One of the major disadvantages of flying robots is that they’re generally useless when they’re not in the air, and the amount of time they can spend in the air is severely limited by their capacity for fuel… Plus, more fuel means more weight means bigger engines means more fuel, and so forth. EPFL has developed a robot called Eyebot that, while based on the familiar quad-rotor helicopter design, includes an innovative extra feature: the ability to stick to ceilings with a magnet. The gives the Eyebot the capability to loiter indefinitely in advantageous positions while providing surveillance with a pan and tilt camera system. It can autonomously attach and detach itself from ceilings, and uses simple optical flow analysis to detect drift and infrared sensors to avoid obstacles. Apparently, the designers envision the Eyebot stationing itself on a ceiling with a bunch of its friends and FIRING FRIKKIN’ LASERS AT PEOPLE:

Okay, not really, but the Eyebot will include a laser pointer with which it can locate and illuminate predefined targets, either helping you find your suitcase or, potentially, doing other things.

You’ve maybe guessed by now that Eyebot is designed to work as a part of the same Swarmanoid distributed robotic system as Handbot. Swarmanoid is still way too cool to include as an afterthought on this post, so we’ll have more on Swarmanoid on Monday… +2 points if you can guess what the third bot in the swarm does.

And before you say anything, I know this was the third “Eye” robot in as many days, and I don’t have any more, I promise.

[ Eyebot ]

BEAR stands for Battlefield Extraction-Assist Robot. Developed by Vecna Robotics, BEAR is a humanoid torso on tank treads designed to bring mobile heavy lifting capabilities to military, search and rescue, and industrial applications. BEAR’s two arms can lift 500 pounds between them, and it’s capable of dynamically balancing on its unfoldable tracked “legs.”

BEAR can lift so much thanks to its GINORMOUS hydraulic backpack, which is conspicuously absent from the animated robots in the video below… I’m not sure how Vecna ultimately plans to stuff so much lifting power into such a small space, but if they can pull it off, they’ll have an extremely capable and versatile robot that would be able to easily adapt to become useful in a variety of different situations. I’m sure it won’t be cheap, but it’ll definitely be handy.

There’s certainly a lot of potential here, and it’s interesting to note how battlefield robots seem to be diverging from medium-sized PackBots and such to things that are larger and more powerful (like the Warrior) and things that are smaller (like the EyeDrive). Why the specialization? If I had to guess, I’d say that if you’re going to commit to the cost and associated infrastructure that even a medium-sized robot requires, you might as well get something big and mean. Otherwise, you just want something cheap that anyone can carry around and operate when necessary. BEAR is decidedly in the former category, and I bet it would be even better than the Warrior at tossing PackBots through windows.

[ Vecna Robotics BEAR ]

Thanks Andy!

This is Handbot, a robot designed by EPFL to climb vertical surfaces and manipulate objects. It’s not quite strong enough to climb using just its grippers and arm, so it’s got a magnetic rope launcher that it fires up to the ceiling to carry most of its weight. Handbot is designed to work as a part of a distributed robotic system called Swarmanoid (the successor to the swarm-bots project), which aims to emulate the functionality of a single humanoid robot with a swarm of more specialized robots that work together. That’s way too cool to include as an afterthought on this post, so we’ll have more on Swarmanoid tomorrow.

[ Handbot ]

We mentioned a little robotic eyeball in yesterday’s post about the ODF EyeDrive surveillance robot, and eagle-eyed readers who sat through the whole video might have noticed in the very last scene that the EyeDrive robot drops a little round ball. ODF also makes a passive surveillance robot called the EyeBall R1, which is a little round bot about the size of a baseball that’s designed to be rolled or thrown into areas where you need some recon.

The EyeBall is designed to bounce around a little bit and then self-right itself on a flat base. Once it does, it rotates every 15 seconds, sending back streaming video and audio. You can pan the EyeBall manually as well as zoom in, and it’s got IR LEDs for night vision. EyeBalls are cheap, at only about $5k for two balls and a control system, and you can even accessorize them little camo jackets or a disguise to make the EyeBall look like a rock. Cute!

As nifty as this little robot is, is does have some substantial drawbacks, as Danger Room reported last month. What tends to happen is that the robot gets thrown into, say, a room, and it ends up rolling underneath a piece of furniture and is rendered useless. Plus, people tend to notice something the size of a baseball come sailing in through a window, and as Danger Room’s source points out, you might as well “just throw a flash-bang, instead.”

