全球导航卫星系统应用
伪距
计算机科学
算法
人工智能
预处理器
符号
数学
全球定位系统
算术
电信
作者
Tao Li,Ling Pei,Yan Xiang,Yu Wang,Trieu‐Kien Truong
出处
期刊:IEEE robotics and automation letters
日期:2022-07-01
卷期号:7 (3): 7021-7027
被引量:11
标识
DOI:10.1109/lra.2022.3180441
摘要
Precise Point Positioning (PPP), a cutting edge GNSS technology, can achieve high-precision positioning without base station assistance. Visual-Inertial Odometry (VIO) realizes a more robust local pose estimation than Visual-SLAM. Based on PPP and VIO, we propose a tightly-coupled PPP/INS/Visual SLAM system, P $^{3}$ -VINS. It fuses GNSS raw measurements (pseudorange, carrier phase, and Doppler) with visual and inertial information for accurate and robust state estimation. All raw data is modelled and optimized under a factor graph framework. To eliminate ionospheric effects and utilize carrier phase measurements, P $^{3}$ -VINS uses the ionosphere-free (IF) model by dual-frequency observations and adds phase ambiguity into the estimated states. Finally, P $^{3}$ -VINS is evaluated on both public datasets and real-world experiments. It significantly outperforms benchmarks (GVINS and PPP) in terms of accuracy and smoothness. This result demonstrates that the high precision carrier phase substantially helps the GNSS/INS/Visual SLAM system reduce noise and improve accuracy.
科研通智能强力驱动
Strongly Powered by AbleSci AI