Home > Published Issues > 2021 > Volume 10, No. 4, April 2021 >

A Tightly-Coupled Visual-Inertial System with Synchronized Time for Indoor Localization

Nguyen H. K. Tran and Vinh-Hao Nguyen
Ho Chi Minh City University of Technology, VNU-HCM, Ho Chi Minh City, Vietnam

Abstract—Indoor localization remains a difficlut problem due to the lack of global positioning facilities. In this paper, we present an indoor localization system based on tightly coupling the stereo camera and the inertial measurement unit (IMU). To achieve the correct fusion, we strictly synchronized the sensors by compensating for all of system delays, including the varied exposure time of camera. The calibration datasets were carefully recorded to obtain the exact model of the stereo camera as well as the relationship between the camera and the IMU. After preparing, we performed the localization by utilizing the keyframe-based visual-inertial odometry. This algorithm works by minimizing the IMU error jointly with the camera reprojection, to get an optimal estimation of robot states. The visual landmarks are calculated by keypoint matching and triangulation between current frames and keyframes. The keyframes are selected to view keypoints from different angles to improve the depth uncertainty of landmarks. The algorithm was put into practice by an embedded computer which has enough processing ability, while presented in a small size for convenient deployment on mobile robots. The experimental results indicated that the developed system could achieve sufficient accuracy and robustness under real-world conditions. 

Index Terms—visual-inertial odometry, stereo camera, IMU, time synchronization, indoor localization, mobile robot

Cite: Nguyen H. K. Tran and Vinh-Hao Nguyen, "A Tightly-Coupled Visual-Inertial System with Synchronized Time for Indoor Localization," International Journal of Mechanical Engineering and Robotics Research, Vol. 10, No. 4, pp. 202-208, April 2021. DOI: 10.18178/ijmerr.10.4.202-208

​Copyright © 2021 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.