EMV-LIO is an Efficient Multiple vision aided LiDAR-inertial odometry system based on LVI-SAM, which introduces multiple cameras in the VIO subsystem to expand the range of visual observation to guarantee the whole system can still maintain the relatively high accuracy in case of the failure of the monocular visual observation. Apart from this, an efficiency-enhanced LVIO system is also introduced to increase the system’s efficiency, including removing LiDAR’s noise via range image, setting condition for nearest neighbor search, and replacing kd-Tree with ikd-Tree.
Our implementation will be available upon acceptance
Ubuntu 64-bit 16.04 or 18.04. ROS Kinetic or Melodic. ROS Installation and its additional ROS pacakge:
Follow Ceres Installation.
sudo apt-get install -y libgoogle-glog-dev
sudo apt-get install -y libatlas-base-dev
wget -O ~/Downloads/ceres.zip https://github.com/ceres-solver/ceres-solver/archive/1.14.0.zip
cd ~/Downloads/ && unzip ceres.zip -d ~/Downloads/
cd ~/Downloads/ceres-solver-1.14.0
mkdir ceres-bin && cd ceres-bin
cmake ..
sudo make install -j4
Install the dependencies
sudo apt-get install libboost-all-dev
sudo apt-get install cmake
sudo apt-get install libtbb-dev
Compile the GTSAM's code
git clone https://bitbucket.org/gtborg/gtsam.git
cd gtsam/
mkdir build &&cd build
cmake ..
make check
sudo make install
You can use the following commands to download and compile the package.
cd ~/emv_ws/src
git clone https://github.com/BingqiShen/EMV-LIO.git
cd ..
catkin_make -j4
The datasets used in the paper can be downloaded from Baidu Drive. The data-gathering sensor suite includes: HESAI PandarXT-32 LiDAR, DAHENG MER2-202 camera, and Xsens MTi-300 IMU.
url：https://pan.baidu.com/s/1QjQzn1ZwN1SHqHPghYN2tw
code：sreu
- Configure parameters:
Configure sensor parameters in the .yaml files in the ```config``` folder.
You can select the number of camera used by change NUM_OF_CAM in params_lidar.yaml
- Run the launch file:
roslaunch emv_lio run.launch
- Play existing bag files:
rosbag play iplus.bag
Our repository mainly develops from LVI-SAM, where the visual-inertial odometry module is adapted from Vins-Mono and the lidar-inertial odometry module is adapted from LIO-SAM. Besides, our implementation also use the codes of ikd-Tree, M-LOAM, and Cartographer.