First clone this repository using this command:
git clone --recursive <reponame>
Then pull the weights and the bag using this command:
sh get-resources
Available in the FT-Autonomous repo here
This module depends unconditionally upon:
- geometry_msgs (ros)
- sensor_msgs (ros)
- OpenCV
To use the ZED camera rather than the prerecorded bags, you need the ZED SDK, which requires CUDA.
To use the c++ clustering algorithm, you need a version of OpenGL that supports compute shaders as the c++ clustering paralellises a significant amount of the workload on the GPU.
If you are using X11, you can query your OpenGL version using:
glxinfo | grep -i 'Max opengl core profile'
pip3 install -r requirements.txt
If your device does support cuda, first delete the ignore file in the zed_capture folder.
rm src/zed_capture/COLCON_IGNORE
The run the followng:
colcon build
If your device doesn't support cuda, just use the build command directly.
colcon build
. install/setup.sh
To run everything, use ros2 launch ft_peception_synthesis py.
Before building, download the bag from slack/github and then extract it in the folder src/ft_perception_synthesis/bags/static.
