M2DGR-plus

Extension and update of M2DGR: a novel Multi-modal and Multi-scenario SLAM Dataset for Ground Robots

First Author: Jie Yin

Figure 1. Acquisition Platform and Diverse Scenarios.

NOTICE

The full dataset with GT trajectories and calibration results will be made public upon paper acceptance. Due to space limitation, dataset descriptions are not elaborated in the main paper. Instead, the details are shown in this website.

This paper has been accepted by ICRA2024! So I decide to release the dataset with its calibration results and GT trajectories right now. Feel free to utilize this full dataset to facilitate your research on SLAM. Please give me a star if you like it.

1.LICENSE

This work is licensed under MIT license. International License and is provided for academic purpose. If you are interested in our project for commercial purposes, please contact us on [email protected] for further communication.

2.SENSOR SETUP

The calibration results are here. All the sensors and track devices and their most important parameters are listed as below:

• LIDAR Robosense 16, 360 Horizontal Field of View (FOV),-30 to +10 vertical FOV,10Hz,Max Range 200 m,Range Resolution 3 cm, Horizontal Angular Resolution 0.2°.
• GNSS Ublox F9p, GPS/BeiDou/Glonass/Galileo, 1Hz
• V-I Sensor,Realsense d435i,RGB/Depth 640*480,69H-FOV,42.5V-FOV,15Hz;IMU 6-axix, 200Hz
• IMU,wheeltec,9-axis,50Hz;
• GNSS-IMU Xsens Mti 680G. GNSS-RTK,localization precision 2cm,100Hz;IMU 9-axis,100 Hz;
• Motion-capture System Vicon Vero 2.2, localization accuracy 1mm, 50 Hz;

The rostopics of our rosbag sequences are listed as follows:

• 3D LIDAR: /rslidar_points

• 2D LIDAR: /scan

• Odom: /odom

• GNSS Ublox F9p:
  /ublox_driver/ephem ,

/ublox_driver/glo_ephem ,

/ublox_driver/range_meas ,

/ublox_driver/receiver_lla ,

/ublox_driver/receiver_pvt

• V-I Sensor:
  /camera/color/image_raw,
  /camera/imu

• IMU: /imu

3.DATASET SEQUENCES

Sequence Name	Collection Date	Total Size	Duration	Features	Rosbag	GT
Switch	2023-8	9.5g	292s	indoor-outdoor switch	Rosbag	GT
Tree	2023-8	3.7g	160s	Dense tree leave cover	Rosbag	GT
Bridge_01	2022-11	2.4g	75s	Bridge, Zigzag	Rosbag	GT
Bridge_02	2022-11	16.0g	501s	Bridge, Long-term,Straight line	Rosbag	GT
Street_01	2022-11	1.7g	58s	Street, Straight line	Rosbag	GT
Street_02	2022-11	3.9g	126s	Bridge, Sharp turn	Rosbag	GT
Parking_01	2022-11	3.3g	105s	Parking lot, Side moving	Rosbag	GT
Parking_02	2022-11	5.4g	149s	Parking lot, Rectangle loop	Rosbag	GT
Building_01	2022-11	3.7g	120s	Building, Far features	Rosbag	GT
Building_02	2022-11	3.4g	110s	Building, Far features	Rosbag	GT

4. EXPERIMENTAL RESULTS

We test methods with diverse senser settings to validate our benchmark dataset. Results shown that our dataset is a valid and effective testfield for localization methods.

And in some cases, our Ground-Fusion achieves comparable performance to Lidar SLAM!

Figure 2. The ATE RMSE (m) result on some sequences.

Figure 3. The visualized trajectory.

5.DEVELOPMENT TOOLKITS

5.1 Extracting Images

• For rosbag users, first make image view

roscd image_view
rosmake image_view
sudo apt-get install mjpegtools

open a terminal,type roscore.And then open another,type

rosrun image_transport republish compressed in:=/camera/color/image_raw raw out:=/camera/color/image_raw respawn="true"

• For non-rosbag users,just take advantage of following script export_tum,export_euroc and get_csv to get data in formats of Tum or EuRoC.

5.2 Evaluation

We use open-source tool evo for evalutation. To install evo,type

pip install evo --upgrade --no-binary evo

To evaluate monocular visual SLAM,type

evo_ape tum bridge1.txt your_result.txt -vaps

To evaluate LIDAR SLAM,type

evo_ape tum bridge1.txt your_result.txt -vap

To test GNSS based methods,type

evo_ape tum bridge1.txt your_result.txt -vp