The last time NASA scientists sent a robot into the crater of a volcano was 1994.
It’s name was Dante II, an autonomous, eight-legged crawler packed with video cameras, lasers and other sensors. It was designed by scientists from Carnegie Mellon University’s Robotics Institute to rappel and hobble down the inside of the active Alaskan volcano Mount Spurr – a proof-of-concept for encounters with the types of hostile environments that NASA robots might deal with in space.
But a tumble towards the end of Dante’s mission and subsequent helicopter rescue offered a stark reminder that “the possibility of catastrophic failure is very real in severe terrain,” the robot’s designers wrote. Even with today’s technology – we have self-driving cars now! – there hasn’t been another Dante since.
“To get a robot to go over the varied and often difficult terrain is very challenging. Robotics has come a long way since Dante, but […] it’s just not quite at the level where they can handle volcanic terrain yet,” explained Carolyn Parcheta, a volcanologist and NASA postdoctoral fellow sponsored by Tennessee’s Oak Ridge Associated Universities. It’s part of the reason that the U.S. Geological Survey still believes that "experienced volcanologists are a better and more cost-effective alternative for monitoring dangerous volcanoes” than robots – at least, for now.
In a volcanic environment, there are myriad materials of different sizes and shapes. You’ll find small round rocks where each step is like walking on the shifting sands of a beach. On the more extreme end of the spectrum is lava that’s sharp and jagged, making it near impossible to find space both flat and wide enough for a human foot. You’re always walking at an angle. In the middle, you have what Parcheta describes as “the slow, oozing, ropy looking stuff” that’s still difficult to walk on, but less so than the jagged stuff.
“Volcanic terrain is much more complicated than just a set of stairs or an inclined slope, because it’s often all those different things combined,” Parcheta explains. “There’s no regular pattern to the landscape. It feels random. And to the robot it will be random. It needs to learn how to assess that before it can take its steps, and humans do this on the fly, naturally.” This is, as you might expect, difficult – and one of the big problems that Dante’s designers had. So, for years, humans have instead sufficed.
But there’s also another reason that volcano crawling robots haven’t exactly been subject to pressing demand. According to Dr. Peter Cervelli, associate director for science and technology at the USGS Volcano Science Center, his agency has had “limited need for ground based robotics” – in large part because the majority of volcanoes in the United States don’t presently pose a threat to human volcanologists.