DARPA Challenge: Underground war robots
We've been vying for the inside scoop on DARPA's latest robotics challenge since it was announced in late 2017. Teams are now getting a taste of what lies ahead in early integration exercises, and one thing's certain: This challenge will take robots where they've never gone before.
The Subterranean (or SubT) Challenge promotes the development of next-gen war robots that can navigate and map underground tunnels and caves. The effort aligns with broader trends in robotics development toward systems that can be deployed in difficult-to-reach areas, including in infrastructure inspection, disaster relief, and mining.
I got a chance to lob questions at Dr. Tim Chung of DARPA's Tactical Technology Office and the program manager of the SubT. Turns out the defense industry won't be the only beneficiary of an army of mole bots.
GN: Why take robots underground? What needs gave birth to the SubT Challenge?
Dr. Chung: Like all interesting ideas, the genesis is a convergence of a number of needs and opportunities. The SubT Challenge is motivated primarily by warfighter needs, as well as those of the first responder communities. We are inspired by the need to conduct search and rescue missions, whether in response to an incident in a highly populated area, a natural disaster, or for mine rescue.
Subterranean environments present serious impediments, from autonomy and perception to networking and mobility, and any one of these areas are rich areas for ongoing research by leading technologists. The SubT Challenge is a mechanism to seek potential breakthroughs at the seams and through the holistic integration of the best innovations in all of those areas.
We are focusing on the subterranean domain in part because it is so challenging and can also lead to new technologies that allow us to navigate other complex environments such as underwater or urban areas.
GN: Tunnels pose unique challenges for robots. What are among the biggest? Any challenges that might surprise us?
Dr. Chung: Teams participating in our SubT Integration Exercise in April encountered some of these challenges firsthand. Degraded communication is one of the most significant problems to overcome because technologies we rely on above ground to relay information are no longer available. Coupled with low light, high dust, and unpredictable terrain, the scale, magnitude, and level of difficulty these environments pose require resilient systems that can adapt. That is, some technologies might work in one part of the course, but might be ineffective in other segments, requiring solutions that can re-attack the problem in different ways, even in the absence of inputs from the human supervisor.
The Tunnel Circuit is first, but we want competitors to think ahead to the Urban and Cave Circuits, as well. I often compare it to a triathlon. We are not looking just for the strongest swimmer, runner or cyclist. We want integrated solutions that can do all three.
GN: Do any robots work underground now?
Dr. Chung: Robots have been deployed underground, for instance, to aid in mine rescue efforts or to clear IEDs in conflict zones.
GN: Why is now the right time, technologically speaking, to spur development in SubT robots?
Dr. Chung: What is really exciting right now is that there are so many advancements being made in all of the component technologies. We see major advancements being made in the maturity and longevity of air and ground robotic platforms, unprecedented onboard computational power, multi-modal sensing approaches, team-level distributed autonomy, and communications technologies that are all starting to tackle the challenges of complex and large-scale underground environments. Our goal for the SubT Challenge is to spur the research and development that it will take to bring all of these technologies together into functional systems, and in fact, bring together a community of researchers, innovators, and engineers from all of the various disciplines that it's going to take to tackle all the challenges.
GN: Who will compete in the SubT? Any sense at the outset whether we'll see drastically diverging technology stacks from team to team? Anything you can say about those?
Dr. Chung: We have a variety of participants, from technology companies and university researchers to hobbyists. I would not want to bias potential solutions by commenting on specific technologies, but we have seen diverse solutions – ranging from novel platforms to interesting concepts of operation of robot teams -- which reflects the diversity of competitors.
Thus far, eleven Systems teams and five Virtual teams have qualified to compete in the Subterranean Challenge Tunnel Circuit. We are accepting proposals for self-funded teams for the Virtual track of the Tunnel Circuit through June 10.
We will continue to accept team registrations over the coming months from self-funded teams for both the Systems and Virtual tracks for the Urban Circuit (February 2020) and Cave Circuit (August 2020).
GN: How were the challenges designed and why were they designed the way they are?
Dr. Chung: Our main objective in designing the challenge is to guide the development of compelling solutions that can meet the real-world challenges that warfighters and first responders experience every day.
The most important design element is surprise. In real-world scenarios, you often do not know the layout of the environment, or what could be around the next corner, or if the air is breathable. We model that scenario of interest by withholding the sites until a couple months before each event, limiting access to the sites before each event, adding terrain features and obstacles, and even modifying the layout of the sites between runs. Teams really have no idea what they are going to encounter until their run begins and they send in their robots.
Teams score when their systems accurately locate, identify and report artifacts placed along the path, all inspired by real-life scenarios. Artifacts include thermal manikins, red backpacks, power drills, cell phones and fire extinguishers, and we will introduce additional artifacts in future events to keep things interesting and relevant.
GN: DARPA competitions spur development. What crossover, if any, do you expect we'll see in civilian robotics applications following the SubT?
Dr. Chung: The primary and most natural crossover is to first responders. In addition to working with DoD agencies and service partners to identify needs, organizations like the Fire Department of New York have provided valuable feedback about the technological challenges of responding to incidents in the urban underground.