Wheel-INS: A Wheel-mounted MEMS IMU-based Dead Reckoning System

Unlike the conventional odometer-aided inertial navigation system (ODO/INS) where the IMU is always placed either on the vehicle body or in the coach, in Wheel-INS, the IMU is mounted on the wheel of the ground vehicle. There are two major advanatges by doing so: 1) the wheel velocity can be obtained by the Wheel-IMU thus replace the traditional odometer (or wheel encoder); 2) the rotation modulation can be leveraged to mitigate the error accumulation of INS. Particularly, the wheel velocity calculated by the gyroscope outputs and the wheel radius is treated as an external observation with non-holonomic constraint (NHC) to fuse with INS through an extended Kalman filter (EKF).

💥News💥

🎉🎉 Oct. 2023 A completely new version of the code supporting both Linux and Windows is released!

🎉🎉 Nov. 2022 Wheel-INS has been extended to SLAM! Check out our paper of Wheel-SLAM accepted to IEEE Robotics and Automation Letters. The source code is also released.

🎉🎉 Nov. 2022 Our paper on multiple IMUs-based wheeled robot localization (Wheel-INS2) has been accepted to IEEE Transactions on Intelligent Transportation Systems. Check it out!

Run it

1. Linux

We recommend you to use g++ compiler with Ubuntu 20.04. The build-essential libraries should be installed first:

sudo apt-get install cmake
sudo apt-get install build-essential

After preparing the compilation environment, you can clone the repository and run it as follows:

# Clone the repository
git clone git@github.com:i2Nav-WHU/Wheel-INS.git ~/

# Build Wheel-INS
cd ~/Wheel-INS
mkdir build && cd build

cmake ..
make -j10

# Run demo dataset
cd ~/Wheel-INS
./bin/Wheel-INS config/robot.yaml

2. Windows

Here we show how to run the code with Visual Studio Code (VSCode), but you can also use other IDEs, e.g., Visual Studio.

Install VSCode and the extensions: C/C++, C/C++ Extension Pack, CMake, and CMake Tools.
Install CMake and Microsoft Visual C/C++ Build Tools.
Open Wheel-INS with VSCode.
Set compiler: open the Command Palette (Ctrl+Shift+P) and type "CMake: Select a Kit", select the correct build tool according to your system.
Configure CMake: type and click "CMake: Configure" in the Command Palette.
Compile Project: type and click "CMake: Build" in the Command Palette.

Once a execuble file Wheel-INS.exe is generated, the compilation is done. Then, you can run it via the terminal in VSCode as following:

.\bin\Release\Wheel-INS.exe config/robot.yaml

You can then run plot.py in utils to plot the trajectory estimated by Wheel-INS as well as the raw Wheel-IMU data. Here is an example plot for the robot dataset.

Datasets

Two sets of example data with ground truth are provided (see dtaset). Please refer to the ReadMe.pdf for details. If git clone is too slow, please try to download the .zip file directly.

Related Papers

Based on the study of Wheel-IMU, we have published three papers. 1) The original Wheel-INS paper where we proposed a wheel-mounted IMU-based dead reckoning system and investigated its characteristics. 2) A thorough and complete comparison on three different measurement models in Wheel-INS with both theoretical analysis and experimental illustration. 3) A multiple IMUs-based localization system for wheeled robots by obtaining different dynamic information of the vhehicle and taking advantage of the relative spatial constraints among the inertial sensors.

If you find our studies helpful to your academic research, please consider citing the related papers.

X. Niu, Y. Wu and J. Kuang, "Wheel-INS: A Wheel-mounted MEMS IMU-based Dead Reckoning System," IEEE Transactions on Vehicular Technology, doi: 10.1109/TVT.2021.3108008, 2021. (pdf) (IEEE Xplore)

@ARTICLE{niu2021tvt,
  author={Niu, Xiaoji and Wu, Yibin and Kuang, Jian},
  journal={IEEE Transactions on Vehicular Technology}, 
  title={{Wheel-INS}: A Wheel-Mounted {MEMS IMU}-Based Dead Reckoning System}, 
  year={2021},
  volume={70},
  number={10},
  pages={9814-9825},
  doi={10.1109/TVT.2021.3108008}
}

Y. Wu, X. Niu and J. Kuang, "A Comparison of Three Measurement Models for the Wheel-mounted MEMS IMU-based Dead Reckoning System," IEEE Transactions on Vehicular Technology, doi: 10.1109/TVT.2021.3102409, 2021. (pdf) (IEEE Xplore)

@ARTICLE{wu2021tvt,
  author={Wu, Yibin and Niu, Xiaoji and Kuang, Jian},
  journal={IEEE Transactions on Vehicular Technology}, 
  title={A Comparison of Three Measurement Models for the Wheel-Mounted {MEMS IMU}-Based Dead Reckoning System}, 
  year={2021},
  volume={70},
  number={11},
  pages={11193-11203},
  doi={10.1109/TVT.2021.3102409}}

Y. Wu, J. Kuang and X. Niu, "Wheel-INS2: Multiple MEMS IMU-based Dead Reckoning System for Wheeled Robots with Evaluation of Different IMU Configurations," IEEE Transactions on Intelligent Transportation Systems, 2022. (pdf)(IEEE Xplore)

@ARTICLE{wu2022tits,  
    author={Wu, Yibin and Kuang, Jian and Niu, Xiaoji},  
    journal={IEEE Transactions on Intelligent Transportation Systems},   
    title={{Wheel-INS2}: Multiple {MEMS IMU}-Based Dead Reckoning System With Different Configurations for Wheeled Robots},   
    year={2022}, 
    pages={1-14},  
    doi={10.1109/TITS.2022.3220508}
}

License

The code is released under GPLv3 license.

Acknowledgement

We would like to thank i2Nav group for sharing KF-GINS, which is referenced by this code.

For any questions, please feel free to contact Mr. Yibin Wu (ybwu at whu.edu.cn) or Dr. Jian Kuang (kuang at whu.edu.cn).

yaoli1992 / Wheel-INS