Awesome GINS Dataset

An awesome vehicle GINS dataset for GNSS/INS integration applications

This repository provide an awesome GINS dataset for GNSS/INS integration applications. The biggest feature of the dataset is that four different grade of MEMS IMUs are included, which provide an opportunity for the researchers to comprehensively evaluate their algorithms.

The dataset was collected in an open-sky industrial area, where the GNSS RTK was well satisfied. The duration of the whole dataset is 1617 seconds, including the raw IMU data, GNSS RTK positioning results, and the ground-truth for each IMU.

Authors: Hailiang Tang, Xiaoji Niu, and Tisheng Zhang from the Integrated and Intelligent Navigation (i2Nav) Group, Wuhan University.

Related Paper:

• Hailiang Tang, Tisheng Zhang, Xiaoji Niu, Jing Fan, and Jingnan Liu, “Impact of the Earth Rotation Compensation on MEMS-IMU Preintegration of Factor Graph Optimization,” IEEE Sensors Journal, 2022.

If you use this dataset for your academic research, please cite our related papers; and give acknowledgement like:

English version: “The authors would like to acknowledge Dr. Xiaoji Niu and the Integrated and Intelligent Navigation (i2Nav) group from Wuhan University for providing the GNSS/INS dataset that was used in the paper.”
中文模板：“本文作者感谢武汉大学多源智能导航实验室和牛小骥教授提供了本文所需的GNSS/INS数据集。”

Explanation: GitHub do not support Latex formula, and we suggest you read the document with a local Markdown editor, e.g. Typora.

1 The MEMS IMU

Four MEMS IMUs are included in this dataset, and the details about these IMUs are listed as follows:

We also provide a set of noise parameters for each IMU, which are obtained by conducting parameter tuning. Specifically, the IMU measurements are defined as follows: $$ \boldsymbol{\tilde{f}}^{\mathrm{b}}=\boldsymbol{f}^{\mathrm{b}}+\boldsymbol{b}a+\boldsymbol{n}a, \ \boldsymbol{\tilde{w}}{\mathrm{ib}}^{\mathrm{b}}=\boldsymbol{w}{\mathrm{ib}}^{\mathrm{b}}+\boldsymbol{b}_g+\boldsymbol{n}_g $$ where the additive noise $\boldsymbol{n}_a$ and $\boldsymbol{n}_g$ are modeled as Gaussian white noise processes; $\boldsymbol{b}_a$ and $\boldsymbol{b}g$ are respectively the accelerometer and gyroscope biases, which are modeled as first-order Gaussian Markov processes. The IMU noise model can be expressed as follows: $$ \boldsymbol{n}g\sim \mathcal{N} \left( \mathbf{0},\sigma {g}^{2}\boldsymbol{I} \right) , \ \boldsymbol{n}a\sim \mathcal{N} \left( \mathbf{0},\sigma {a}^{2}\boldsymbol{I} \right) , \ \delta \boldsymbol{\dot{b}}{g_t}=-\frac{1}{\tau {b_g}}\delta \boldsymbol{b}{g_t}+\boldsymbol{n}{b_g}, \boldsymbol{n}{b_g}\sim \mathcal{N} \left( \mathbf{0},\sigma {b_g}^{2}\boldsymbol{I} \right) , \ \delta \boldsymbol{\dot{b}}{a_t}=-\frac{1}{\tau {b_a}}\delta \boldsymbol{b}{a_t}+\boldsymbol{n}{b_a}, \boldsymbol{n}{b_a}\sim \mathcal{N} \left( \mathbf{0},\sigma _{b_a}^{2}\boldsymbol{I} \right) $$

The reference noise parameters are listed as follows:

2 Installation parameters

The IMU body frame (b-frame) is defined as forward-right-down (FRD) coordinate. The GNSS geodetic positioning is defined in WGS-84 ellipsoid model. The GNSS antenna lever-arms in b-frame are listed as follows:

3 File format descriptions

Three type of files are contained in the dataset, including the IMU file, the GNSS positioning file, and the ground-truth file.

The IMU text file format (*.txt) is defined as:

The GNSS positioning result text file (*.pos, 7 columns) is defined as:

The ground-truth text file (*.nav) is defined as: