If you want to explore the terrain of Mars, or the surface of a foreign moon or asteroid, you need to know what you're getting into. Destinations like these are bathed in lethal radiation, with no breathable atmosphere, likely in profound cold (or lead-melting heat). In short, it doesn't make a lot of sense to send human beings. At least not for a while.
The obvious solution is to commission robots and unmanned rovers to do our exploring and planetary science for us. But there's a difficulty here that has possibly never occurred to you: How do you move the robot around? And what happens if it gets stuck?
A lot of the robots you're likely to see in an industrial setting don't even need to go from one place to another. They're just stationary arms welding car doors or lifting boxes onto freight pallets, or Cartesian coordinate robots punching holes in sheet metal along an assembly line. Locomotion (the act of traveling from one place to another) can be a surprisingly difficult problem in robotics.
Especially if the nearest maintenance technician is 250 million miles away.
The two most obvious choices for locomotion are wheel-based designs and leg-based designs.
Wheels are simple and energy efficient, and that's probably why most rover designs to this date have been wheel-based vehicles. But wheels are not without problems. What if the rover needs to maneuver over a pile of rocks? Or what if it has to cross a crack in the ground that is wider than the diameter of the wheel? Or what if you simply wanted to guide it through a pit of loose sand? It's very possible that in any of these scenarios, a wheeled rover could get stuck. In fact, one of these scenarios has already happened.
The NASA rover Spirit was a huge success, far outliving its designed mission length, operating for more than six years on the surface of Mars. But in the spring of 2009, Spirit's wheels became stuck in a sand pit, and despite valiant effort, NASA engineers were unable to help the rover escape. Eventually, its solar panels became stuck out of ideal alignment with the sun. This led to electronic malfunctions, and eventually Spirit stopped communicating with controllers on Earth.
So wheels have problems.
OK, so what about legs? Biologically inspired legs might be more versatile when it comes to scrabbling over difficult terrain, but they have problems of their own. Legs are more mechanically complicated than wheels, which would seem to put them at greater risk for damage or malfunction (not to mention the fact that they're just more expensive and difficult to build). Also, a legged robot would probably require more energy to move around, and on Mars, where there is nigh a gas station or wall socket to be found, energy is a precious thing.
So if we're talking about exploring planetary surfaces, we might want to go with a robot that is a little bit less MechWarrior and a little bit more Marble Madness.
I want to introduce you to three tumblers, or whatever you want to call them - robots designed for full-body rolling locomotion. To clarify, I doubt any of these is likely to end up exploring Mars or any other far-off body in its current form. Instead, I think we should think about these robots as more general design-types that might inspire the rovers of the future.
Imagine a beach ball being rolling across the dunes, driven by nothing more than a stiff breeze. Now imagine this beach ball is full of instruments that take scientific readings. Now imagine it's on Mars.
Not a bad idea, right? As far as the general concept goes, the so-called "tumbleweed" rover design is nothing new. NASA has been investigating designs like this for more than a decade, and the general idea is traceable at least to the 1970s, if not earlier. A rover in the style of a tumbleweed could overcome several common rover problems at once. Because of its size, shape and mass, it can easily tumble over and around many obstacles. And because it's powered by the wind, it doesn't rely on solar panels (which can become stuck in sub-optimal positions or covered in dust) or exhaustible power sources. Because it is inflatable rather than rigid, it also seems less likely to become damaged during landing. And if it needs to settle down in one place for a while, it can partially deflate itself to prevent rolling.
One possibility with the tumbleweed-inspired rover is swarm exploration, which I've discussed previously in this blog post. The idea is pretty simple: Instead of covering a little bit of ground with one expensive, powerful, slow-moving rover, explore lots of ground with a team of many simpler, cheaper rovers that fan out over the landscape and each report their findings back to a central base vehicle. This approach doesn't just explore more territory, it also provides redundancy. If you have two dozen wind-blown tumbleweeds and one of them gets wedged in a shallow crevice, you still have 23 bots ready move on and keep exploring. If your one powerful rover gets stuck or incapacitated, the mission is pretty much a bust. The best you can do is convert it to a stationary observation platform.
