Unique Realistic & Challenging Datasets

Have a look at YouTube TRADR channel

The TRADR use cases involve response to a medium to large scale industrial accident by human rescuers team and robot team consisting of both UGVs and UAVs. For exercise and evaluation purpose, we simultaneously run the entire TRADR-system in sites such as shut down blast furnace, destroyed hospital complex, but also actual earthquake areas. The sorties are characterized by the following unique facts:

  • The operation environments are real industry and disaster sites restricted for civilians. Thus, they are highly realistic regarding environment structure. Shut down industrial site contains typical industrial structure with complex architectural and machinery structures indoors and outdoors. An earthquake site show a lot of complex damaged unstructured terrain and buildings.

  • Our sorties run under realistic challenging sensing conditions. For example, flexibly combining various sensors depending on the spontaneous need, the spatial relation among the sensors are often unknown. Also, the datasets exhibit partial data absence which can occur in real scenarios because of technical issues or environmental conditions such as visibility condition. We e.g. use fog machines for our exercises.

Our intention is to exploit all the various types of sensor data acquired during the exercise and evaluation sorties to provide realistic and challenging evaluation opportunities for USAR(urban search and rescue) robotics. In particular, we are interested in environment modeling and analysis applications using multiple types of sensors.

There are two different kinds of robots involved in a TRADR mission: Unmanned Ground Vehicles (UGVs) and Unmanned Aerial Vehicles (UAVs). The TRADR ground robots are flexibly equipped with different types of sensors like RGB-camera, laser scanner, stereo-camera, thermo-camera etc.


ROS Bag-Files

Tomas Svoboda has organized the current available datasets here.  The tables can be sorted by clicking onto the column headers. Bag-file names lead to more detailed pages with topic list, “Download” link, and overview video at the bottom of the page when suitable imagery is available.


Radio Signal Strength Data

The CRAWDAD dataset kth/rss  contains Radio Signal Strength data from the TRADR ground robot along with odometer in indoor and outdoor environments.


Team Communication Data

The robot sorties described above were carried out as part of human-robot team missions simulating response to medium-size disaster. The TRADR Team Communication Corpus contains audio recordings and transcriptions of the verbal communication among the human team members during these missions, e.g., mission commander, team leader, UGV.UAV operators, an infield rescuer.


Where TRADR datasets have been used?

V. Kubelka, M. Reinstein, T. Svoboda, "Tracked Robot Odometry for Obstacle Traversal in Sensory Deprived Environment". IEEE/ASME Transactions on Mechatronics, submitted, 2018TEval Rotterdam 2017
A. Gawel, R. Dube, H. Surmann, J. Nieto, R. Siegwart, C. Cadena. "3D Registration of Aerial and Ground Robots for Disaster Response: An Evaluation of Features, Descriptors, and Transformation Estimation". In Proceedings of the 2017 IEEE International Symposium on Safety, Security and Rescue Robotics (SSRR). Shanghai, China, October 2017.TEval Dortmund 2016 (Powerplant) - TRADR Review 2017
R. Dube, A. Gawel, H. Sommer, J. Nieto, R. Siegwart, C. Cadena. "An Online Multi-Robot SLAM System for 3D LiDARs". In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vancouver, Canada, Sep 2017.TEval Dortmund 2016 (Powerplant)
R. Dube, M. Gollub, H. Sommer, I. Gilitschenski, R. Siegwart, C. Cadena, J. Nieto. "Incremental Segment-Based Localization in 3D Point Clouds". In IEEE Robotics and Automation Letters (RA-L), 2018.TEval Dortmund 2016 (Powerplant)


License Information

These datasets are released under the Creative Commons license. It is free to share and adapt the datasets for non-commercial purposes in the way specified in the license terms (https://creativecommons.org/licenses/by-nc-sa/3.0/).