A Filter-Based Integration of GNSS, INS, and Stereo Vision in Tight Mode With Optimal Smoothing

• DOI: 10.1109/jsen.2023.3295816
• Publication Date: 2023-07-20T18:01:55Z
• Authors: Xingxing Li; Chunxi Xia; Shengyu Li; Yuxuan Zhou; Zhiheng Shen; Zeyang Qin
• Keywords: Smoothing; Mode (computer interface)

In recent years, it is increasingly prevalent to integrate global navigation satellite system (GNSS) with visual-inertial systems (VINS) provide high-precision and continuous solutions. However, its accuracy robustness are challenged by GNSS signal blockages, inertial (INS) error accumulation, noisy visual measurements, especially in severe environments such as urban downtowns. To achieve accurate consistent pose estimation complex scenarios, we innovatively propose a filter-based GNSS/INS/Vision integrated system, which fuses pseudorange, carrier phase, IMU, measurements at the observation level. Meanwhile, full set of collected heterogeneous sensors utilized for optimal through RTSS-FBS smoother, can simultaneously enhance performance terms both relative absolute post-processing mode. The results vehicular experiments partly blocked challenging indicate that proposed decimeter-level centimeter-level position without obvious convergence processes. Combining advantages tight integration mode effective method could improve about 80% (root mean square reduction) compared state-of-the-art VINS-Fusion. noticeable improvements also be obtained attitude estimation.