Start by going into the SLAM session created in the SLAM step (check 3. Run the SLAM)

cd ./slam_session_tutorial

Create a dedicated Loop closure sub-directory and go in it:

mkdir ./1_loop_closure
cd ./1_loop_closure

The exwayz_loop_closure.exe executable expects

• A directory of frames created by exwayz_slam.exe
• A support trajectory that will be used to “guess” where are the loops to optimize. This trajectory could be the traj_odometry.ply trajectory output by the exwayz_slam, but it could also stem from a pre-georeferencing step (check Pre-georeferencing for helping the loop closure)

Run the loop closure with the following command

exwayz_loop_closure -i ../frames -t ../traj_odometry.ply -p Odometry -d 5.0

In your console, you should observe that loops are detected between pairs of trajectory points.

As an output, you’ll have in the current directory

• Loop closure observations, that can be provided in the next step to the trajectory optimizer
• An optimized trajectory

Once the Loop closure is over, the SLAM session should be structured as follow :

📂 slam_session_<hour>_<date>

• 📂 1_loop_closure
• 📂 frames
• 📂 log