# FASTEX **Repository Path**: wrd666/FASTEX ## Basic Information - **Project Name**: FASTEX - **Description**: No description available - **Primary Language**: Unknown - **License**: GPL-3.0 - **Default Branch**: main - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2026-04-21 - **Last Updated**: 2026-08-21 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README

FASTEX: Fast UAV Exploration in Large-Scale Environments Using Dynamically Expanding Grids and Coverage Paths

Xiaoxun Zhang, Peiming Duan, Lanxiang Zheng, Junlong, Huang, and Hui, Chengā€ 

ā€ Corresponding Author

Paper License IROS 2025 ROS Noetic Ubuntu 20.04

--- ## šŸ“– Overview **FASTEX** is an efficient and robust autonomous exploration framework for quadrotor UAVs operating in large-scale, complex environments such as caves, dense forests, and cluttered indoor spaces. The system leverages **dynamically expanding grids** for scalable environment representation and **incremental coverage path planning** to achieve rapid, complete exploration. --- ## šŸ“¦ Prerequisites ### System Requirements - **OS**: Ubuntu 20.04 (recommended) - **ROS**: [ROS Noetic](http://wiki.ros.org/noetic/Installation/Ubuntu) - **Docker** (optional but recommended): 20.10+ ### Local Dependencies (Non-Docker) ```bash # System libraries sudo apt update && sudo apt install -y \ git tmux cmake wget build-essential \ pcl-tools libpcl-dev libdw-dev libtbb-dev libelf-dev \ libglew-dev libglfw3-dev libblosc-dev libeigen3-dev \ libspdlog-dev libopenexr-dev libarmadillo-dev liblog4cplus-dev \ libatlas-base-dev libsuitesparse-dev libgoogle-glog-dev \ libignition-common3-graphics-dev libignition-common3-profiler-dev \ python3-tk python3-pip python3-pandas python3-wstool python3-matplotlib \ python3-catkin-tools # ROS packages sudo apt install -y \ ros-noetic-rosfmt ros-noetic-rviz ros-noetic-navigation \ ros-noetic-lms1xx ros-noetic-message-filters ros-noetic-tf # NLopt v2.7.1 (required for trajectory optimization) git clone https://github.com/stevengj/nlopt.git --branch v2.7.1 cd nlopt && mkdir build && cd build cmake .. && make -j$(nproc) && sudo make install ``` --- ## šŸ”Ø Build ### Option 1: Docker Build (Recommended) All dependencies are handled inside the container. No local installation required. ```bash cd docker make build ``` See [`docker/README.md`](docker/README.md) for detailed Docker usage. See [`docker/Makefile`](docker/Makefile) and [`docker/cave_exp.Dockerfile`](docker/cave_exp.Dockerfile) for build customization. ### Option 2: Native Build ```bash # Create and configure catkin workspace mkdir -p ~/fast_ex_ws/src cd ~/fast_ex_ws/src git clone https://github.com/zxx01/FASTEX.git # Build cd ~/fast_ex_ws catkin config -DCMAKE_BUILD_TYPE=Release catkin build source devel/setup.bash ``` --- ## šŸŒ Environment PCD Files The environment point-cloud files used by the released scenes can be downloaded here: - [Google Drive: FASTEX environment PCD files](https://drive.google.com/drive/folders/1pDnEowW5wwxa1IzXnkwg7daS7pv43kwY?usp=sharing) Please place the downloaded scene files under [`marsim/map_generator/resource/`](marsim/map_generator/resource/) before running the `cave`, `forest`, or `indoor` experiments. Expected filenames for the maintained scenes are: - `cave`: `cave.pcd` - `forest`: `forest_ds.pcd` - `indoor`: `indoor.pcd` --- ## šŸš€ Run ### Quick Start Launch exploration directly via `roslaunch`: ```bash # Cave exploration source devel/setup.bash && roslaunch exploration_manager cave_exploration.launch # Forest exploration source devel/setup.bash && roslaunch exploration_manager forest_exploration.launch # Indoor exploration source devel/setup.bash && roslaunch exploration_manager indoor_exploration.launch ``` ### Docker Compose (Multi-Scene) First ensure the Docker image is built (see [Build](#-build) above), then: ```bash # Single scene bash docker/run_lca_scenes.sh start forest # Multiple scenes bash docker/run_lca_scenes.sh start cave forest indoor ``` See [`docker/run_lca_scenes.sh`](docker/run_lca_scenes.sh) and [`docker/lca_scenes.compose.yml`](docker/lca_scenes.compose.yml) for scene orchestration details. ### Launch File Arguments Key ROS parameters available in launch files ([`fastex_planner/applications/exploration_manager/launch/`](fastex_planner/applications/exploration_manager/launch/)): | Parameter | Default | Description | |-----------|---------|-------------| | `is_launch_rviz` | `true` | Launch