CMU team takes on Subterranean DARPA Challenge

Credit: Courtesy of Carnegie Mellon University Credit: Courtesy of Carnegie Mellon University

If you were in Pittsburgh last year, you might remember seeing the self-driving Ubers cruising around the more gentrified areas of the city. That project seems to have been sidelined after a few of Uber’s autonomous vehicles self-drove into pedestrians, but they wouldn’t have even existed in the first place without the 2007 Urban Challenge. The Urban Challenge, the second such project from the Department of Defense’s (DoD) research arm, the Defense Advanced Research Projects Agency (DARPA), challenged teams to create autonomous vehicles that could operate in an urban environment. The Carnegie Mellon Tartan Racing team won the $2 million dollar grand prize in the challenge, which helped lay the groundwork for the recent self-driving explosion, from Tesla to Google to the aforementioned Uber. For the newest DARPA challenge, the Carnegie Mellon team has a new environment to navigate: underground.

The DARPA Subterranean Challenge will test teams on their ability to navigate three underground spaces: tunnels, caves, and the “urban underground.” According to the challenge’s website, the goal is to “drive novel approaches and technologies to allow warfighters and first responders to rapidly map, navigate, and search dynamic underground environments.”
The Carnegie Mellon team is led by Sebastian Scherer and Matt Travers, senior systems scientist and systems scientist, respectively, at the Robotics Institute. The team is composed of a core group of 15-20, including members from Oregon State University, and also includes classes focused on solving key challenges, as well as many undergraduate and graduate students. Scherer, who has worked on several Carnegie Mellon DARPA Challenge teams, says that the Subterranean Challenge “is really about sending back a map and a location of where different objects are [without] using biological systems.”

The team’s approach will be to “have a [modular] perception system which tells us...where we are going, [while] at the center it creates a map and classifies [objects],” using robots of varying mobilities, from crawling to flying. Scherer says that a positive element of the DARPA challenges is that they “really kind of force you to put [existing technologies] together,” which can result in interesting and innovative new approaches.

The Carnegie Mellon team is one of six receiving up to $4.5 million to “develop the robotic platforms, sensors, and software necessary to accomplish unprecedented underground missions,” according to a Carnegie Mellon press release, making this project one of the many at Carnegie Mellon receiving funding from the military. When asked about the possible applications of the research, Scherer said that “the challenge is really general,” with possible applications in “ and rescue after [things like] earthquakes,” and that the military’s strategy is “to get ahead in they invest a lot in science and technology.”

The technologies resulting from the previous challenges have already been used to develop autonomous vehicles for the military, and it stands to reason that the novel, interesting solutions to these problems will as well. Bob Hastie, a senior technical adviser at the Defense Threat Reduction Agency, was quoted by C4ISRNET as saying that the goal of these projects is to “take the human out of [dangerous jobs] and use autonomous systems and robotic systems to...protect our warfighters.”

Whether the new technologies resulting from the Subterranean Challenge will be used for war is unclear, but probable: you don’t invest in something without expecting a return. The fact that the DoD, rather than the National Science Foundation, funds these projects means that the DARPA challenges, whatever their possible commercial or humanitarian benefits, are geared towards solving military problems.