LVI_SAM

To run

• In one terminal:

cd ~/lvisam_ws
source devel/setup.bash
roslaunch lvi_sam Husky.launch

• In the other terminal:

cd ~/rosbags/lvi-sam
rosbag play husky.bag

• Useful TF commands:

rosrun rqt_tf_tree rqt_tf_tree
rosrun tf view_frames

To evaluate

# pcd转txt，位于~/lvi-sam/src/LVI-SAM-Easyused/pcd2tum.py
# pip install pyntcloud
python pcd2tum.py

• 安装evo评估工具

# https://github.com/MichaelGrupp/evo
pip install evo --upgrade --no-binary evo

• 拷贝真值轨迹, gt.txt

~/lvi-sam/results/gt.txt

• 计算误差

# -r full (旋转+平移) trans_part (平移m) angle_deg (旋转deg)
evo_ape tum gt.txt lvisam.txt -r full -va --plot --plot_mode xy --save_plot ./lvisamplot
# 多轨迹绘制
# evo_traj tum lvisam.txt fastlio2.txt kissicp.txt --ref=gt.txt -va -p --plot_mode=xy --save_plot ./trajall

Parameters

• Husky_camera.yaml and Husky_lidar.yaml
• Camera intrinsics
• Camera-IMU, LiDAR-IMU extrinsics

Acknowledgement

• Original LVI-SAM
• Modified LVI-SAM-Easyused