Vision capabilities and camera handling for the Humanoid Robot project "Energirobotter".
Clone this repository into a workspace/src/
folder, along with realsense-ros (run commands from root of workspace
):
git clone [email protected]:energinet-digitalisering/energirobotter-vision.git src/energirobotter-vision/
git clone https://github.com/IntelRealSense/realsense-ros.git src/realsense-ros/
Setup RealSense SDK:
sudo apt install ros-humble-librealsense2*
In worspace
root, source ROS and install ROS dependencies with rosdep:
source /opt/ros/humble/setup.bash
rosdep install -i --from-path src --rosdistro $ROS_DISTRO --skip-keys=librealsense2 -y
Python modules not included in rosdistro can be installed from root of workspace with:
pip install -r src/energirobotter-vision/requirements.txt
Download face detection model yolov8n-face.pt from the yolo-face repository. Move the model into the src/energirobotter-vision/face_detection/models/
directory.
Build workspace
with:
colcon build --symlink-install
Use the vision_bringup
package's vision.launch.py
to start the camera and face detection:
source install/setup.bash
ros2 launch vision_bringup vision.launch.py use_compressed:=true
If no camera is available, run example with:
ros2 launch vision_bringup vision.launch.py use_mock_camera:=true
Note: The first time
face_detection
is runultralytics
will install som extra packages, and require a restart of the node.