1. stereo vo

1.1 steps

• get image sequence
• feature detection
• feature matching or tracking
• motion estimation(3d-2d correspondence)
• Local Optimiation(Bundle adjustment)

1.2 Building

cd catkin_ws/src
git clone [email protected]:libing64/stereo_vo.git
cd ..
catkin_make -DCATKIN_WHITELIST_PACKAGES="stereo_vo"

1.3 Running with kitti dataset

modify the dataset_folder in stereo_vo_kitti.launch

soure devel/setup.bash
roslaunch stereo_vo stereo_vo_kitti.launch

1.5 Test Environment

Ubuntu 20.04 + ros noetic

2. TODO

2.1 How to make the estimator more robust and accurate?

• local sparse bundle adjustment
• record screen to gif
• mapping
• fusing imu