Coco-LIC

Coco-LIC: Continuous-Time Tightly-Coupled LiDAR-Inertial-Camera Odometry using Non-Uniform B-spline

The following are three main characters of 🥥 Coco-LIC [Paper] [Video] :

• dynamically place control points to unlock the real power of the continuous-time trajectory
• tightly fuse LiDAR-Inertial-Camera data in a short sliding window based on a factor graph
• support multimodal multiple LiDARs and achieve great performance in degenerated cases

Prerequisites

• ROS（tested with noetic）
• Eigen 3.3.7
• Ceres 2.0.0
• OpenCV 4
• PCL >= 1.13
• livox_ros_driver
• yaml-cpp

Install

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws/src
git clone https://github.com/Livox-SDK/livox_ros_driver.git
cd ~/catkin_ws && catkin_make
cd ~/catkin_ws/src
git clone https://github.com/APRIL-ZJU/Coco-LIC.git
cd ~/catkin_ws && catkin_make
source ~/catkin_ws/devel/setup.bash
cd ~/catkin_ws/src/Coco-LIC && mkdir data

Run

• Download R3LIVE dataset or FAST-LIVO dataset or NTU-VIRAL dataset or LVI-SAM dataset.

• Configure parameters in the config/ct_odometry_xxx.yaml file.

  • log_path: the path to log
  • config_path: the path of config folder
  • bag_path: the file path of rosbag

• Run on R3LIVE dataset for example.

  roslaunch cocolic odometry.launch config_path:=config/ct_odometry_r3live.yaml

  The estimated trajectory is saved in the folder ./src/Coco-LIC/data.

Supplementary1 - non-uniform verification

1 control point per 0.1 seconds 🥊 adaptively placing control points per 0.1 seconds.

The different colors of the trajectory correspond to different densities of control points.

Supplementary2 - comparison on NTU-VIRAL

We additionally compare Coco-LIC with our previous work CLIC on NTU-VIRAL dataset, employing 1 LiDAR.

The best results are marked in bold. It can be seen that Coco-LIC stably outperforms CLIC.

TODO List

• serve as the front-end of incremental 3D Gaussian Splatting（Gaussian-LIC）
• optimize the code architecture (rosbag play mode) and support ikd-tree for acceleration

Citation

If you find our work helpful, please consider citing 🌟:

@article{lang2023coco,
  title={Coco-LIC: continuous-time tightly-coupled LiDAR-inertial-camera odometry using non-uniform B-spline},
  author={Lang, Xiaolei and Chen, Chao and Tang, Kai and Ma, Yukai and Lv, Jiajun and Liu, Yong and Zuo, Xingxing},
  journal={IEEE Robotics and Automation Letters},
  year={2023},
  publisher={IEEE}
}

@article{lv2023continuous,
  title={Continuous-time fixed-lag smoothing for lidar-inertial-camera slam},
  author={Lv, Jiajun and Lang, Xiaolei and Xu, Jinhong and Wang, Mengmeng and Liu, Yong and Zuo, Xingxing},
  journal={IEEE/ASME Transactions on Mechatronics},
  year={2023},
  publisher={IEEE}
}

Acknowledgement

Thanks for Basalt, LIO-SAM, Open-VINS, VINS-Mono, R3LIVE and FAST-LIVO.

LICENSE

The code is released under the GNU General Public License v3 (GPL-3).