⚔️CLID-SLAM: A Coupled LiDAR-Inertial Neural Implicit Dense SLAM with Region-Specific SDF Estimation

Mesh	Neural Points

TODO 📝

• Release the source code
• Enhance the README.md
• Include the theory derivations

Installation

Platform Requirements

• Ubuntu 20.04
• GPU (tested on RTX 4090)

Steps

1. Clone the repository

   git clone git@github.com:DUTRobot/CLID-SLAM.git
   cd CLID-SLAM

2. Create Conda Environment

   conda create -n slam python=3.12
   conda activate slam

3. Install PyTorch

   pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126

4. Install ROS Dependencies

   sudo apt install ros-noetic-rosbag ros-noetic-sensor-msgs

5. Install Other Dependencies

   pip3 install -r requirements.txt

Data Preparation

Download ROSbag Files

Download these essential ROSbag datasets:

• Newer College Dataset
• SubT-MRS Dataset

Convert to Sequences

1. Edit ./dataset/converter/config/rosbag2dataset.yaml.
2. Run:

   python3 ./dataset/converter/rosbag2dataset_parallel.py
   

Run CLID-SLAM

python3 slam.py ./config/run_ncd128.yaml

Acknowledgements 🙏

This project is built upon the open-source project PIN-SLAM, developed by PRBonn/YuePanEdward. A huge thanks to the contributors of PIN-SLAM for their outstanding work and dedication!