Last August, we posted about ECCE, an “anthropomimetic” robot designed to attempt to duplicate the muscular and skeletal functionality of a human. Kojiro is a robot from the University of Tokyo’s JSK Robotics Laboratory which was built based on the same kinds of ideas. At this stage, Kojiro’s highlight is his spine, which can bend and flex like ours do to allow the robot to move its torso. While it’s true that Kojiro’s spine doesn’t offer a range of motion that’s significantly different from (say) a couple axes worth of conventional servos, the big advantage at this point is that having a structure of artificial muscles, tendons, and bones makes the robot much more forgiving of impacts to both its own structure and whatever it may be trying not to crush. The disadvantage, however, is that such a complex structure is tricky to get to move just exactly the way you want it to, since you’ve got all kinds of different artificial muscles tugging different ways at the same time.

All of this research is focused on building robots that are more human-like, but I’m still wondering just why this is such an appealing thing to roboticists. I totally understand why learning from biological systems is important, but most of the successful applications that I’ve seen are bio-inspired rather than biomemetic, taking elements of biological design and using them to build better robots as opposed to building robots that are modeled on their biological counterparts. As both EECE and Kojiro show, humans are ridiculously complicated inside, and as research towards a potential anthropomimetic endpoint, it’s a good goal. My question, though, is that as a development perspective for practical robotics, are we better off (at this point, anyway) just letting robots be unapologetically robotic?

[ JSK Robotics Lab ] VIA [ IEEE ]

The future of robotics is here. It’s **drum roll** cockroaches? Texas A&M University’s Nuclear Security Science and Policy Institute (say that 10 times fast!) is developing technology that allows cockroaches to be controlled via a tiny chip on a cockroach’s back that sends electrical signals to make the roach move. “It’s like a cattle prod for cockroaches,” says William Charlton, an associate professor at Texas A&M.

But why cockroaches? Well, the same chip that communicates remotely with a computer to prod the roach has several types of radiation sensor, meaning that it can detect whether conditions in a given area are safe for humans.

“Cockroaches really are the perfect medium for this,” says William Charlton, an associate professor of nuclear engineering at the university and a principal investigator on the project. “They can go for extraordinarily long periods of time without food. They exist on every continent except Antarctica. They’re very radiation resistant, and they can carry extremely large amounts of weight compared to their body mass.”

If Charlton gets his way, we’ll have mini-armies of 20 or so roaches surveying areas as large as one square kilometer, all controlled by remote operators, all reporting data about chemical conditions in the area.

I, for one, welcome our new insect overlords

It may look almost cartoonish in proportion, but the picture above is the real deal. How long till we see “Control Your Own Cockroach” kits for the kids?

[ NDIA ] VIA [ Wired ]

From the folks at RSLSteeper in the UK comes BeBionic, “The next generation of fully articulated myo-electric hands.” Being myoelectric means that these hands can be controlled by electrical signals from the human brain. Not only can these bionic digits produce still art straight out of a zombie flick, they’re actually quite… dare I say… handy? “Complete with a range of naturally compliant grip patterns that provide repeatable accuracy, our powerful new hands combine innovative technology with life-like appearance.” In other words these things are AALLIIIVVVEEEEE… or at least pretty darn close to it. RSLSteeper has also released a nifty teaser video showing off just how accurate these things are:

Tell me those things don’t look awesome!

Now, maybe I shouldn’t be talking, seeing as I’ve made some pretty destructive robots, but as cool as these bionic hands look, I always find robots that are capable of actually hurting someone (read: Kung-Fu grip) the slightest bit creepy, and this is no exception. I mean, if we’ve learned anything from The Addams Family, it’s that hands with a mind of their own can only mean bad news.

At the same time, I do have to applaud RSLSteeper- if the BeBionic hands are as nimble as the video demonstrates, then they could have some very important applications, both in helping amputees, and in making more life-like androids.

What’s your take? Handy, or creepy?

[ BeBionic ] VIA [ Engadget ]

This robot (its name is CARL) may not look especially biological (aside from the adorable little ears), but inside, it’s thinking with a computerized model of a rodent brain and interacting with the world through a “biologically plausible nervous system.” The programming for the rodent brain comes from brain recordings of real rats, and when CARL’s brain learns to replicate the behaviors of those rats, researchers hope that they can analyze what CARL’s brain is doing to make inferences about how biological brains work. Specially, CARL (and the real rats) are put in situations where the locations of stimuli that predict food rewards change abruptly, so CARL and the rats have to quickly learn how to adapt. It’s this capacity adaptation that we don’t really understand on the neuronal level, and in addition to offering insights into human behavior, it’s hoped that the research will lead to robots that are better able to decide what to do complicated and changing environments.

