on the public low-dynamic and high-dynamic datasets. The results show that the proposed method greatly increases localization accuracy in high-dynamic environments. Additionally, the absolute trajectory error (ATE) and average RMSE root mean square error (RMSE) of our ID-LIO can be improved by 67% and 85% in the UrbanLoco-CAMarketStreet dataset and UrbanNav-HK-Medium-Urban-1 dataset, respectively, when compared with LIO-SAM.
Subject
Electrical and Electronic Engineering,Biochemistry,Instrumentation,Atomic and Molecular Physics, and Optics,Analytical Chemistry
Reference46 articles.
1. An integrated GNSS/LiDAR-SLAM pose estimation framework for large-scale map building in partially GNSS-denied environments;He;IEEE Trans. Instrum. Meas.,2020
2. Monocular visual odometry based on depth and optical flow using deep learning;Ban;IEEE Trans. Instrum. Meas.,2020
3. Lu, Q., Pan, Y., Hu, L., and He, J. (2023). A Method for Reconstructing Background from RGB-D SLAM in Indoor Dynamic Environments. Sensors, 23.
4. Nonparametric Background Model-Based LiDAR SLAM in Highly Dynamic Urban Environments;Park;IEEE Trans. Intell. Transp. Syst.,2022
5. Method for registration of 3-D shapes;Besl;Proceedings of the Sensor Fusion IV: Control Paradigms and Data Dtructures,1992
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献