UT-MM Dataset
To access our dataset, visit Hugging Face. To see videos of the content, please visit our [Project Page].
A script to convert the UT-MM rosbag data to usable .txt files is included in
scripts/bag2data.py at this repository. Note that ROS
is required to read the camera-to-imu transformation.
python scripts/bag2data.py --path <rosbag dir path> --scene <scene name>
However, the transformation is constant between our datasets, which we have provided.
./tf/tf.txt
Command Line Arguments for scripts/bag2data.py
--path
Path to the directory containing rosbags
--scene
Name of the scene to read
Rosbag topics
Combined, the datasets contain 8387 images, 2796 LiDAR scans, and 27971 IMU measurements. The following is a quick overview of the topics that can be found within each bag file:
/vrpn_client_node/Jackal_Latest/pose : geometry_msgs/PoseStamped (Pose Coordinates, 100hz)
/realsense/color/image_raw/compressed : sensor_msgs/CompressedImage (RGB Image, 30hz)
/realsense/depth/image_rect_raw : sensor_msgs/Image (Depth Image, 30hz)
/microstrain/imu/data : sensor_msgs/Imu (Imu Measurements, 100hz)
/ouster/points : sensor_msgs/PointCloud2 (LiDAR Point Clouds, 10hz)
To visualize the dataset, ROS is needed. Some scripts are provided in the UT_MM_Scripts directory.
roscore
rosrun rviz rviz -d UT_MM_Scripts/configs/jackal.rviz
rqt --clear-config --perspective-file UT_MM_Scripts/configs/rqt_plots.perspective
rosbag play --clock *.bag --pause
python3 imu_preintegration.py
Citation
If you find our paper useful or interesting, please consider giving a star ⭐ and citing the following paper 📝.
@misc{sun2024mm3dgs,
title={MM3DGS SLAM: Multi-modal 3D Gaussian Splatting for SLAM Using Vision, Depth, and Inertial Measurements},
author={Lisong C. Sun and Neel P. Bhatt and Jonathan C. Liu and Zhiwen Fan and Zhangyang Wang and Todd E. Humphreys and Ufuk Topcu},
year={2024},
eprint={2404.00923},
archivePrefix={arXiv},
primaryClass={cs.CV}
}