In the new era of robustness and perception, the Visual-Inertial Odometry (VIO), which is tightly coupled by the camera and the Inertial Measurement Unit (IMU), can obtain high-precision local pose results in unknown environment. Its low cost and miniaturization have received widespread attention. However, due to the limitation of the measurement principle, in the long-term runs, error will still accumulate. In addition, the outdoor large-scale environment is also a major challenge facing VIO. The Global Navigation Satellite System (GNSS) can provide accurate global estimates for VIO in an open outdoor environment and correct drift caused by long-term operation. Similarly, VIO can still perform in environments where GNSS is denied, which makes it possible for seamless indoor and outdoor navigation. Therefore, this paper proposes a visual-inertial SLAM algorithm assisted by GNSS. Taking the optimized tightly coupled VIO as the main body, and the pose information obtained by GNSS is combined with the VIO solution result to enhance the global positioning while ensuring the accuracy of the local pose accuracy. To this end, a simulation experiment based on the KITTI data set was carried out. The results show that the VIO system with the aid of GNSS can achieve the accuracy of 1.687 m error average, 1.176 m standard deviation, and 2.056 m root mean square error, which is nearly 80% higher than that without assistance. And it can also play a role in the environment where GNSS is denied, and the robustness of the system is also enhanced.
Research on Visual-Inertial SLAM Technology with GNSS Assistance
Lect. Notes Electrical Eng.
China Satellite Navigation Conference ; 2022 ; Beijing, China May 22, 2022 - May 25, 2022
China Satellite Navigation Conference (CSNC 2022) Proceedings ; Kapitel : 36 ; 425-434
2022-05-10
10 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Research on Visual-Inertial SLAM Technology with GNSS Assistance
TIBKAT | 2022
|EM-SLAM With Inertial/Visual Applications
IEEE | 2017
|Stereo Visual Inertial SLAM with Semi-direct Residual
IEEE | 2021
|