Saturday, December 4, 2010

Simple Robots

Hey Susan,

This is only tangentially related to evolution, but our conversations have made me think about something. I've been reading about invasive species and such, and I was thinking that it would be nice to have a way of getting rid of them. But there's a reason they're invasive, right? They have no natural predators, or they compete better than whatever's filling that local biological niche. So it's hard to git em. 

Well, I mentioned to my friend Ben that we should just make robots that go out and kill the buggers we don't like. He was immediately dismissive. I was taken aback my his assuredness, and so I asked why he thought that. He said that we can barely get a "roomba" (irobot.com) to work. And that was the sum total of his answer. 

Of course, I hadn't been thinking of a roomba, or a Terminator (http://www.imdb.com/title/tt0088247/) -stlye humanoid gun-toting machine. I don't know, I was thinking more of a ball that moves around. Like a hamster ball (http://www.youtube.com/watch?v=98M1mqKGJJc) without the hamster. Hell, it could even have a hamster making it move. The goal is to kill the thing to be killed, not to be cute or cool while doing it, and certainly not to be clever while doing it. 

Let's say you want to get rid of squirrels: let loose a bunch of rolling things that roll around until they bump into the squirrel. There needs to be a way to positively identify the squirrel, and then you kill the offending animal some way. Poisoned food, injections, explosions, entrapment...something. Boom! No squirrel. 

Ok, there are lots of logistical problems with this. But here comes the fun part: trying (and failing) to solve one problem sometimes presents a whole new line of reasoning. So this is it: a different way to make a robot. An evolutionary robot. 

Maybe the reason that we've had such a hard time making robots is that we've been trying to make mammal-like robots. Robots that move around on two or four legs (on wheels or not), making carefully calibrated movements to get from point A to point B and then accomplish it's tasks. But that's hard! As any evolutionary scientist will tell you, it takes a long time to get that kind of precision down. It took the earth 4 billion years, for gosh sake! Man is clearly speeding up the process, but we're failing miserably at the same time.

So why don't we start simple? Like in the rolling ball scenario?

Jon (who is a huge instigator of biological inquiry in my life) was telling me about some single-celled organisms. Probably bacteria. These things are so dumb! Basically a ball with a tail this guy will spin its tail clockwise, which propels it forward. After a short while the guy stops and "tastes" its surroundings to see if the amount of food in the area has increased or decreased. If it has increased, he spins his tail clockwise again and continues in the same direction. If, on the other hand, the food supply has decreased, he spins his tail counter-clockwise, which has the effect of rotating the creature crazily, making it point in a different direction. And then it spins its tail the clockwise again to head off in this new direction.

In this way, using simple randomness and probability, the creature gets enough food to survive and reproduce.

I love the simpleness and stupidity of this thing. It's not the most efficient way of getting to food. Many wrong turns are made. But it gets the job done! 

Even if my idea of making a swimming robot that is nothing but a ball with a tail (and maybe a deadly hypodermic needle) isn't actually going to manage to kill off an invasive fish species, it's an interesting idea for future problems that might be solved by simple, stupid machines. 

I just thought of one! Send one to Mars! I mean, that rover is so meticulous and slow, and ground control cheers whenever it hurdles over a pebble. If they sent along 3,000 randomly rolling balls with cameras, they'd perhaps get to see more of the planet. And take more risks. Maybe they could have a clear plastic shell for protection with solar panels with apertures for cameras just underneath. They could beam their info to the rover, which could beam it up to earth.

In fact, they could have a different ball for each thing they were trying to accomplish. A ball for testing dirt samples. A ball for sensing air quality. An explosive ball that sends info about buried minerals back to the rover right before it succumbs to immolation.

As an aside, I've heard people theorize about "nano-robots" that we'd place in our blood to find and destroy diseases. I can only imagine that these guys would be much more like my robots than like Johnny 5 ( tinyurl.com/29oagz4 ) . 

That's all I got,
Brandon

PS I guess this is what happens when I'm typing with all ten of my fingers instead of just my thumbs!

No comments:

Post a Comment

Total Pageviews

Contributors

Followers