Based one open source SLAM framework VINS-Mono.
The approach contains
- Depth-integrated visual-inertial initialization process.
- Visual-inertial odometry by utilizing depth information while avoiding the limitation is working for 3D pose estimation.
- Noise elimination map which is suitable for path planning and navigation.
However, the proposed approach can also be applied to other application like handheld and wheeled robot.
This dataset is part of the dataset collection of the STAR Center, ShanghaiTech University: https://star-datasets.github.io/
A video showing the data is available here: https://robotics.shanghaitech.edu.cn/datasets/VINS-RGBD
Shan, Zeyong, Ruijian Li, and Sören Schwertfeger. "RGBD-inertial trajectory estimation and mapping for ground robots." Sensors 19.10 (2019): 2251.
@article{shan2019rgbd,
title={RGBD-inertial trajectory estimation and mapping for ground robots},
author={Shan, Zeyong and Li, Ruijian and Schwertfeger, S{\"o}ren},
journal={Sensors},
volume={19},
number={10},
pages={2251},
year={2019},
publisher={Multidisciplinary Digital Publishing Institute}
}
1.1. Ubuntu 16.04 or 18.04.
1.2. ROS version Kinetic or Melodic fully installation
1.3. Ceres Solver Follow Ceres Installation
1.4. Sophus
git clone http://github.com/strasdat/Sophus.git
git checkout a621ff
Recording by RealSense D435i. Contain 9 bags in three different applicaions:
Note the rosbags are in compressed format. Use "rosbag decompress" to decompress.
Topics:
- depth topic: /camera/aligned_depth_to_color/image_raw
- color topic: /camera/color/image_raw
- imu topic: /camera/imu
The source code is released under GPLv3 license.