Visual-inertial odometry (VIO) is known to suffer from drifting and can only provide local coordinates. In this paper, we propose a tightly coupled GNSS-VIO system based on point-line features for robust and drift-free state estimation. Feature-based methods are not robust in complex areas such as weak or repeated textures. To deal with this problem, line features with more environmental structure information can be extracted. In addition, to eliminate the accumulated drift of VIO, we tightly fused the GNSS measurement with visual and inertial information. The GNSS pseudorange measurements are real-time and unambiguous but experience large errors. The GNSS carrier phase measurements can achieve centimeter-level positioning accuracy, but the solution to the whole-cycle ambiguity is complex and time-consuming, which degrades the real-time performance of a state estimator. To combine the advantages of the two measurements, we use the carrier phase smoothed pseudorange instead of pseudorange to perform state estimation. Furthermore, the existence of the GNSS receiver and IMU also makes the extrinsic parameter calibration crucial. Our proposed system can calibrate the extrinsic translation parameter between the GNSS receiver and IMU in real-time. Finally, we show that the states represented in the ECEF frame are fully observable, and the tightly coupled GNSS-VIO state estimator is consistent. We conducted experiments on public datasets. The experimental results demonstrate that the positioning precision of our system is improved and the system is robust and real-time.

We propose a synergistic lidar-visual-inertial state estimation system, LIVIO, which achieves high-accuracy and robust localization results. LIVIO is composed of two sub-systems: a lidar-inertial system (LIS) based on Iterated Error-State Kalman Filter and a visual-inertial system (VIS) based on nonlinear optimization. In the prediction stage of LIS, a midpoint integration approach is used to obtain the discrete states and covariance propagation, significantly enhancing pose estimation accuracy. For the update stage, LIS employs normal distribution transform registration coupled with sub-maps for efficient map matching, resulting in lower time consumption. On the other hand, to improve the localization accuracy, VIS strategically integrates LIS odometry as a prior constraint in its factor graph. Another key innovation of ours lies in the novel feature velocity factor, establishing a geometric constraint between the 3D velocity in the camera frame and its corresponding projected 2D velocity on the normalized image plane. This formulation effectively bounds localization drift through physical motion consistency. Finally, we validate our proposed LIVIO on both public and collected datasets. Our method achieves relative pose errors below 0.253 m and time consumption under 60 ms across all datasets, outperforming state-of-the-art localization systems in nearly all evaluated scenarios. The experimental results demonstrate that LIVIO is accurate, robust, real-time, and highly applicable.