RViz for visualization | | `is_logging` | `false` | Enable exploration data logging | | `is_launch_data_vis` | `true` | Launch data visualization tools | | `auto_start` | `true` | Automatically start exploration | | `max_vel` | `2.0` | Maximum flight velocity (m/s) | | `max_acc` | `1.5` | Maximum acceleration (m/sĀ²) | | `sdf_resolution` | `0.4` | ESDF map resolution (m) | --- ## āš™ļø Configuration ### Exploration Parameters Located in [`fastex_planner/applications/exploration_manager/launch/`](fastex_planner/applications/exploration_manager/launch/) ā€” adjust per-scenario launch files to tune exploration behavior: - [`launch/cave/cave_exploration.launch`](fastex_planner/applications/exploration_manager/launch/cave/cave_exploration.launch) - [`launch/forest/forest_exploration.launch`](fastex_planner/applications/exploration_manager/launch/forest/forest_exploration.launch) - [`launch/indoor/indoor_exploration.launch`](fastex_planner/applications/exploration_manager/launch/indoor/indoor_exploration.launch) ### Map Preprocessing Parameters Located in [`fastex_planner/planning/map_process/resources/`](fastex_planner/planning/map_process/resources/): | File | Scene | |------|-------| | [`config_cave.yaml`](fastex_planner/planning/map_process/resources/config_cave.yaml) | Cave exploration | | [`config_forest.yaml`](fastex_planner/planning/map_process/resources/config_forest.yaml) | Forest exploration | | [`config_indoor.yaml`](fastex_planner/planning/map_process/resources/config_indoor.yaml) | Indoor exploration | Key tuning sections include: - **DynamicExpandingGrid** ā€” Grid resolution, expansion bounds, consistency cost - **FrontierManager** ā€” Visible voxel thresholds, cluster radius, viewpoint sampling - **WSRoadmap** ā€” Roadmap sampling density, connectivity range, clearance constraints ### Exploration Data Logging Run-time metrics are stored under [`fastex_planner/applications/data_log_manager/exploration_data_files/`](fastex_planner/applications/data_log_manager/exploration_data_files/), including per-iteration preprocessing time and planning time for performance analysis. --- ## šŸ“š Citation If you use FASTEX in your research, please cite our paper: ```bibtex @inproceedings{zhang2025fastex, title = {{FASTEX}: Fast {UAV} Exploration in Large-Scale Environments Using Dynamically Expanding Grids and Coverage Paths}, author = {Zhang, Xiaoxun and Duan, Peiming and Zheng, Lanxiang and Huang, Junlong and Cheng, Hui}, booktitle = {2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, pages = {3541--3548}, year = {2025}, organization = {IEEE} } ``` --- ## šŸ“ Notes 1. Exploration and planning parameters are concentrated in [`fastex_planner/applications/exploration_manager/launch/`](fastex_planner/applications/exploration_manager/launch/). Map preprocessing parameters reside in [`fastex_planner/planning/map_process/resources/`](fastex_planner/planning/map_process/resources/). Adjust them according to your specific scenario. 2. For detailed parameter descriptions, refer to [`fastex_planner/planning/map_process/resources/config_cave.yaml`](fastex_planner/planning/map_process/resources/config_cave.yaml) for annotated comments. 3. The [`fastex_planner/applications/data_log_manager/exploration_data_files/`](fastex_planner/applications/data_log_manager/exploration_data_files/) directory stores preprocessed timing data and per-iteration planning time logs for performance evaluation. 4. If you encounter Docker permission issues, follow the setup steps in [`docker/README.md`](docker/README.md) and [`docker/Makefile`](docker/Makefile) to configure user groups. --- ## šŸ™ Acknowledgments We would like to express our gratitude to the following projects, which have provided significant support and inspiration for our work: - [**FUEL**](https://github.com/HKUST-Aerial-Robotics/FUEL) ā€” An efficient framework for fast UAV exploration, leveraging Frontier Information Structure (FIS) and hierarchical planning. - [**MARSIM**](https://github.com/hku-mars/MARSIM) ā€” A light-weight point-realistic simulator for LiDAR-based UAVs, supporting diverse realistic maps and sensor models. - [**FALCON**](https://github.com/HKUST-Aerial-Robotics/FALCON) ā€” Fast autonomous aerial exploration using coverage path guidance. --- ## šŸ“§ Contact For questions and collaborations, please contact the corresponding author or open an issue on GitHub.