?

Log in

No account? Create an account
corvi
corivax
.:::. ..::.: .:.::..:.::. .::::

October 2008
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31

corvi [userpic]
robot robot hop!

We are just now doing the final editing pass on the TEDP for NASA, and I wanted to share this impossibly cute image lilsquishy drew of two flight team members and one of the NASA zero-g techs handling a certain aggravating cubical robot.



It looks so cute! And so professional!

And derangedly optimistic! Nobody's space sick, and everyone is paying attention. I predict at least one team member will spend most of the time doing zero-gravity somersaults and bouncing off the walls and yelling, "The enemy's gate is down!" randomly at other research teams. La la la. But of course, I have no idea who and wouldn't dream of it myself. Not at all.

I'm at that stage of punchdrunk where I'm inordinately amused by how much one of the tiny, cute little allen screwdrivers is well-weighted for knife-throwing. And team members keep sitting in front of our corkboard. I want so. very. badly. to throw the screwdrive and bury it in the cork next to someone's ear with a nice loud TWANG-thunkathunkathunka. Those've you who've never been on a big project with me just don't know what you're missing! And thank god for that. :)

xmurf says, "I must be tired. I haven't won a game of solitaire yet."

Current Mood: amused4 AM! Everything's funny!
Comments

I'm curious about this robot... What is its purpose? and what does GYRE stand for?

We always thought we'd come up with an expansion for GYRE eventually. I maintain it stands for 'GYRE: Your Robotic Explorer". This usually makes my teammates throw bits of foam padding at me. We're all hoping NASA never gets around to asking what it means.

Moving around in zero-gravity is actually remarkably tricky. There's the whole 'wheels don't work' problem. There's the fact that you can't trust where you are or whether you're right-side up - everything floats around. There's always the possibility of completely overshooting your goal, and then backing up and overshooting it again.

So every robot NASA has right now has to be controlled by a human, someone who can watch what's going on and decide what thrusters to fire and for how long and when - it has to be watched every step of the way.

What we're trying to do is create a robot with enough built-in intelligence that it can be given high-level commands (go to the flight cabin, hold still, carry this wrench to the other astronaut, return to your cradle) and will be able to work out all the stupid little details of thruster fire and orientation and direction itself without intense human supervision. It would be awfully helpful.



It has three cameras (the blue and white thing on one of the bottom edges), and three very fast, specialized computers to analyze the images (they're inside the silver box with the ports on the back) and tell how it's moving at any given time, and where it is. COuple of pressurized paintball tanks (big brown things), and twelve miniature thrusters (silver cone in the middle of each edge) for propulsion.

We only have thirty seconds of zero-gravity at a time, so actual tests can't be so complicated. What we're doing on this run is getting into zero gravity, having someone spin it around, and then having the onboard artificial intelligence use the cameras, computations, and thrusters to bring itself to a stop (reletive to the plane). EVen this is a lot harder than it sounds.

Aren't you sorry you asked? :)

You wish that all we threw was bits of foam.

Who still has an enticingly weighted little allen screwdriver.

Who is still sitting in front of the corkboard?

Yaah. I thought at much.

Not only am i not sorry that i asked, i'm fascinated :) It sounds like a very interesting problem to tackle. The complexity of figuring out where you are, where you're going, your orientation, and then controlling that with no gravity, is quite immense.

Do you use some kind of highly visible markers for it to see on surfaces to help work out where it is (the reflective blobs on motion-capture suits spring to mind), or is it smart enough to simply handle images of any room you choose?

Very, very cool.

We're trying to do it without them. We don't know whether it will work yet. We do know from the last time we did this that it's possible to get correct data *most* of the time without them. But we don't know if that small amount of bizarre, horrible errors is going to be enough to confuse navigation beyond repair. Our experiments this year are designed to test feedback/response software.

Have you thought of three sensor probes in the space and two transmitters on the robot and finding position and orientation based on reception?

I know they were doing something like that for the VR studies in the UW HITL.

We could do something like that, or attach color targets to the walls, but all that sort of thing misses the point of this research. The whole idea is to make an adaptable, intelligent robot, one that can operate anywhere - a space station, the shuttle, out in space...

If we make it dependant on external beacons or targets, we lose that flexibility utterly.

Ah.. so does it have edge recognition?

And lasers (distance recognition)? Everything is better with lasers :)

We tested dge recognition two years ago and found it insufficienty accurate. No lasers, either, though that would be a lot of fun.

Then we'd get to fill out the laser sections of the TEDP! Can we work in an IRB review at the same time?

The plane pitches over a parabola for freefall, doesn't it? What does this mean for a robot trying to come to a rotational stop?

To be at rest relative to the inside of the plane is the goal.

I'd never actually thought of it before, but I didn't notice any coriolis-ish forces when I went up last year. The freefall is pretty rough, though -- lots of little bumps from turbulence that twitch the exact acceleration back and forth by several centi-gs. A slow rotation across the entire parabola could easily be lost in the noise.

It's hard to remember once you're in there what exactly is going on outside the plane. I meant to go watch out one of the (very few, very small) passenger windows during a parabola, but forgot. Maybe I'll remember this year during the lunar or martian gravity cycles.

Wheeee!

How long has it been since you slept? That particular state of manic lunatic glee generally comes on about the twenty-fifth hour, or the fourth shot of expresso in my mocha... Not that I've ever, ah, sat at an all-night cafe with a crowd of friends working through statics or physics homework, or some team project while someone builds bridges with creamer packets and stirrers, and actively contributed to transcribing the interaction in the format of playing a MUD, while somebody else's conversation wandered off into trebuchets versus catapults.

*grins* best try hitting the corkboard first without someone there, no matter how nifty it would be to try it with them there first. Can you encourage them to go fetch coffee or something so the space gets vacated?

Have tons of fun!