README.md

Instructions for Collecting Dataset from Offline Terrasentia SVO Files

[T#] represents a unique terminal open on the computer. We assume the working directories of your terminals are at decompteleop/ros/catkin_ws and each terminal has the terra-torch3d Conda environment activated.

• Download SVO files collected from the Terrasentia driven under desired conditions
  • Follow the link to explore the Terrasentia Dataset here
• Update paths in svo2sync.py for desired SAVE_DIR and FILENAME representing where extracted data should be saved (the FILENAME should change for every new SVO file)
• [T1] Run roscore
• [T2] Run python src/data_collection/svo2sync.py
• [T3] Run python src/combined_node/publish_transform.py
• [T4] Run source devel/setup.bash
• roslaunch the desired downloaded SVO file to collect frames from
  • Launch with visualization from RTAB-MAP
    • [T4] Run roslaunch rtab_mapping rtab_launch.launch rtabmap_args:="--delete_db_on_start" rgb_topic:=/zed2/zed_node/rgb/image_rect_color depth_topic:=/zed2/zed_node/depth/depth_registered camera_info_topic:=/zed2/zed_node/depth/camera_info odom_topic:=/zed2/zed_node/odom imu_topic:=/zed2/zed_node/imu/data visual_odometry:=false frame_id:=base_link approx_sync:=true rgbd_sync:=true approx_rgbd_sync:=false svo_file:="</path/to/.svo>"
  • Launch with minimal visualization (without RTAB-MAP)
    • [T4] Run roslaunch rtab_mapping rtab_launch_minimal.launch rtabmap_args:="--delete_db_on_start" rgb_topic:=/zed2/zed_node/rgb/image_rect_color depth_topic:=/zed2/zed_node/depth/depth_registered camera_info_topic:=/zed2/zed_node/depth/camera_info odom_topic:=/zed2/zed_node/odom imu_topic:=/zed2/zed_node/imu/data visual_odometry:=false frame_id:=base_link approx_sync:=true rgbd_sync:=true approx_rgbd_sync:=false svo_file:="</path/to/.svo>"
• Data will be collected at the designated folder for the current playback
• Once the desired number of frames have been collected for the current SVO file, close the terminals in reverse order (T4, T3, T2, T1)