Description Contributions Citation Recording platform Challenges Downloads How to play Baselines Related work Privacy License Funding Acknowledgment

😉 2023-06-01: All download links have been fixed!

😱 2023-05-10: Our data hosting server is experiencing some hardware issues and it will take some time to fix it. Sorry for the inconvenience caused.

😍 2022-06-29: Data and annotations for some challenging scenarios are available!

😎 2021-09-29: We have a new all-in-one download mirror hosted by the Chronorobotics Lab @ CTU.

😉 2021-01-04: Everything is back to normal.

😱 2020-12-19: The data download links are broken due to some unknown cloud server issues, we will check it as soon as possible after our university reopens, and sorry for any inconvenience caused.

😷 2020-03-14: Due to the current context, our university is closed from March 14 2020. As a result, we are affected by this measure and cannot, temporarily, no longer update our dataset. We apologize for any inconvenience caused.

EU Long-term Dataset with Multiple Sensors for Autonomous Driving

--- Collected by our very own UTBM robocar ---

Description

This dataset was collected with our robocar (in human driving mode of course), equipped with eleven heterogeneous sensors, in the downtown (for long-term data) and suburban (for roundabout data) areas of Montbéliard in France. The vehicle speed was limited to 50 km/h following the French traffic rules. For the long-term data, the driving distance is about 5.0 km (containing a small and a big road loop for loop-closure purpose) and the length of recorded data is about 16 minutes for each collection round. For the roundabout data, the driving distance is about 4.2 km (containing 10 roundabouts with various sizes) and the length of recorded data is about 12 minutes for each collection round. In addition to enjoying the typical scenery of eastern France, users can also feel the daily and seasonal changes in the city. For a quick overview, please refer to the following videos.

Contributions

This dataset provides:

Robot Operating System (ROS) rosbag files that record environmental data from two Velodyne HDL-32E lidars, an ibeo LUX 4L lidar, a SICK LMS100-10000 laser rangefinder, a Continental ARS 308 radar, two stereo cameras (a Bumblebee XB3 and a Bumblebee2), two Pixelink PL-B742F cameras with Fujinon FE185C086HA-1 fisheye lens, a Magellan ProFlex 500 GNSS receiver with an RTK base station, and an Xsens MTi-28A53G25 IMU;
Data covering day/dusk/night/week/season, ans as it captures daily and seasonal changes, this dataset is especially suitable for long-term autonomy research;
Intensive data for roundabout challenge, a very common road condition in France as well as in other European countries, which is not easy to handle even for humans;
Baseline methods especially for the lidar odometry, with ground-truth trajectory recorded by GPS-RTK.

Citation

If you publish work based on, or using, this dataset, we would appreciate citations to the following:

@inproceedings{eu_longterm_dataset,
   author = {Zhi Yan and Li Sun and Tomas Krajnik and Yassine Ruichek},
   title = {{EU} Long-term Dataset with Multiple Sensors for Autonomous Driving},
   booktitle = {Proceedings of the 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
   year = {2020}
}

Recording platform

Our multi-sensor platform design mainly follows two principles: 1) strengthen the visual scope as much as possible, and 2) maximize the overlapping area perceived by multiple sensors. In particular:

Two stereo cameras, a front-facing Bumblebee XB3 and a back-facing Bumblebee2, are mounted on the front and rear of the roof, respectively.
Two Velodyne HDL-32E lidars are mounted on the front portion of the vehicle roof, side by side.
Two Pixelink PL-B742F industrial cameras with Fujinon FE185C086HA-1 fisheye lens are installed in the middle of the roof, facing the lateral sides of the vehicle.
An ibeo LUX 4L lidar is embedded into the front bumper close to the y-axis of the car.
A Continental ARS 308 radar is mounted in a position close to the ibeo LUX lidar.
A SICK LMS100-10000 laser rangefinder (i.e. 2D lidar) facing the road is mounted on one side of the front bumper.
A Magellan ProFlex 500 GNSS receiver is placed in the car with two antennas on the roof.
An Xsens MTi-28A53G25 IMU is also placed inside the vehicle.

For more details, please refer to our paper.

Challenges

Many new research challenges have been introduced in this dataset, such as:


sloping road	shared zone	construction bypass	roundabout	night

snow	right overtaking*	crossing	pigeon	police

^*Aggressive driving / rule breaking behavior

Downloads

Please note that all rosbags are compressed, please decompress them as needed.
Camera calibration (done with camera_calibration) files are available here.

