When teams participating in the DARPA Robotics Challenge (DRC) were announced last year, almost all of them provided reasonably detailed renderings that gave us a good idea of the robots that they were working on.
The notable exception was NASA's Johnson Space Center (JSC), which only released a piece of concept art that appeared to show a Robonaut-like humanoid, but didn't give much detail. And since then, NASA JSC has been extraordinarily secretive about what they've been working on. Naturally, we got a little bit curious, and back in October, IEEE Spectrum went to Houston for a preview of NASA JSC's DRC robot, Valkyrie.
Today, NASA is ready to share it with the world.
When we visited the JSC team to meet their DRC robot, we had no idea what to expect. Or rather, we were expecting to see something like Robonaut's torso on top of a pair of beefy legs. After having our passports scrutinized in exchange for JSC visitor badges, we were escorted to an enormous, windowless building that could have easily housed a modestly-sized spacecraft, and probably did at some point in the past. Inside, after negotiating a maze of dimly lit, narrow passages and a series of locked doors, we emerged into a room the approximate size of an IMAX theater, where NASA JSC team leader Nicolaus Radford introduced us to this:
Valkyrie (officially designated "R5" by NASA) is a 1.9 meter tall, 125 kilogram, 44 degree of freedom, battery-powered humanoid robot. A team from NASA's JSC in Houston, in partnership with the University of Texas and Texas A&M and with funding from the state of Texas itself, built the robot for the DRC, which will hold a preliminary competition later this month. JSC is a Track A team in the DRC; along with five other Track A teams with their own robots, JSC will be competing against Track B and C teams, each one of which will have an ATLAS robot from Boston Dynamics. In addition, Track D teams (which have no DARPA funding) will be entering their own robots.
The challenge created by DARPA involves tasks like walking over uneven terrain, climbing a ladder, using tools, and driving. This means that Valkyrie has to be capable of operating in the same spaces that a person would operate in, under the control of humans who have only minimal training with robots, which is why the robot's design is based on a human form. The overall goal of the DRC is to help drive innovation towards robots that are able to take over from humans directly, without needing any special accommodations. In that context, a human form makes sense because we're humans, and these robots will be doing the jobs that we don't want to be doing because they're too dangerous.
To that end, Valkyrie has seven degree of freedom arms with actuated wrists and hands, each with three fingers and a thumb. It has a head that can tilt and swivel, a waist that can rotate, and six degree of freedom legs complete with feet equipped with six-axis force-torque sensors. Unlike the ATLAS robots, Valkyrie is battery powered and operates without a tether. A removable battery in its backpack is good for about an hour of activity, and a human can swap in a fresh battery for a spent one in a matter of minutes. Also removable are Valkyrie's limbs: in just a few more minutes, a damaged arm can be swapped out for a new one, and the left arm can even be swapped with the right arm, since they're identical in construction. Things are bound to go wrong during the DRC, and the ease with which Valkyrie can be fixed is a potentially significant advantage.
While the ultimate goal for Valkyrie (and indeed all of the DRC robots) is to be as autonomous as possible, the trial in December will likely be relying on human teleoperation as well as assistive autonomy. Valkyrie is equipped with a staggering amount of sensors: cameras and LIDAR in the head, more cameras and sonar in the abdomen, and yet more cameras in the forearms, knees, and feet. All of these data won't be transmitted back to Valkyrie's operators all at once, but being able to take advantage of whatever sensor is most relevant to the robot's current task should help Valkyrie be fast and efficient, whether it's being teleoperated or functioning more autonomously.
NASA JSC has a huge amount of experience building robots. One of its flagship robotics programs is Robonaut, a sophisticated humanoid upper torso that's currently being teleoperated on the International Space Station. "When we were thinking about the DARPA Robotics Challenge, we were really leveraging a lot of the knowledge and technology that has been developed throughout the years with Robonaut," Radford, the team leader, explains. However, there are plenty of reasons why a robot designed for space isn't an ideal template for a robot designed for a disaster scenario on Earth, and Valkyrie is much, much more than a newer version of Robonaut with a pair of legs.
NASA JSC team leader Nicolaus Radford with Valkyrie.
Robonaut was primarily designed to assist astronauts in space, up to and including EVAs outside of the ISS. It's modeled after a large astronaut form, which offers a lot of advantages in microgravity, but isn't ideal for getting around on Earth. And in space, walking isn't an issue, and you also don't have to worry about falling down, something that Valkyrie is almost certainly going to have to contend with.
Besides these differences in hardware requirements, the use case for Valkyrie demands more autonomy and less telepresence. Robonaut does explore supervised autonomy a little bit, but Valkyrie will need to be able to complete tasks under much more severe latency and bandwidth restrictions, which means that it has to be much smarter, with a huge amount of embedded intelligence.
So, even though Valkyrie is based on a lot of what NASA learned from the Robonaut program, Valkyrie should really be thought of as a completely new robot, packed with completely new technology, and sporting a completely new, and very distinctive, look.
It's a She?
While NASA's official position is that Valkyrie is a genderless humanoid (as is Robonaut), the robot does have some features that we would call unmistakably female. For example, there's the name of the robot. "Valkyrie" (the roboticists call it "Val" for short) refers to the female figures in Norse mythology who decided which warriors fought valiantly enough in battle to be taken to Valhalla when they died. A valkyrie is either completely supernatural, or a human with supernatural powers, either of which are appropriate for JSC's superhuman robot. And then, of course, there's the extended chest area, which looks the way it does (at least in part) because it houses the linear actuators that allow the robot to rotate at the waist, while also providing some protection if the robot falls forward.