Demand for this type of robot (something small, cheap, and tossable) is increasing, seemingly moreso than for larger, more capable robots like PackBots… At AUVSI last week, the Marine Corps announced that they’re in the market for a throwable surveillance robot for day and night reconnaissance. It’s easy to understand why: most of the time, all you really need an unmanned system for is getting that first look around an unknown and potentially dangerous area, and the ideal robot for that task is something simple, portable, and expendable that anyone can put into action quickly and reliably.

[ ODF EyeBall R1 ]

Image via Defense Update

This is gonna sound crazy, but just bear with me… Imagine a world in which millions of people stay at home, living out their lives by controlling artificial representations of themselves that let them be whoever they want to be and do things they could never do in real life.

Yeah, I know, totally crazy. But that’s the premise for Surrogates, a robot action flick starring Bruce Willis that’s due out September 25. And of course, something goes horribly wrong with the robots and Bruce has to save the day. IRL. Which is, I guess, scary in the future. And why shouldn’t it be? Robots today exist to take over tasks that are dull, dirty, and dangerous, and oftentimes, real life is all of those things and worse. Project that philosophy into the future, and you’ve got robots living our lives for us, while we control them from someplace comfortable and safe, like a bomb shelter filled with pillows.

Surrogates looks to be basically another Matrix, except with hardware instead of software and everybody knows that it’s going on. More realistic? Maybe. People are experimenting (successfully) with neural control of robots, but reading brains is the easy part: it’s writing data that’s hard. Even so, it’s not impossible… Direct brain stimulation has been used to allow blind people to see, for example.

I’m probably overthinking this movie, but the premise does seem interesting, and if nothing else, it appears to contain ample amounts of sex, violence, and robots… What else could you possibly want?

[ Surrogates ] VIA [ io9 ]

In a mere 50 virtual generations, swarm bots (remember them?) using genetic software evolved the capacity to lie to other robots about the location of a source of food. Initially, the robots were programmed as a group to search for an object that represented food, and they gradually learned to emit a blue light when they found the food to show other robots where it was. Researchers at EPFL in Switzerland evolved and mixed the programming of the most successful foragers until they had a bunch of robots who were very good at finding food, and then gave the virtual genes of each individual robot control over their blue light that signified food.

You might expect that the robots would learn not to signal when they found the food to reduce competition, which is passive deception, but they also evolved an actively deceptive behavior, where some robots would deliberately travel away from the food and signal their blue light, drawing other robots in the wrong direction. Crafty little buggers. Interestingly, this deceptive behavior didn’t make much of a difference to the overall fitness of the group strategy of following blue lights… Some robots always tell the truth with their blue lights, which means it’s always advantageous for a clueless robot to follow a blue light as opposed to just wandering randomly.

So why do some robots keep telling the truth if deception can effectively lure other robots away from the food? It’s fairly simple, as I understand it… If all of the robots are deceivers, any new robot will quickly learn that avoiding blue lights is the best way to find food. And in that case, any robot that starts signaling its blue light when it does find food (through a “genetic mutation” in its software) will increase its own fitness by repelling other robots from the food it finds. As it passes this behavior on to its virtual children, there will be more and more truthful robots until it once again becomes more advantageous to be deceptive.

There are, however, populations of truthful and deceptive robots such that the combination of behaviors reaches a stable point. In this particular experiment, the stable evolutionary endpoint (after 500 generations) was that 60% of the robots were deceivers and 10% told the truth. Furthermore, about a third of the robots were slightly attracted to blue lights, another third were strongly attracted, and the final third avoided them completely. This type of experiment, its progression, and the results are particularly fascinating to me because the robots are behaving and evolving in much the same way as populations of animals do. Examples of both altruism and tactical deception can be found in many different species of animals as well as (of course) in humans, but these little robots offer a unique opportunity to study (and tweak) the evolution of behavior in real time.

[ EPFL ] VIA [ Not Rocket Science ]

EyeDrive is a portable, tossable surveillance robot that distinguishes itself with cameras on all four sides that allow it to send back a streaming 360 degree video (and audio) panorama of its surroundings. It’s also got a fifth camera in the front that pans and zooms in the vertical plane… All of this stuffed into a robot about the size of a large hardcover book. It can carry its own weight in payload, drives upside-down, sees in the dark, and has a point ‘n click drive interface. Plus, it comes with a nifty custom vest, with a pocket for the robot in the back and all of its gear in the front:

In addition to the carrying vest, you get a protective throwing case that you can put the robot in before you chuck it through (say) a second floor window, removable treads, and any number of modular payloads including additional cameras, armaments, and little round robotic eyeball that also serves as a range extender for the robot (more on that later). The video we’ve got of the EyeDrive is a little bit long, but it’s worth the watch, ’cause there’s a weapons integration test (with an explosion, yay!) starting at around 4:00.