Long-term data:

Date	Local Time (Paris)	Sensors	Image Data	Non-image Data
2018-05-02 (Wed, evening)	20:40-20:54 (14'30")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3	rosbag	rosbag^/
2018-05-02 (Wed, night)	21:28-21:42 (13'55")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3	rosbag	rosbag^/
2018-07-13 (Fri, sunny)	14:16-14:33 (16'59")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3 / Bumblebee2	rosbag	rosbag
2018-07-16 (Mon, sunny)	16:10-16:26 (15'59")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3 / Bumblebee2	rosbag	rosbag
2018-07-17 (Tue, sunny)	15:40-15:56 (15'59")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3 / Bumblebee2	rosbag	rosbag
2018-07-18 (Wed, sunny)	15:04-15:21 (16'39")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag⁺	rosbag
2018-07-19 (Thu, sunny)	16:15-16:31 (15'26")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag^-	rosbag
2018-07-20 (Fri, cloudy)	14:35-14:51 (16'45")	2 × Velodyne / ibeo / SICK / IMU / GPS-RTK / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag	rosbag
2019-01-10 (Fri, snow)	09:06-09:17 (10'59")	1 × Velodyne / ibeo / SICK / IMU / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag^*	rosbag
2019-01-31 (Fri, snow)	08:54-09:10 (15'59")	1 × Velodyne / ibeo / SICK / IMU / GPS / Bumblebee XB3 / fisheye	rosbag^*	rosbag
2019-04-18 (Thu, sunny)	11:07-11:22 (14'55")	1 × Velodyne / ibeo / SICK / IMU / GPS-RTK / radar / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag	rosbag
⁺Only partial GPS-RTK data.
^-Best data quality, recommended for evaluation.
^*Only part of the itinerary recorded due to adverse weather conditions.
^/You might want to add_header_time_offset.py

Roundabout data:

Date	Local Time (Paris)	Sensors	Image data	Non-image data
2019-04-12 (Fri, cloudy)	18:14-18:26 (12'10")	1 × Velodyne / ibeo / SICK / IMU / GPS-RTK / radar / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag	rosbag
2019-04-18 (Thu, sunny)	12:03-12:15 (11'59")	1 × Velodyne / ibeo / SICK / IMU / GPS-RTK / radar / Bumblebee XB3 / Bumblebee2 / fisheye	rosbag	rosbag

How to play

roslaunch utbm_dataset_play.launch bag:=path_to_your_rosbag

By using the provided launch file, you will have:

Point cloud conversions for the Velodyne lidars, and without any spurious data (caused by blockage or reflections from the opposing sensors).
The output of the object tracking functionality of the ibeo LUX lidar.
A complete and aligned tf tree.

Baselines

https://github.com/epan-utbm/utbm_robocar_dataset

Related work

Yingqiu Chen. Data annotation for the EU long-term dataset. UTBM Student Development Project, June 2022. [PDF]
Tao Yang, Xiaofei Chang, Hang Su, Nathan Crombez, Yassine Ruichek, Tomas Krajnik, and Zhi Yan. Raindrop removal with light field image using image inpainting. IEEE Access, March 2020. [PDF | Dataset]
Hongkun Zheng. Image data anonymization. CIAD Internship Report, February 2020. [PDF | Code]
Li Sun, Zhi Yan, Anestis Zaganidis, Cheng Zhao, and Tom Duckett. Recurrent-OctoMap: Learning state-based map refinement for long-term semantic mapping with 3-D-lidar data. IEEE Robotics and Automation Letters, July 2018. [BibTeX | PDF | Video]

Privacy

We take privacy very seriously and handle personal data in line with the General Data Protection Regulation (GDPR) (EU) 2016/679. To this end, we used deep learning-based methods to post-process the images in order to blur face and license plate information. However, if you still find yourself or your personal belongings in the data, please contact us and we will immediately remove the corresponding information from the dataset.

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Copyright (c) 2018-2023 Zhi Yan, Li Sun, Tomas Krajnik, and Yassine Ruichek.

Funding

This work was supported by the Quality Research Bonus (BQR) of the University of Technology of Belfort-Montbéliard (UTBM), the Contrat de Plan État-Région (CPER) 2015-2020 (Mobilitech), the CZ MSMT project (No. FR-8J18FR018) / PHC Barrande project (No. 40682ZH) (3L4AV), the OP VVV funded project CZ.02.1.01/0.0/0.0/16\_019/0000765 (Research Center for Informatics), and the NVIDIA GPU grant program.

Acknowledgment

The authors would like to thank Abdeljalil Abbas-Turki, Olivier Lamotte, Jocelyn Buisson, and Fahad Lateef for their help in building the dataset, the Lincoln Centre for Autonomous Systems (L-CAS) for previously hosting the dataset, and the four reviewers of ICRA 2020 and the three reviewers of IROS 2020 in helping improve the manuscript.