PINS-datasets-based-on-Xsens-IMU

This is a MATLAB dataset of pedestrian inertial navigation systems measured by Xsens MTi-10 IMU.

This repository is the usage page of the PINS-dataset. Since the positioning of GNSS is affected by the multipath effect in complex scenes such as urban canyons, the use of IMUs for positioning has become one of the hotspots in the field of positioning and navigation. Although there are many datasets on IMUs in the current GitHub, there are still fewer datasets on pedestrian navigation systems using foot-tethered IMUs. The data in this dataset are collected from Xsense MTi-10 series IMU developed by Beijing BDStar Navigation Co., Ltd. in circular flower bed and staircase scenes respectively.

This repository is dedicated to provide a public dataset for PINS researchers, if you have any questions about using the dataset, please feel free to contact [email protected].

IMU Parameters

	Gyroscopes	Accelerometers
Standard full range	±450°/s	±20g
Initial bias error	0.2°/s	5mg
In-run bias stability	18°/h	15μg
Bandwidth (-3dB)	415Hz	375Hz
Noise density	0.03°/s√Hz	60μg/√Hz
Non-linearity	0.03%	0.1%

1) Long-distance Promenade Scenario

There are two scenarios for our experimental validation. Firstly, to validate the effectiveness of our method in suppressing the cumulative errors generated by the long-time iterative calculation process, we first conducted a set of long-distance walking experiments in a promenade. In the experiments, a pedestrian walked in a straight line about 197.43 m with the IMU mounted at the top of her foot.

Fig.1. Long-distance Promenade

2) Loop Closure Scenario

In addition, we also set up a set of loop closure experiments. The experimenters also used a foot-mounted IMU to walk around a rectangular square to test the performance of the proposed method in suppressing directional drift.

Fig.2. Loop Closure

Description of the data set format

This experiment mainly uses the three-axis accelerometer and three-axis gyroscope of the IMU, the sampling rate is 500Hz, and the samples are shown in Figure 3. The first three columns of data are respectively the x, y, z-axis acceleration data in the process of pedestrian movement recorded in the first three columns; the second three columns of the x, y, z-axis angular velocity movement data in the process of pedestrian movement recorded in the second three columns.

Fig.3. The samples of dataset

Citations

This dataset is collected for our PINS positioning research, if you use this dataset please cite our paper which will available next year.