PDA

View Full Version : Slice of Rat's brain learning to fly a simulator.



georgeo76
10-23-2004, 01:56 PM
check it out (http://www.wired.com/news/medtech/0,1286,65438,00.html?tw=newsletter_topstories_html )

georgeo76
10-23-2004, 01:56 PM
check it out (http://www.wired.com/news/medtech/0,1286,65438,00.html?tw=newsletter_topstories_html )

VVanks
10-23-2004, 02:13 PM
I don't really get how that works.

ddelarosa
10-23-2004, 02:14 PM
The picture in the article shows X-Plane as the software being used. Austin Meyer wrote a very valuable research tool in ways he probably didn't even expect.

T_O_A_D
10-23-2004, 03:00 PM
Oh great soon a "Rat on a slider" http://forums.ubi.com/images/smilies/35.gif

antifreeze
10-23-2004, 03:07 PM
Oh yeah. This may be the Forgotten Battles AI of the future. When you buy Oleg's new flight sim in 2010, included you'll get a petri-dish of brain with a fire-wire cable dangling from it!

TheJoyStick
10-23-2004, 04:30 PM
That can't be real.. They say that the brain can fly a plane in different weather conditions, and even learn how to stabilize it based on how crappy the plane is flying.


That would all be based on visual input, and AFAIK there were no eyes to visually input anything.

Unless theyre feeding the stuff straight to the brain, and using a monitor to view what the brain is doing, I don't see how that'd be possible.

antifreeze
10-24-2004, 05:10 AM
It doesn't quite work like that. The cells don't know they're hooked up to a flight sim; that's just a jovial choice on the part of the doctor I think.

All that's happening is that in their normal state, the cells are in some sort of electrical and chemical balance. The doctor is applying unique electrical stimuli depending what is happening to the plane, and the cells change in order to reach a balance again.

This is different from a conscious brain, because your brain uses stimuli to make decisions. The cells aren't making decisions, they are just trying to maintain a specific state. So this experiment is more akin to brain-stem cells automatically maintaining internal body functions, and keep everything in balance. It has nothing to do with conscious thought.

So he's rigged the flightsim to other equipment so that when the plane moves laterally or vertically it feeds an electrical stimulus to the cells. The cells do whatever they have to do to bring themselves back to a balanced state. And the impulse from the cells is fed back to the simulator. If they made the right change, the plane will once more be in balance and so the stimuli will stop.

So, say a 'virtual' gust of wind blows the plane left. A unique electrical impulse is sent to the cells; it is unique because it depends on the strength and direction of the wind. The cells start making electrical and chemical changes to themselves to counter that stimuli and remain in balance.

In the beginning they would do this more or less at random. But in a kind of evolutionary way, neuron paths are made which 'remember' successes and failures (a failure here is that the cells are unstable for a longer period of time. A success is retaining the balance state). This is how the brain 'learns' normally; it actually does change itself physically like that.

So eventually the cells 'learn' how to counter all the stimuli that they are given.
Note that his next step is to include a horizon. This means that at the moment, the cells have only learnt to fly in a straight line; they just want to stop the plane from moving laterally or vertically from it's initial vector. But that vector might not be stable horizontal flight. That's why he calls it an 'autopilot'.

But the flightsim is just a cool way to feed the stimulus to the cells. He could have used any software or hardware where a balanced state is desirable.
Note also that the difficult piece of work here isn't to make the cells learn to control the plane, afterall, that's a natural process. The difficult part is studying the changes in the cells so that the doctor can make a much more refined feedback loop from the cells to the flightsim. The cells are currently generating masses of information, but the doctor can only understand and use a tiny amount of that in his 'network'. Conversely, his electrical input into the cells is also probably incredibly crude also.
The cells can only have more interaction with the outside world when the doctor learns how to 'communicate' with them better. The focus is much more about what the doctor learns about the cells than what the cells learn about 'flying'.

triggerhappyfin
10-24-2004, 08:30 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by VVanks:
I don't really get how that works. <HR></BLOCKQUOTE>

Just fly online and you´ll meet brains smaller than that http://forums.ubi.com/images/smilies/88.gif http://forums.ubi.com/images/smilies/mockface.gif

Zatorski
10-24-2004, 11:13 AM
I think that is why it gets frustrating on-line, especially playing shooters. I'm almost 40 years old, I go around a corner in a shooter and try to identify targets, aim and shoot at them. 12 year olds don't think, they just run around the corner and pull the trigger at the slighest bunch of pixels that could be a target. Using a young person's brain would be unethical so the rat brain has all the criteria for a scientifically notable experiment.

Maple_Tiger
10-24-2004, 11:27 AM
WTF is our world comming to?

We are all doomed!

It's the end of the world!

http://forums.ubi.com/images/smilies/35.gif

TheJoyStick
10-24-2004, 03:54 PM
That makes more sense, but that still makes me wonder what the significance of the weather effects are.

In any case, I can only imagine what technology like this could be used for, but nevertheless, it is rather fascinating

antifreeze
10-24-2004, 04:16 PM
&gt; That makes more sense, but that still makes me
&gt; wonder what the significance of the weather
&gt; effects are.

