In the second half of February 2020, ten robotic teams gathered at the nearly-built but unfinished nuclear power plant near Elma, Washington. The robots were sent into the underground maze of the huge concrete-steel in a quest to find and locate predetermined 20 artifacts. Researchers, graduate and undergraduate students of our department together with colleagues from the Departments of Cybernetics and Northern Robtoics Laboratory, University Laval won the competition among the self-funded teams as CTU-CRAS-NORLAB team and scored 3rd overall, surpassing three out of five teams which were directly funded by DARPA.
Subterranean Challenge, called SubT, sponsored by the Defense Advanced Research Projects Agency (DARPA), is testing cutting-edge technology in autonomous robotics. Motivated, among several others, by the tasks of first responders, a team of robots is required to go into a complex, hostile, and unknown environments and find and localize predetermined objects. Urban Circuit was the second in the row, following Tunnel Circuit in August 2019. The Cave Circuit and the Grand Challenge itself, all kinds include challenges that are yet to come.
Objects in Urban Circuit included humans (real size mannequins), mobile phones, backpacks, gas leakages, and ventilation vents. Objects were located on several floors and the robots had to descend stairs. Only very little human assistance is allowed which furthermore complicated with limited or no wireless connection. Mission time is limited to one hour and also the team preparation time is restricted to 30 minutes. All these should faithfully simulate a real mission deployment.
Our CTU-CRAS-NORLAB team performed very well at the challenge while being accompanied by distinguished world-top robotic teams including CMU, JPL-NASA, MIT, ETH, UCB, CalTech, KAIST, OSU. As the best self-funded team we walked away with the main prize of 500 000 USD. Aside the prize money, which will help in further development of the robotics at our university, we also highly value the sensory data gathered during the competition runs.
Tomáš Svoboda, co-lead of the CTU-CRAS-NORLAB team and the chair of Department of Cybernetics commented on the challenges we encountered during the competition: "The limited-time for making the robots ready and for the search and locate mission itself is an important feature of the competition. Plainly speaking, the robots have to work at the first launch. No time for tweaking, repeating actions, fixing bugs. The task is very complex and calls for flawless cooperation of many hardware and software components. Robustly balanced hardware and software components are on the main strength of our CTU-CRAS-NORLAB team. The competition requirements go intentionally beyond the current state of the art. DARPA SubT stimulates developments and opens new research questions in multi-robot cooperative exploration. It is also a great opportunity for undergraduate and graduate students. They will become a new generation of roboticists. Working with smart students has been a great experience for me."
Jan Faigl, Head of the Computational Robotics Laboratory, described our strategy and the deployed technology: "A large, multi-floor environment was one of the main challenges in the Urban Circuit. We succeeded in deploying all kinds of our robots; tracked, wheeled, flying, and walking. The robots performed simultaneously, they computed 3D maps, run object detection algorithms, explored unknown scenes, planned paths, executed them and also composed a necessary mesh-like communication infrastructure. We are glad we successfully verified our multi-channel, multi-purpose communication solution that proved, together with several levels of autonomy driving modes as very beneficial."
The Challenge is also and perhaps foremost a unique experience for undergraduate and graduate students. They see a real deployment and importance of many methods and algorithms they were taught at the courses. Jiří Kubík, master student of the Open Informatics program says: "It was a unique opportunity how to experience robot deployment in a real mission. I appreciate the team spirit and the challenge of working under pressure. I deepened my knowledge of the Robot Operating System (ROS) which I had known from the course Artificial Intelligence in Robotics." Tomáš Rouček, master student of Cybernetics and Robotics program adds: "I never dreamed about having a chance of participating at such an amazing event as the DARPA Robotics Grand Challenge. My personal biggest challenge was the need for reacting to new requirements, coding solutions and integrating them into a complex multi-robot system. It helped me to master agile development methods."
See all the photos shot by our team members.