HLoc

For the MuYuan contract, improvements have been made to the Hloc algorithm.

Reconstruction: Block reconstruction and scene merging are aided by estimating camera poses from other sources.
Localization: A coarse-to-fine localization approach is adopted to enhance localization speed.

Data collection platform

Estimated Pose vs. Real Pose Comparison Chart

Hardware operation commands

Data Collection

docker(env)

Creating a container through mirroring:

docker run -it  --net=host  \
    --name collect \ 
    -v /data:/data  \
    -v /tmp/.X11-unix:/tmp/.X11-unix  \
    -v /dev:/dev \
    --privileged=true \
    --ipc=host \
    -e DISPLAY=$DISPLAY \
   harbor.cvgl.lab/library/vslam-data-collection-platform:v1.1.3

Starting a container:

docker start collect

Running a Docker image:

docker exec -it collect bash

ROS

Basic Command

Configuring the workspace:

source /ws/install/setup.bash

Viewing all current topics:

ros2 topic list

Viewing the frequency of information sent for a specific topic:

ros2 topic hz /imu/data

Printing the content of a specific topic:

ros2 topic epoch /imu/data

Data collection

Start the left camera node.

ros2 launch pylon_ros2_camera_wrapper 1_pylon_ros2_camera.launch.py

Start the right camera node.

ros2 launch pylon_ros2_camera_wrapper 2_pylon_ros2_camera.launch.py

Record camera data packets.

cd /data/bagfiles 
ros2 bag record /my_camera_1/pylon_ros2_camera_node/image_raw \
                /my_camera_2/pylon_ros2_camera_node/image_raw

ros2 bag record /my_camera_1/pylon_ros2_camera_node/image_raw \
                /my_camera_2/pylon_ros2_camera_node/image_raw \
                /imu/data /filter/positionlla

It is recommended to use the method described in the "multi-cameras" section.

Start the LiDAR node.

ros2 launch livox_ros_driver2 msg_MID360_launch.py

Record LiDAR data packets.

ros2 bag record /livox/lidar

Start the IMU node.

usermod -aG dialout root
chmod 777 /dev/ttyUSB0
ros2 launch bluespace_ai_xsens_mti_driver xsens_mti_node.launch.py

Record IMU data packets.

ros2 bag record /imu/data

Record GPS data packets.

ros2 bag record /gnss

Data visualization.

Data visualization is only supported for playback of camera-recorded data packets.

Execute a specific data packet.

ros2 bag play *.db3 --loop  # The `--loop` option is used to specify looping playback.

Start RViz for visualization.

ros2 run rviz2 rviz2

# You may need to execute the following command on the local host to grant permissions to the X server.
xhost +

Displaying data packet information.

ros2 bag info *.db3

Data decompression.

Install GitHub - Kyungpyo-Kim/ROS2BagFileParsing: ROS2 Bag file parsing first.

You can use the following command for decompression:

source /ROS2BagFileParsing/dev_ws/src/ros2bag_parser/install/setup.bash
# alias DC

ros2 run ros2bag_parser ros2bag_parser_node

Usage: 
/ROS2BagFileParsing/dev_ws/src/ros2bag_parser/install/ros2bag_parser/lib/ros2bag_parser/ros2bag_parser_node \
<db3 file path> <output dir>

It is important to note that when decompressing camera data packets, it is required to have a "data" folder specified in the target directory. However, for the IMU-generated CSV files, they do not need to be placed within the "data" folder.

Camera calibration.

Method One

Start the corresponding Docker container.

docker stop wonderful_turing
docker start wonderful_turing

Interactively execute Docker commands.

docker exec -it wonderful_turing bash

And

source /catkin_ws/devel/setup.bash

rosrun kalibr kalibr_calibrate_cameras \
    --target /catkin_ws/src/target.yaml \
    --models pinhole-equi pinhole-equi \
    --topics /my_camera_1/pylon_ros2_camera_node/image_raw /my_camera_2/pylon_ros2_camera_node/image_raw \
    --bag /path/to/your/bag \
    --bag-freq 4.0 \
    --show-extraction

The path after "bag" needs to be replaced with your own path. If you are using ROS2 recording, you need to use rosbags-convert to convert the .db3 file to a .bag file (you can refer to https://github.com/ethz-asl/kalibr/wiki/ROS2-Calibration-Using-Kalibr for more details). The specific command is:

rosbags-convert <ros2_bag_folder> --dst calib_01.bag

Method Two

Run the following command directly in the collect_data container:

# Monocular.
ros2 run camera_calibration cameracalibrator --size 10x10 --square 0.06 --ros-args --remap image:=/my_camera_2/pylon_ros2_camera_node/image_raw --remap camera:=/my_camera_2/pylon_ros2_camera_node

# Stereo.
ros2 run camera_calibration cameracalibrator --size 10x10 --square 0.06 --ros-args --remap right:=/my_camera_2/pylon_ros2_camera_node/image_raw --remap left:=/my_camera_1/pylon_ros2_camera_node/image_raw --remap left_camera:=/my_camera_1/pylon_ros2_camera_node --remap right_camera:=/my_camera_2/pylon_ros2_camera_node

Record a video in real-time using a chessboard grid. Once all the progress bars turn green, proceed with the calibration.

Camera time synchronization.

It is recommended to follow the method described in the "multi-cameras" section, which does not require executing the following steps.

Execute the following commands sequentially to synchronize the camera timestamps with the host computer, record the data packets, and obtain the image titles as timestamps after decompression:

ros2 service call /my_camera_1/pylon_ros2_camera_node/enable_ptp std_srvs/srv/SetBool "{data: true}"

ros2 service call /my_camera_1/pylon_ros2_camera_node/set_periodic_signal_period pylon_ros2_camera_interfaces/srv/SetFloatValue "{value: 50000}"

ros2 service call /my_camera_1/pylon_ros2_camera_node/set_periodic_signal_delay pylon_ros2_camera_interfaces/srv/SetFloatValue "{value: 0}"

ros2 service call /my_camera_1/pylon_ros2_camera_node/set_trigger_selector pylon_ros2_camera_interfaces/srv/SetIntegerValue "{value: 0}"

ros2 service call /my_camera_1/pylon_ros2_camera_node/set_trigger_mode std_srvs/srv/SetBool "{data: true}"

ros2 service call /my_camera_1/pylon_ros2_camera_node/set_trigger_source pylon_ros2_camera_interfaces/srv/SetIntegerValue "{value: 5}"



ros2 service call /my_camera_2/pylon_ros2_camera_node/enable_ptp std_srvs/srv/SetBool "{data: true}"

ros2 service call /my_camera_2/pylon_ros2_camera_node/set_periodic_signal_period pylon_ros2_camera_interfaces/srv/SetFloatValue "{value: 50000}"

ros2 service call /my_camera_2/pylon_ros2_camera_node/set_periodic_signal_delay pylon_ros2_camera_interfaces/srv/SetFloatValue "{value: 0}"

ros2 service call /my_camera_2/pylon_ros2_camera_node/set_trigger_selector pylon_ros2_camera_interfaces/srv/SetIntegerValue "{value: 0}"

ros2 service call /my_camera_2/pylon_ros2_camera_node/set_trigger_mode std_srvs/srv/SetBool "{data: true}"

ros2 service call /my_camera_2/pylon_ros2_camera_node/set_trigger_source pylon_ros2_camera_interfaces/srv/SetIntegerValue "{value: 5}"

# alias PTP

ros2 run rqt_image_view rqt_image_view

isxulu / hloc Goto Github PK

hloc's Introduction