Parallax VI-SLAM

This is the MATLAB code for our paper "Parallax Visual-Inertial SLAM: Parallax Bundle Adjustment with IMU and Linear Submap Joining".

This paper first proposes a new method for Visual-Inertial SLAM. It uses a parallax angle for feature parametrization [1], and the feature observation and the preintegrated IMU information are used together to formulate a nonlinear least squares problem.

In addition, it proposes a linear submap joining method using the Linear SLAM framework [2], where the local submaps are built using the Parallax Visual-Inertial SLAM. Then, these submaps are joined together through linear least squares and nonlinear coordinate transformations.

Our proposed Parallax Visual-Inertial SLAM and linear submap joining algorithms are evaluated using multiple KITTI [3] datasets, demonstrating nice convergence, robustness and high accuracy.

---

To run Parallax VI-SLAM, 'PBAwIMU_LocalMap_main.m ' can be executed to build a full-batch or local map. Dataset ('KITTI-06', 'KITTI-07', 'KITTI-09') and Start and End states can be decided.

Standard VI-SLAM, using XYZ parametrization, can be found in 'XYZIMU' folder. Same procedure as PBAwIMU, in 'SBAwIMU_LocalMap_main_LM.m', different datasets can be decided. In Standard VI-SLAM, the 3D feature position is initialized utilizing the parallax angle, which yields the same initial objective function as PBAwIMU.

Linear Map Joining can be executed with 'Main.m' in 'LinearSLAM' folder.
Datasets ('KITTI-06', 'KITTI-07', 'KITTI-09') with different numbers of local maps can be chosen.

It can be plotted using rpg_trajectory_evaluation (https://github.com/uzh-rpg/rpg_trajectory_evaluation) with Ground-Truth (stamped_groundtruth.txt) [4].

---

References

1. L. Zhao, S. Huang, Y. Sun, L. Yan, and G. Dissanayake, “ParallaxBA:Bundle adjustment using parallax angle feature parametrization,” International Journal of Robotics Research, vol. 34, no. 4-5, pp. 493–516, 4 2015
2. L. Zhao, S. Huang, and G. Dissanayake, “Linear SLAM: Linearising the SLAM Problems using Submap Joining,” Automatica, vol. 100, pp.231–246, 9 2018. [Online]. Available: http://arxiv.org/abs/1809.06967
3. A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Vision meets robotics: The kitti dataset,” The International Journal of Robotics Research, vol. 32, no. 11, pp. 1231–1237, 2013.
4. Z. Zhang and D. Scaramuzza, “A Tutorial on Quantitative Trajectory Evaluation for Visual(-Inertial) Odometry,” in IEEE International Conference on Intelligent Robots and Systems. Institute of Electrical and Electronics Engineers Inc., 12 2018, pp. 7244–7251.

---