If it turns out that Val was deliberately designed from the start with female characteristics, that's a very bold (and very laudable, in our opinion) choice that the JSC team has made. Most robots, especially humanoid robots, are generally viewed as either genderless or male, and this is especially true when it comes to robots designed for utility, as the DRC robots are. In December, Valkyrie will be competing against robots like ATLAS and THOR, which are unambiguously male names that happen to evoke ideas of strength and power. But why can’t robots with unambiguously female names evoke strength and power as well? Considering how many girls are interested in STEM in general and robotics specifically, it's about time that someone decided to say, "hey, look, robots can be big and powerful and awesome and female, all at the same time."
Whether or not we're looking at a female robot from NASA, Valkyrie certainly has an aesthetic that's been carefully thought out by its designers. Most of the other DRC robots are starkly utilitarian: ATLAS, for example, is protected by cages of steel tubing. Valkyrie is completely different, in that it has "clothing" of a sort, made up of custom designed and fitted panels of fabric-wrapped foam armor.
"We take our soft goods very seriously," says Radford. "Our robot is soft. If you brush against it while you're working, you don't want to feel this cold, hard metal. You want it to feel natural, like you're working next to another human being. The soft goods, the clothes we put on the robot, give it that feel, that appearance of being more comfortable to be near." The clothing also performs the essential function of protecting Valkyrie from falls and impacts.
Part of the reason that JSC is one of the few groups putting functional clothes on robots is that it's a lot more difficult than it might seem. It's a continuous (and sometimes contentious) compromise between design and engineering, making sure that the soft goods are functional (and good looking) while still allowing for ready access to the underlying mechanics. The only way to make this work was for JSC to develop their own soft goods lab, where they design and construct all of Valkyrie's clothing on-site. There is a dedicated design staff, and outside experts are involved as well: Valkyrie will be sporting footwear custom made by DC Shoes.
Building a Team
The development environment where the JSC team works, called "The Bunker," is an enormous room in a building that has been used for spacecraft design and testing since the Apollo era. It's a completely open space, filled with software engineers, mechanical engineers, and electrical engineers all working side-by-side. In addition to a core team of people at JSC who have been working on Robonaut (and other robotics projects) for upwards of a decade, NASA hired about 20 new people (some of them straight out of college) specifically to work on Valkyrie. That brings the total size of the team to 55 people, and according to Radford, "it's been an amazing experience watching everyone grow and work together."
Radford intentionally structured the JSC DRC team on the Skunk Works model that famed aeronautical engineer Kelley Johnson so successfully implemented at Lockheed in the 1940s: a diverse group of people all working together under one roof in very tight secrecy. The team works on two shifts, 22 hours per day, turning the lights on at 7am and keeping everything running until 5am the next morning. It's an aggressive schedule, to put it mildly, but that's what's required to start from scratch and end up with a functional robot in less than a year.
Even with all of the Robonaut experience that JSC has, building Valkyrie has been a demanding process on a nearly impossible deadline. JSC managed to design and build Valkyrie in only nine months, which is, as Radford puts it, "insane." When IEEE Spectrum visited JSC in October, we arrived just a few days after the end of a 16-day U.S government shutdown, which meant a two week forced vacation for the entire Valkyrie team. "It probably came at the absolute worst possible time it could have come," Radford says. "To say that the schedule is thin is to put it lightly: this project has been crazy. So, you take two week chunk out of that, and that's a significant amount of work time. But we took it professionally, and when the government came back on, we hit the ground running. We're not going to let a two week slip get in the way of the amazing amount of work that our team has done over the last 12 months."
Preparing for the Future
The DARPA Robotics Challenge Trials take place in Florida just a few weeks from now, and Valkyrie will be there competing against 16 other teams. Radford is optimistic about how it's going to go, and IEEE Spectrum will be there too, following up with the JSC team throughout the event.
There's a lot more to Valkyrie than just the DRC, however. In the short term, as Radford explains: "in the same way that Valkyrie has benefited from the 15 years of robotics development in the Robonaut program and NASA's exoskeleton, we'll continue the development cycle, with technologies that we've developed for Valkyrie feeding back into Robonaut, making our space-faring robots even better than they are now."
Valkyrie is, of course, not a space-faring robot, which is why NASA JSC's development promise is even more exciting, as Radford tells us:
NASA saw considerable overlap in what the DRC is trying to accomplish, in NASA's goals as an agency, and in our vision for space exploration. NASA wants to get to Mars, and in order to get to Mars, NASA will likely send robots ahead of the human explorers. These robots will start preparing the way for the humans, and when the humans arrive, the robots and the humans will work together. Technologies like Valkyrie, and the other advancements that the DRC is making, are feeding directly into programs that NASA sees as critical for long-term human space exploration.
As revolutionary as the technology inside Valkyrie is, and as important as the robot is the future of space exploration, the team's focus on design as well as engineering has made Valkyrie into something more. "If it's worth doing, it's worth doing cool," says Radford. "[Valkyrie] has a little bit of a superhero feel to it, because honestly, that's what DARPA's requesting: they want a superhero robot." Team NASA JSC has definitely worked hard toward that goal, and later this month, we'll have a chance to see just what kind of superhero Valkyrie will be.
[ Valkyrie ]
Special thanks to Nicolaus Radford, Daniel Huot, and the NASA JSC DRC team for giving us the chance to visit.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.