[ EyeDrive ]

There’s a big potential market for small, simple, and cheap man-portable UAVs that individual soldiers can rely on for on-demand reconnaissance. The Point and Toss UAV is a little robotic plane which, while certainly small (at a pound and a half with a three foot wingspan) and cheap (“very” cheap, I was told), may very well be able to claim unambiguous victory in the simple category. The operation of the Point and Toss is literally that simple: you point it in the direction you want it to go, click a hand controller to lock the heading and start the engine, and then toss. The UAV automatically stabilizes itself, flies out for about half a mile at 250 feet taking pictures and video to an SD card, and then autonomously turns around and lands at your feet. All done.

While I love the simplicity of the Point and Toss, there’s a major trade-off in flexibility. The UAV doesn’t have a loiter capability, so if you’re in an evolving environment, you’re going to have to keep chucking the thing out to keep up to date on what’s going on. You’re also restricted to one direction at a time. And although it’s handy to have the UAV store data on an SD card, it’s handier to have one that sends information back realtime.

I guess the point, though, is that the Point and Toss is easy enough for anyone to operate with practically no training, cheap enough for everyone to have access to one, and small enough to make that not unrealistic.

[ Point and Toss UAV ]

What makes this robotic hand, from the neurobotics lab at the University of Washington, special is that it’s designed to be an exact replica of a human hand. The shape of the bones, the attachment points of the tendons, even the musculotendon passive viscoelasticity (whatever the heck that is) are functionally identical. The hand is a testbed for investigating the potential for complex neural control; if your brain sends a signal to a robot hand, the hand better be able to interpret the signal and move just the way you’re expecting it too.

Now, I still don’t think that modeling robots on human anatomy is a very efficient way to make robots more effective, but it is a good way to figure out how humans (most humans) are inherently able to be as dexterous as we are. And that, in turn, could help us to design better robots. Not necessarily more human robots, but robots that make use of (or potentially improve upon) specific attributes of human anatomy.

[ UW Neurobotics ]
Thanks Alan!

This is ECCE, an anthropomimetic robot designed to test embodied cognition in compliant engineering. In other words (normal words), it’s a robot designed to emulate a human, both inside and (eventually) out, the idea being that this will make the robot more human-like. ECCE uses artificial bones, muscles, and tendons, all wired together in the same way that yours are, in the hope that the robot will be able to move less artificially.

“Standard humanoid robots mimic the human form, but the mechanisms used in such robots are very different from those in humans, and the characteristics of the robot reflect this. The concept of anthropomimetic robots, such as ECCE, will compensate these drawbacks by not just replicating the appearance of a human, but duplicating the inner structures and mechanisms of humans.”

I’m not exactly sure what drawbacks are being referred to… Yes, robots end up being robot-like, and most anthropomorphic robots tend to strike that uncanny valley chord, at least to some extent. But I don’t think the solution is to structure robots based on humans. I mean, the whole point (okay, one of the points) of robots is that they have the capability to do things that humans can’t do, or at least can’t do very well. They’re not humans, and the most effective robots are generally the most specialized and therefore least human. Arguably, the only reason to model a robot on a human at all is that robots have to operate in a world designed by and for humans, and those humans have some hangups when it comes to interacting with things that aren’t humans. And most of the time, it’s just not worth it. As the video rightly points out, using human muscle style motors is way, way more complicated than using a servo or two. Yeah, it’s (potentially) a more “natural” motion, but who cares? You can generally emulate the same thing with some tastefully programmed servos. I would guess that with some kind of skin on, you’d be hard-pressed to tell the difference between a robot using artificial bio-inspired muscles versus a well-programmed combination of servo motors.

That said, there’s a lot to be learned about motion, and humans are as good a place to start as any. My only worry is that if you imagine the endpoint of this project, you end up with an extremely complicated robot that isn’t significantly less robot-y than a bot with more traditional innards, while sacrificing the power, speed, and precision that we’ve grown to expect from robot limbs.

[ ECCEROBOT ] VIA [ IEEE Spectrum ]