There is no significance to the cells.
The weather (ie. wind and turbulance in this case) is just a parameter that can be used to apply an external stimulus. The other external stimulus is the flight model of the plane. All they need do in the experiment is move the aircraft away from a balanced state, either in a random way like the wind, or in a more uniform way like the flight model (ie. when the aircraft gets faster it always climbs).

It could just as well be the Window's 'joystick configuration window' instead of a flight sim.
Imagine the cells want the cross in your configuration window to be in the centre; the balanced state. If you moved your joystick left and held it there, the cells would go crazy trying to get the cross back to the middle.
In the beginning you would see the cross go all over the place, until it reached the middle probably by accident, but then it would stop there. If you kept on doing that same movement, eventually the neuron paths would be created that allow the cells to move the cursor straight back to the centre; eventually you'd move the joystick left and the cross wouldn't even move.

Now you move the stick right, and because the cells haven't encountered that before, they go crazy again, and the cross takes a long time to find the centre; the cells just randomly change chemically and electrically until they are at balance again. But eventually they will 'learn' how to keep the cross in the centre when you move the joystick to the right.
After lots of 'training', you could waggle your joystick around all over the place, but the cross wouldn't move because the cells would be adjusting as fast as you could move the stick. That's how the cells are keeping the aircraft straight.

Actually, I'm saying all this, but most of it is supposition. It just seems the most logical scenario from reading that article.

x6BL_Brando
10-24-2004, 05:31 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>I can only imagine what technology like this could be used for, but nevertheless, it is rather fascinating <HR></BLOCKQUOTE>

Well...perhaps not some kind of weapon or pilotless aircraft. When I had the accident that lost me my arm, I was unlucky enough to have three major nerves pulled out of my spinal cord. The end result is that I still feel like I have an arm, but it feels like an elephant is stood on it http://forums.ubi.com/groupee_common/emoticons/icon_frown.gif - so any research that gets done on how to retrain the neural pathways would be a plus in my book!

Zatorski
10-24-2004, 06:00 PM
This is really interesting. The experiment was down to the celluar level. There is a lot of work being done on stem cell research right now that will benefit mankind. My take on the research is with the rat brain finding celluar equilibrium, involuntary reponses can be restored in those that do not possess the pathways due to injury, and science can help those with jobs that require extreme reflexes like for instance, fighter pilots.
My earlier statement about 12 year olds don't think has to do with atheletism. Ever pitched a game of baseball? Thrown a bowling ball or darts or drove a race car or a fighter plane? You can't aim a baseball and throw a great curve, you can't clinically pilot a rally car. "in the groove" and "on top of your game" comes from a nervous level that science does not understand yet. Fighter aces have it, sports heros have it, and I submit, young people have it.

Good Luck Brando, I hope you find your life is enhanced by fate, fortune, love and science.

WTE_Galway
10-24-2004, 06:22 PM
surely if it involves flying a budgy brain would work better than a rat http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

ddelarosa
10-24-2004, 10:09 PM
A lot of people who read this story may come away with the impression that airplanes are so easy to fly a rat can do it.

In reality, autopilot devices have been around almost as long as airplanes. The first was introduced in 1914, and used gyroscopes. Straight and level flight is tedious and boring making it a task well suited to automated machines/computers. There is more to flying than just that task however.

x6BL_Brando
10-25-2004, 07:14 AM
Thank you for your kind words Zatorski <BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Good Luck Brando, I hope you find your life is enhanced by fate, fortune, love and science.
<HR></BLOCKQUOTE>
I'm lucky enough to have the love angle worked out http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif and I'm involved in the "bleeding-edge" of pain control with a team of 'world-class' doctors and surgeons, albeit as a lab-rat! Which kinda ties in....

The problem is that this work is mostly aimed at 'terminal' patients in the last stages of cancer and other illnesses. Getting feedback from those poor souls is very difficult so a talkative, relatively young person is what they need. I fit the bill (wish i didn't, but...) and I'm due for a "Borg implant" that will stimulate parts of my brain that "ordinary washing-powders won't reach". I hope it works, for me and all those people who deserve a peaceful end to their lives. Here's hoping.

PS. The reason they don't use guinea-pigs: they just want to stuff the cockpit with shredded paper and go to sleep! http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

oFZo
10-25-2004, 08:00 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by triggerhappyfin:
<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by VVanks:
I don't really get how that works. <HR></BLOCKQUOTE>

Just fly online and you´ll meet brains smaller than that http://forums.ubi.com/images/smilies/88.gif http://forums.ubi.com/images/smilies/mockface.gif <HR></BLOCKQUOTE>

Nail-on-the-head. http://forums.ubi.com/images/smilies/16x16_smiley-tongue.gif

TX-EcoDragon
10-25-2004, 01:51 PM
:-/

triggerhappyfin
10-26-2004, 05:58 AM
<span class="ev_code_RED">Slice of Rat's brain </span> learning to fly a simulator.

Stop posting threads like this...

My wife got a new name to call me http://forums.ubi.com/images/smilies/1072.gif

That mean woman http://forums.ubi.com/images/smilies/cry.gif