Ground-Fusion

[ICRA2024]Ground-Fusion: A Low-cost Ground SLAM System Robust to Corner Cases

First Author: Jie Yin 殷杰   📝 [Paper] / [Arxiv]   ➡️ [Dataset]   ⭐️ [Presentation Video]   🔥[News]

[!TIP] Check out the presentation video above for a quick overview of this work!

News

• 🔥2026/06/14: We are excited to release a preview version of Ultra-Fusion system, which unifies WIO, VIO, LIO, and LVIO in one configurable optimization framework, with optional wheel/GNSS fusion and online calibration.

• 🔥2025/06/16: Introducing our new paper on IROS2025:"Towards Robust Sensor-Fusion Ground SLAM: A Comprehensive Benchmark and A Resilient Framework" is accepted to IROS2025!code: Ground-Fusion++, dataset: M3DGR!

NOTICE

We warmly welcome and highly recommend the integration of the Ground-Fusion system into your projects for several compelling reasons:

1. 🔥Comprehensive Sensor Suite: The Ground-Fusion system is equipped with a multitude of sensors (RGBD-IMU-Wheel-GNSS), facilitating an easy onset for enhancements to any module. This richness in sensory input streamlines the process of adapting and refining components within the system.

2. ⭐️Open-Source Ecosystem: Both the Ground-Fusion algorithm and associated datasets such as M2DGR-plus and the Ground-Challenge are openly available, forming a comprehensive bechmark suite. Welcome to beat Ground-Fusion on M2DGR and Ground-Challenge!

3. 🚀Proven Performance: The Ground-Fusion algorithm has been rigorously validated across diverse datasets, establishing itself as SOTA in lidar-less SLAM algorithms. Outperforming Ground-Fusion on these benchmarks would significantly bolster the credibility of your proposed method.

Introduction

We introduce Ground-Fusion, a low-cost sensor fusion simultaneous localization and mapping (SLAM) system for ground vehicles. Our system features efficient initialization, effective sensor anomaly detection and handling, real-time dense color mapping, and robust localization in diverse environments. We tightly integrate RGB-D images, inertial measurements, wheel odometer and GNSS signals within a factor graph to achieve accurate and reliable localization both indoors and outdoors. To ensure successful initialization, we propose an efficient strategy that comprises three different methods: stationary, visual, and dynamic, tailored to handle diverse cases. Furthermore, we develop mechanisms to detect sensor anomalies and degradation, handling them adeptly to maintain system accuracy.

The preprint version of paper is here. The dataset is at M2DGR-plus, Ground-Challenge and M2DGR.

⚙️ 1. Prerequisites & Installation

1.1 OS & ROS Support

✅ Ubuntu 18.04 + ROS Melodic + OpenCV3
✅ Ubuntu 20.04 + ROS Noetic + OpenCV4

1.2 OpenCV

This package requires OpenCV 3/4 and some features of C++11.

1.3 Eigen, Ceres, and PCL

This package requires Eigen 3.3.7, Ceres 1.14,Sophus and PCL 1.10 or 1.11. You need to download they in your thirdparty folder, and then:

sudo apt-get update
sudo apt-get install -y cmake libgoogle-glog-dev libgflags-dev libatlas-base-dev libsuitesparse-dev 
cd thirdparty/eigen
mkdir -p build && cd build
cmake ..
sudo make install
cd ../../ceres-solver
mkdir -p build && cd build
cmake ..
make -j$(nproc) 
sudo make install
sudo apt-get install -y libflann-dev libvtk6-dev libboost-all-dev ros-noetic-pcl-ros (for ubuntu20.04) libfmt-dev
cd ../../pcl
mkdir -p build && cd build
cmake ..
make -j$(nproc)
sudo make install
cd ../../Sophus
mkdir -p build && cd build
cmake ..
make -j$(nproc) 
sudo make install

1.4 Gnss_comm

This package also requires gnss_comm for ROS message definitions and some utility functions.

