Last August, we posted about a clever concept for an asteroid exploration rover from Stanford, JPL, and MIT that uses internal reaction wheels to flip its spiny cubalicious shape around without needing legs, wheels, rocket engines, force fields, tractor beams, or anything else. As of our 2014 article, NASA had funded this thing to Technological Readiness Level 3.5, which is somewhere in between proof-of-concept and laboratory validation, which left us optimistic that something might come of it.
A few months ago, we got a chance to check out the latest prototype of this robot, and we’re excited to say that it’s made it all the way to a fully armed and operational prototype: Hedgehog, as it’s called, has its core mobility hardware fully integrated and has been undergoing microgravity testing on parabolic flights. We spoke with Rob Reid from JPL and Ben Hockman and Marco Pavone from Stanford about what they’ve been up to over the last year, and then we definitely didn’t sneak* the robot into Smithsonian’s National Air and Space Museum in Washington, D.C., for a little photoshoot.
The reason that Hedgehog exists is to explore planetary bodies where there’s very little gravity, like asteroids, comets, and small moons. For example, on Phobos (one of Mars’ moons), you’d weigh about as much as a tennis ball, and a world-class sprinter could run fast enough to launch themselves off of the moon entirely. An even better example is Itokawa, the asteroid visited by Japan’s Hayabusa spacecraft. There, you’d weigh as much as a paperclip, you could reach escape velocity with a violent sneeze, and it would take you well over 2 minutes to free fall to the surface from a height of one meter. “It’s not like regular gravity,” says Reid. “It’s more like two objects in space, floating side by side. There’s very little attraction pulling these objects together.”
Exploring a place like this is incredibly difficult, because the microgravity is seriously micro, and trying to use something like a wheel would most likely either slip ineffectively or just flip whatever robot it was attached to end over end and your mission would be over. Legs aren’t much better, because they need gravity to help them anchor themselves to be effective. Even robots that are specifically designed for microgravity exploration, like the Philae lander on comet 67P, have all kinds of issues even when mobility is the exact opposite of what they’re designed to be doing.
Figuring out the best way to do this exploration hasn’t been easy, as Pavone explains:
“Over the past 10 years or so, there’s been increasing interest in exploring small bodies—asteroids, comets, anything that’s smaller than a planet. We’ve spent quite a bit of time with scientists at JPL trying to understand what you need for that. The figures that came back from the scientists were that in general, you want to be controllable within 20 percent: if you want to move to a spot 10 meters away, you should be able to get there within 2 meters. Small bodies have surfaces that are heterogeneous, but within small patches, they’re homogenous, so you don’t need exact precision, although you need some sort of control. With that in mind, we designed this robot.
Sometimes when you work in robotics, you try to come up with a design that is cool. Here, this design is completely science driven. We tried to design something as simple as possible, given the requirements. But just because it’s a simple robot doesn’t mean it’s an easy robot. Everything about it is complicated. But in our opinion, it’s the least complicated way to do the job.”
Hedgehog is unique because of its form factor and method of mobility. First, the form factor: being a cube (a fetchingly spiky cube), it’s completely symmetrical, and doesn’t care a jot about which of its faces are up. This means that if it’s bouncing around an asteroid, it doesn’t need to worry about landing a certain way: it can just bounce and roll until it comes to a stop and it’s fine. And this is where the mobility comes in: Hedgehog is completely sealed (one of the key reasons why NASA seems to like the design, according to the researchers), and inside, there are three reaction wheels mounted orthogonally, one along each axis. The wheels are electrically driven, and can be abruptly stopped with a band brake (a friction belt that tightens over the surface of the wheel). When you spin one of these wheels up and then stop it abruptly, it imparts torque corresponding to its momentum to whatever it’s attached to.
