DHP-Mapping: A Dense Panoptic Mapping System with Hierarchical World Representation and Label Optimization Techniques
🎊 We plan to integrate Grounded-Segment-Anything, Yolo-World, and OWLv2 into this mapping pipeline, for online open-world semantic-mapping. Additionally, we will support Habitat-Lab to advance developments in embodied-AI. Stay tuned for our upcoming feature releases and don’t forget to give us a star!
🔥 27/12/2024 Release experiment configs and launch files.
🔥 26/11/2024 Release main algorithms!
🤗 15/10/2024 Presented at iros2024-abudhabi
🚀 30/06/2024 Accepted by IROS2024!
📜 26/03/2024 arXiv version paper
SemanticKITTI: Prepare the dataset to desired format, following dataset.
flat: Download the dataset from the ASL Datasets.
sudo apt-get install python3-pip python3-wstool
mkdir -p DHP_ws/src
cd DHP_ws/src
git clone git@github.com:hutslib/DHP-mapping.git
wstool init . ./DHP-mapping/panoptic_mapping_ssh.rosinstall # SSH
cd ..
catkin init
catkin config --extend /opt/ros/noetic
catkin config --cmake-args -DCMAKE_BUILD_TYPE=RelWithDebInfo
catkin config --merge-devel
catkin build panoptic_mapping_utils
source ../devel/setup.zsh
The launch files are located in: panoptic_mapping_ros/launch/iros_exp
The config files are located in: panoptic_mapping_ros/config/mapper/iros_exp
- Update
base_path
andconfig
in the launch file. - Note: For long sequences in SemanticKITTI, adjust
max_frames
in the launch file to limit the number of frames based on your device's memory. - Update the following in the config file:
save_map_path_when_finished
label_info_print_path
save_mesh_folder_path
submap_info_path
Tr
P2
labels: file_name
visualization: colormap_print_path
- Note: The
Tr
andP2
parameters should be modified based on the extrinsics calculated in the data preprocess step.
# run the experiment
roslaunch panoptic_mapping_ros iros_exp_kitti_detectron_07.launch
The code is under cleaning and will be released gradually.
- Initial repo & paper
- Dataset process & player & visualization
- Main algorithms code
- Config and launch files
- Visualization and other tools
- Improve docs
- Feature imporve: integrate open-world segmentation/detection algorithms
- Feature improve: supporting Habitat-Lab
If you find this repo useful, please consider giving us a star 🌟 and citing our related paper.
@misc{hu2024dhpmapping,
title={DHP-Mapping: A Dense Panoptic Mapping System with Hierarchical World Representation and Label Optimization Techniques},
author={Tianshuai Hu and Jianhao Jiao and Yucheng Xu and Hongji Liu and Sheng Wang and Ming Liu},
year={2024},
eprint={2403.16880},
archivePrefix={arXiv},
primaryClass={cs.RO}
}
This implementation is based on codes from several repositories. Thanks to these authors who kindly open-sourcing their work to the community. 💕 Thanks to panoptic_mapping, Voxfield Panmap, Semantic 3D mapping