1.5 Configure gcc (For Ubuntu 20.04)

sudo apt-get install g++-8
sudo apt-get install gcc-8
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-8 20
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-8 20

1.6 Build Ground-Fusion

After install all 3rd parties：

mkdir -p ~/Groundfusion_ws/src
cd ~/Groundfusion_ws/src
git clone https://github.com/HKUST-Aerial-Robotics/gnss_comm
git clone https://github.com/SJTU-ViSYS/Ground-Fusion
cd ../..
catkin_make -j12

💡 Tip: If you encounter Sophus issues, try building both template and non-template versions.

🚀 2. Run Examples

2.1 Ground-challenge dataset

Download Ground-challenge dataset and give a star.

# [launch] open a terminal and type:
source devel/setup.bash
roslaunch vins groundfusion.launch

# [run localization] open another terminal:
source devel/setup.bash
rosrun vins vins_node src/Ground-Fusion/config/realsense/groundchallenge.yaml

# [dense map]open third terminal:
source devel/setup.bash
rosrun dense_map dense_map_node src/Ground-Fusion/config/realsense/groundchallenge.yaml

# [play rosbag]open forth terminal:
rosbag play office3.bag

⚠️ Mapping node is resource-intensive, affecting the real-time performance of the entire system. So it's suggested that do not run this node unless necessary. We are working on optimizing the mapping node currently

2.2 M2DGR-Plus dataset

Download M2DGR-Plus dataset and give a star.

# [launch] open a terminal and type:
source devel/setup.bash
roslaunch vins groundfusion.launch

# [run localization] open another terminal:
source devel/setup.bash
rosrun vins vins_node src/Ground-Fusion/config/realsense/m2dgrp.yaml

# [dense map]open third terminal:
source devel/setup.bash
rosrun dense_map dense_map_node src/Ground-Fusion/config/realsense/m2dgrp.yaml

# [play rosbag]open forth terminal:
rosbag play anamoly.bag

⚠️ Known Issues:

• Ground-Fusion may perform better without GNSS on M2DGR-Plus due to its low (1Hz) GNSS frequency.
• Initialization thresholds may need tuning to avoid early-phase drift.
• A more advanced version of Ground-Fusion is under development — stay tuned!

---

🙏 3.Acknowledgements

This work is supported by:

• National Key R&D Program (2022YFB3903802)
• NSFC (62073214)
• Midea Group

This project is built on GVINS, with inspirations from VIW-Fusion and VINS-RGBD — huge thanks!

---

📄 4. Citation

@inproceedings{yin2024ground,
  title={Ground-fusion: A low-cost ground slam system robust to corner cases},
  author={Yin, Jie and Li, Ang and Xi, Wei and Yu, Wenxian and Zou, Danping},
  booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)},
  pages={8603--8609},
  year={2024},
  organization={IEEE}
}
@article{zhang2025towards,
  title={Towards Robust Sensor-Fusion Ground SLAM: A Comprehensive Benchmark and A Resilient Framework},
  author={Zhang, Deteng and Zhang, Junjie and Sun, Yan and Li, Tao and Yin, Hao and Xie, Hongzhao and Yin, Jie},
  journal={arXiv preprint arXiv:2507.08364},
  year={2025}
}
@article{yin2021m2dgr,
  title={M2dgr: A multi-sensor and multi-scenario slam dataset for ground robots},
  author={Yin, Jie and Li, Ang and Li, Tao and Yu, Wenxian and Zou, Danping},
  journal={IEEE Robotics and Automation Letters},
  volume={7},
  number={2},
  pages={2266--2273},
  year={2021},
  publisher={IEEE}
}
@inproceedings{yin2023ground,
  title={Ground-challenge: A multi-sensor slam dataset focusing on corner cases for ground robots},
  author={Yin, Jie and Yin, Hao and Liang, Conghui and Jiang, Haitao and Zhang, Zhengyou},
  booktitle={2023 IEEE International Conference on Robotics and Biomimetics (ROBIO)},
  pages={1--5},
  year={2023},
  organization={IEEE}
}

5. Star History