In this case, it forces the body of the robot to rotate around the same axis as the wheel but in the opposite direction, flipping itself with an aggressiveness in proportion to how quickly the wheel was spinning before it was stopped. With just a little momentum, the robot can gently twist or change the face that it’s resting on, and with a lot of momentum, it can violently hurl itself all over the place. This isn’t a terribly efficient way to move around in anything but microgravity, but it takes so little energy that the efficiency doesn’t even matter that much, according to Hockman. “If we’re going to a body where the gravity is very low, like a smaller asteroid, we actually don’t really care about the efficiency of the hop, because it takes almost nothing to get yourself off the ground.” You definitely wouldn’t want to use this method of mobility for an Earthbound rover, but it does work:
Beach tumbling is fun, but microgravity is where it’s at for Hedgehog, and it can do some amazing things when you crank down the gravity a whole bunch. Note that in very low gravity, spinning the reaction wheels up also imparts torque on the robot, but if you do it slowly enough, you can make sure that it’s not enough torque to overcome the inertia and friction of the robot while stationary. To test all of this stuff, earlier this year the robot flew on nearly 200 parabolas in a NASA aircraft that, while accelerating terrifyingly towards the ground, can provide about 20 seconds of near zero-g. Of particular interest here is how Hedgehog performs on sand and other granular media, which acts much differently in microgravity. Watch:
A few things to note about this. First, you can hear the spinning and braking of the flywheel(s), which is pretty cool. Also, you can see how important the spikes are for both traction and directional control. The version with the black spikes is from Stanford, while JPL’s version is white and slightly spikier. Even in microgravity, the robot possesses a very fine level of control, and it’s easily scalable to whatever the gravity happens to be, making this one design very efficient. That last “tornado” escape maneuver is very cool, and could be used to escape from craters or sinkholes that the robot might accidentally tumble into. Or maybe to dig a bit of a hole, if you want to see what’s under the surface. And that poor little yellow rubber ducky that got so abused? Blame ICRA 2016.
It’s hard to get a sense of what kind of long distance mobility Hedgehog has in microgravity, and until we send one to the moon (or farther), the best we can do is watch what happens when a simulated Hedgehog is unleashed on a simulated asteroid Itokawa:
Wheeeeeeeee! The simulation is sped up 100x, but still, wheeeeeeeee! There are plans to teach Hedgehog to do its own SLAM and high-level motion planning, which will be necessary if it’s ever to start exploring on its own.
If we start thinking about what might be involved in sending Hedgehog somewhere extraterrestrial, the pieces fall into place pretty easily. The robot can be scaled up or down, from CubeSat all the way up to (we assume) Borg. The sides of the cube are covered with solar panels to generate power (or you could just run the thing off of batteries and assume it’s disposable), and they also offer access to a variety of instruments for surface analysis, like a spectrometer, a microscope, and cameras.
A Hedgehog, or multiple Hedgehogs, would be deployed from a mothership that would take care of communications back to Earth and potentially help out with localization. The mothership wouldn’t have to be that big, or that expensive, and if you could instead send Hedgehogs piggyback on spacecraft that were heading to somewhere like Mars, the entire mission could be done on the (relatively) cheap, as an add-on to a flagship mission that’s heading there already. It’s hard to say how cheap, because we’re not quite at the point where numbers like that are being discussed, but as you can see from the videos, things look very promising, with the readiness level of the robot maturing rapidly.
“In terms of being able to have a mission in the short term, I think we understand enough about the platform and the mobility today that if we were to go out and collect a bunch of relatively high TRL cubesat components, we could stick them together and have a functioning, almost flight-ready piece of hardware,” Reid says. “There are a bunch of other questions, like what science instruments we’d want to put on it, but for having a robot that we could move around on the surface of a small body, we could do that in very short order.” And Pavone agrees: “The general strategy is to make this ready for flight, at least in its most basic configuration, fairly soon, so that whenever a flight opportunity arises, we can raise our hands and say, ‘look, we have an option here with a secondary payload to dramatically increase your science with a minimal cost.’ ”
Special thanks to Ben, Rob, and Marco for meeting with us.