[ UC Irvine ] VIA [ RobotBuzz (Translated) ]

The video that includes Boston Dynamics’ BigDog getting kicked and slipping on ice is nothing new, and it’s still one of the best robot videos I’ve ever seen. The vid above includes that stuff, but also some footage that I don’t think we’ve seen before of BigDog descending slopes and getting its foot caught in a concrete block. It’s all very impressive, but even more importantly, it’s damn funny stuff.

[ BigDog ]

Remember that jumping grasshopper robot from May of last year? It still hasn’t quite figured out how to fly, but it can now make more than one autonomous jump in a row, thanks to a primitive simple but effective self-righting system. It’s the same type of thing we saw on the WeebleCopter: a spherical metal framework with the robot on the bottom, where its weight will cause the whole thing to roll upright:

There are some downsides to this system, including increased bulk and most notably a decrease in jumping height of nearly 25%, but the frame does protect the robot, and if it gets stuck in a tight spot, it can use the frame to bounce off obstacles and get itself pointed in a different direction.

[ Self Deploying Microglider @ EPFL LIS ]

Last month, we posted about Lockheed Martin’s SAMARAI UAV, a monocopter designed to test control principles for a much smaller unmanned reconnaissance drone. The University of Maryland has just demonstrated the results of several years worth of research on monocopters, and they’ve got a fully controllable vehicle that uses a single wing no larger than a seed pod, and packs a camera to boot:

As you can see, the UMD monocopter can take off from the ground, hover, fly controllably, and land without killing itself. The key was a lot of research into the flight characteristics of the seed pods themselves, which enables the monocopter to autorotate just like the real thing. Obviously, the camera isn’t really useful for surveillance at this stage, but that’s solvable. Researchers suggest that the craft could be airdropped, autorotate for a while, and then be controlled remotely for “defense, fire monitoring and search-and-rescue purposes.”

The implication of the Lockheed Martin project (based on this image, anyway) is that their final product will be both the size and form factor of a samara, while the UMD monocopter relies on an external motor and battery pack for propulsion, stability, and control. While it would be totally awesome (and very stealthy) to have a little surveillance robot that looks like a seed pod, for most purposes, having a functionally similar design is just as good, and (most importantly) UMD has an aircraft that works… And it only costs $500, which odds are is going to be a damn sight cheaper than anything that Lockheed Martin produces.

[ UMD Press Release ] and [ Project Website ] VIA [ Boing Boing ]

I guess the bee crisis is worse than it seems, because the National Science Foundation is giving Harvard a cool $10 million to develop a robot bee colony. That’s right, not just one, but an entire colony of robot bees. The bees will buzz around on flapping wings, use optical flow sensors for navigation and obstacle avoidance, sport cute little antennae as well as “pollination and docking appendages,” and use an as yet unspecified power source.

This scheme will seem a whole lot less crazy if you recall that Harvard had a lifesize robotic fly that was operational (mostly) back at the beginning of 2008. Even with that foundation, developing the body of the robot is going to be no small challenge, but there’s also the brain of each robobee plus the organization and communication of the colony as a whole to consider.

All of the aims and goals of the project revolve around “pushing advances” and “spurring innovation” as opposed to creating a robobee army. Which is good, I guess. And, in case you were wondering, the robobees are all gonna be dudes, and consequently sting-free.

[ Harvard RoboBees ] VIA [ Robot Living ]

If a hexapod kit is a little bit out of your price range, how about a little bot that uses one motor and one servo, is made mostly of cardboard, and can be built in an hour? Yeah, sounds good to me too, but you won’t be shocked to learn that it’s a little more complicated than that, although not much more complicated. The robot is called DASH, which stands for Dynamic Autonomous Sprawled Hexapod (remember SprawlBots?), and it’s from the Biomemetic Millisystems Lab at UC Berkeley.

The principle behind this robot is that simple + cheap = just about as good as complicated + expensive, a philosophy that this lab as proven before with RoACH. And it really is simple, with the entire robot buildable in 59 minutes flat. Despite this simplicity, DASH is no slouch, able to motor along at a steerable 15 body lengths per second, which translates into 60 miles per hour if it was my size (how’s that for scary). Despite, or perhaps because of, its simple construction, DASH it can survive a terminal velocity impact of 10 meters per second, which means you could drop it off Mount Everest and it would hit the ground running. Plus, the pliable design and rotary leg motion give it the ability to climb over obstacles taller than it is.

The next steps for DASH involve trying out different leg designs and refining the steering mechanism. With such a simple and cheap robot, it might be fun if they put out a kit and sponsored a competition for the best leg design over different types of terrain or something. Oh well, I can dream, can’t it?

[ Biomemetics @ UC Berkeley ] VIA [ IEEE ]

Robot Watch picked up some video of Nissan’s robot cars that we wrote about on Monday. If anything, they’re cuter than I thought they’d be, with the welcome addition of a bunch of blinky lights.

VIA [ Robot Watch ]