But there are obvious drawbacks to a wind-propelled rover. For one, it doesn't go exactly where you want. You'd just need to drop a bunch of them and hope they go somewhere interesting. Now, you might be able to insert an internal steering mechanism to help influence the direction of a tumbleweed's roll, but this will add to the overall weight of the rover, leading to the next concern: In order to get maximum propulsion out of the natural winds, the ball needs to be as light as possible. This puts a limit on the kinds of instrumentation the ball can carry. Plus, powering electrical components is a concern, since batteries themselves tend to be heavy. In statements over the years, engineers interested in these designs have emphasized that miniaturization is key. The lighter the internal instruments and power sources, the more feasible this kind of rover becomes.
So what about a tumbling robot that incorporates its own method of propulsion?
Ever since I first saw it, I have liked the idea of the Jollbot, which is a portmanteau of this robot's two movement styles: jumping and rolling. You can see its movement below:
The Jollbot was created in 2008 by a PhD student named Rhodri Armour at the University of Bath. Like the tumbleweed concept, it is based on a lightweight, spherical frame. Unlike the tumbleweed, it propels itself by hopping action. A central electric motor stores up potential energy along the length of shaft spanning the diameter of the spherical cage. It then releases suddenly like a spring, causing the bot to hop into the air. It's easy to imagine how this kind of action would be useful not only for generating rolling momentum, but for extracting a robot that becomes stuck against a pile of rocks or in a shallow hole. A fleet of bots like this could perform the same kind of swarming exploration I described with reference to the tumbleweed, only with more deliberate control.
Oh, and if you watch the video and think, "Gee, it can't really jump that high," you shouldremember a couple of things: 1.) This was the first robot of its kind, and engineers may be able to design subsequent prototypes with a more powerful vertical. (In fact, it's possible someone already has and I'm just not aware of it.) 2.) Martian gravity is only about 38% as strong as Earth's. So Mars is a good place to jump.
I'd imagine that implementations of a robot like this depend on some of the same advances in miniaturization as the wind-powered tumbleweed. When saddled with a heavy array of instruments and batteries, its ability to jump and roll would be somewhat encumbered. So the lighter we can make its parts, the likelier a model like this is to see some real extraterrestrial action.
But why should robots like this be so perfectly round? You don't need to be exactly spherical to roll - you just need some creative engineering.
All of these tumblers are interesting, but the proposed robot known as the Super Ball Bot is my favorite of the bunch, and not just because one of the project leaders is a person named "Vytas SunSpiral." The Super Ball Bot has been under active research at NASA's Innovative Advanced Concepts (NIAC) program, and you can see Mr. SunSpiral discussing the idea in his own words in this video:
The structure is fascinating - like the others, it has neither legs nor wheels, but is designed for full-body rolling and tumbling. Instead of bearing a spherical skin or exterior cage, it is a roughly spherical tangle of cables and rods. Because of this odd composition, it is light, flexible and structurally robust all at the same time.
During transport through space, it can fold itself up into a flat, compact structure for easy storage. Once in the landing stage, a tensegrity robot really shines: You can drop a rover like this from a great height without causing damage. The bot simply bounces away from its touchdown site unharmed. And once it's on the ground, it can roll itself along nicely by shortening and lengthening the wires that hold the central rods together, much in the same way your muscles manipulate your skeleton with complex patterns of squeezing and relaxing. What's more, a tensegrity structure appears to be very adaptable to difficult terrain.
But while locomotion systems like wheels and legs are pretty familiar, no one would fault you for wondering, "How do you tell a jumble of rods and cables how to get from one place to another?" I particularly like the way SunSpiral and co-creator Adrian Agogino have approached the problem of programming movement. They have been learning how best to navigate rough and rocky landscapes by studying different movement control algorithms in an evolutionary simulator, forcing different control schemes to compete with one another for survival, leaving only the best plans for movement in the running.
To get a better visual sense of the project, check out this excellent video hosted at the IEEE Spectrum YouTube channel: