Learning Robot is "A Lot Like a Puppy"

I'm excited to report that I just read the first news about real AI that's ever made me just a little bit nervous.

First off: apparently, when artificial neural networks are used in AI, they have to be carefully limited to keep robots from doing dumb things. I'm pretty sure that at this level this just means controlling the number of completely irrational solutions the robot comes up with for a problem on the level of "There's a wall in my way," but looking ahead to the future, I like to think that this could mean one missing line of code could stand between peaceful coexistance and robots eradicating humanity.

On a more serious note, the point of the article is that through cleverly mixing artificial neural networks with the traditional "pre-programmed" approach to AI, they found something that works a lot better than either one works on its own.

Here's a quote, because this is just that awesome:

Working in the EU-funded COSPAL project, Felsberg’s team found that using the two technologies together solves many of those issues. In what the researchers believe to be the most advanced example of such a system developed anywhere in the world, they used ANN to handle the low-level functions based on the visual input their robots received and then employed classical AI on top of that in a supervisory function.

“In this way, we found it was possible for the robots to explore the world around them through direct interaction, create ways to act in it and then control their actions in accordance. This combines the advantages of classical AI, which is superior when it comes to functions akin to human rationality, and the advantages of ANN, which is superior at performing tasks for which humans would use their subconscious, things like basic motor skills and low-level cognitive tasks,” notes Felsberg.

What's really interesting is how they handle the problem of setting a robot free to learn on its own. Essentially the robot has no innate criteria for making decisions, so how does it know when it's done something "right?" Well, we tell it so. A human operator has a device with two buttons.

The good boy, button, and the bad boy button.

Wow. Can you imagine when this becomes a toy, or an ethical issue?

There's actually a video game that functions similarly. I guess it'll be harder without a magical fairy.

Fake plants. No, Really.

One of the strategies for producing clean energy is to do it the way plants do it. Photosynthesis. Imagine! Not only creating energy without pumping any CO2 into the air, but actually consuming CO2 and getting energy in return. Still, it's hard to imagine getting enough energy out of photosynthesis to fuel, say, a car. Plants are quiet, and slow. They can't even read our thoughts! But they are awfully good at making possible the cycle of life on earth. I have to give them that one.

According to a Science Daily article, that's one of the things scientists are working on right now. Their progress is meager; they've managed to stabilize the process of water oxidation, which is just the beginning of a long task list of problems to figure out.

Meanwhile, the clock is ticking, just a little bit. Icebergs are melting, and ecosystems people need and rely on are deteriorating. Long term strategies for reducing carbon output are great, but it's also a good time to be looking into how we're taking care of our water resources. We can't necessarily stop global warming before it can do harm, but we've got a bit more control when it comes to polluting and damming the rivers it puts at risk.

(Image by: Paul)

A Visit to the Past, with Rabbits

I guess in a way we could kind of see this coming, given the reproductive reputation of rabbits, but still. Can you believe that these critters have been around for 53 million years?

That spans the better part of the Cenozoic Era (65 million years ago until now) and means rabbits evolved around the same time as grass. Other contemporaries of rabbits include three-toed horses and whales with legs.

Things were strange on planet Earth when rabbits were new. North America and Europe were just starting to pull apart. Dinosaurs had gone extinct only recently, and mammals were just beginning to thrive. But the world was cooling down and drying out, triggering extinctions of 50-90 percent of certain groups of animals.

...But not the rabbits. They made it. I wonder how much further they'll go.

(Image by: Franie Frou Frou)

Beating the Superbugs

MRSA, otherwise known as antibiotic-resistant bacteria, are some scary bugs. There was a fairly recent local outbreak in the D.C. area, which left a lot of people feeling pretty helpless and apprehensive. Public schools are still dealing with the threat -- but the best they can do is focus on prevention. For the most part, we only have a few options when it comes to treating bacterial infections.

So what happens when bacteria become better than the drugs we use to kill them? Do the bugs just win? I mean, I know they've pretty much been around since the dawn of time, but aren't we smarter than they are?

Yes! (Thank goodness.) A group of researchers led by the University of Warwick have investigated the process these bacteria use to get around penicillin's attack strategy of breaking down the cell wall of the bacterium. Now that they know the protein and the chemical reaction associated with resistance, (the trick seems to be that resistant bacteria have been building their cell walls with more ropes than ordinary bacteria) they can work on ways to mess it up.

Once they can disrupt this deceptively cutely named MurM reaction, penicillin will be able to kill the previously resistant bacteria.

I like it when humans win!

Interactive Space Atlas, Stage 1

So, checking out Google Sky, I just used the "hand" icon to grab a fistful of the known universe and drag it two inches to the left.

I think there's something to be said for that.

The bad news about Google Sky is that you can't really use it to zoom right up to planets and see them in all their high-res glory, which is kind of what I was unrealistically hoping for: cratered planets, burning suns, and "stars" expanding out to become galaxies and all that. It's not like a road map, either, or even like a political map. Nothing's labeled.

It's actually even a little unintuitive trying to get the names and close-ups of the celestial bodies you might be looking at while you randomly browse the universe. The easiest way to figure out what's what is to use the search function and ask it to point you at something in particular --only then does a label show up.

Even then, though, all you get is an awkwardly (almost comically) pasted-in picture of the celestial body over the larger star-field, which, even though it changes size as you zoom in and out, still kind of disrupts your sense of cosmic perspective.

There's also a lot of holes -- places where they didn't have the data to fill in any picture at all. That's the worst part. They really ruin the aesthetic experience, but I suppose it's better than false advertising-- putting "black sky" in places that might turn out to be full of stars.

Still, it's great to have dragging and zooming features applied to something like the night sky, where the distances are so difficult to wrap your mind around in the abstract. There are also some cool features that let you check out the same sights using different spectra. What it needs most, I think, is an optional label overlay so you can browse without everything looking like anonymous colored dots.

Check out Ars Technica for a full review combined with a list of "must-see" destinations in the universe.

Change your skin?

I had no idea we were anywhere near being able to do this, but here you are.

Artificial skin made of carbon nanotubes that scientists say can eventually be linked up to the human nervous system to give people the experience of heat, cold, and pressure. Not only will prosthetics be able to move better, but they'll also be able to start sensing like a real limb.

(Although I imagine they'll still lack the sense of motion-consciousness that lets us know the positions of our limbs without looking at them...)

Still, most exciting thing ever, right? Unfortunately, there's no time scale given for how long it will take to develop this stuff, so while we wait, let's learn more about nanotubes!

I am Rubber, I am Glue...

So they came out with this nifty new kind of rubber that can knit itself back together after being cut, eventually becoming as good as new! Not only can you put it back together just by kinda pushing and squeezing it like clay, but THEN YOU CAN STRETCH IT AGAIN. Full healing takes about 6 hours, but you can get it to hold together almost immediately, and start stretching it again after about an hour.

According to the Scientific American article, the way it works is by utilizing the same sort of weak hydrogen bonds that exist in water -- after the bonds are broken, they still have a latent attraction to one another and can be recombined.

What remains to be seen is how we'll use this. Proposed applications include "rubber duckies" and "healing tires." Surely we can do better than that? As far as I know, we've never really had a material that combined durability with the ability to squish back into its original state, so there ought to be great new things to do with it.

I have to admit, my first thought was "androids..."

Why We Get the Flu in the Winter

This was something I always wondered. Why do people get sick when it's cold? The usual consensus always had to do with the time people spend cramped up indoors among other people, facilitating the passing of germs.

But that's not really totally satisfying, and it seems that the reason why is because there's more to it than that. According to an article at abc.net the flu virus is covered in a fatty substance that actually hardens in the cold. Instead of killing the virus, the hardened coating actually protects the virus from the elements, allowing it to survive longer outside while it waits for a biological host.

What's most interesting to me about this discovery is that before I read the article, I figured someone out there completely understood the subject--even if I didn't. So what I mean is, I noticed here that sometimes there can be a hole in the scientific explanation of how something works that the average person might not even realize is there.

Here's another example. Scientists recently discovered that bacteria in the atmosphere play a significant role in the creation of ice and snow. Bacteria are sent up into the atmosphere by wind, and they wind up becoming the most active "substrate" for gathering ice crystals and water droplets until they form a unit big enough to fall to the ground in the form of rain. Dust and soot can serve the same purpose, but bacteria form a much larger percentage of the substrate than we thought, and can accomplish it at warmer temperatures.

And just like that, there's a biological component to the water cycle. The finding could be extremely useful, too, as it will add another potential tactic for cloud-seeding during a drought.

Either way, I'm still surprised. I just never knew the water cycle was something we didn't fully understand.

Ships you Can't See

Did you know that radar was invented as a sort of consolation prize after Robert Watson Watt determined it would be impossible to invent a death ray?

Anyway, I was in the middle of reading this article about how stealth technology for warships is improving, when I came across something that made me pause and say "Holy crap."

Here it is:

"Metamaterials are tailored to have specific electromagnetic properties not found in nature. In particular, they can bend light around an object, making it appear to an observer as though the waves have passed through empty space."

Come again? Invisibility? What? Huh?

But hold on. Let's hear some more and get disappointed.

From an interview with David R. Smith, one of the physicists working with metamaterials:

TR: So an object inside the shield is actually invisible?

DRS: More or less, but when we talk about invisibility in these structures, it's not about making things vanish before our eyes--at least, not yet. We can hide them from microwaves, but the shield is plain enough to see.

So...only microwaves. Bummer. And Smith also points out that the way metamaterials work isn't really all that conducive to using them to hide away giant objects like ships and planes. Right now, some more likely uses are better data storage and solar energy technology that can run on indirect sunlight.

Still, it seems the biggest barrier right now is the size of the waves -- they're tiny, and it's hard to build tiny. The microwave demonstration was easiest to make, but a demonstration that applies to visible light could be coming. Someday.

Or just more room for special features on our DVDs. And less global warming, with better solar panels. Which is good too.

(Image: The Visby Corvette, a Swedish warship that uses "dazzle painting" among other techniques to make it harder to see and detect. naval-technology.com)