Robot Controlled By Rat Brain

kkleiner writes "Kevin Warwick, once a cyborg and still a researcher in cybernetics at the University of Reading, has been working on creating biological neural networks that can control machines. He and his team have taken the brain cells from rats, cultured them, and used them as the guidance control circuit for simple wheeled robots. Electrical impulses from the bot enter the batch of neurons, and responses from the cells are turned into commands for the device. The cells can form new connections, making the system a true learning machine."

Comments

True learning machine?

[SoupIsGoodFood_42]

If it uses living cells from a rat brain, then it's not really a machine.

Sentient cells?

[Twinbee]

What doesn't seem too clear after listening to the videos is why the rat's cells wouldn't want to crash the robot it's controlling, into the wall. Did the scientists program that in (perhaps wall crashes give the cells some kind of negative electrical stimulation), or did the cells have a mind of its own on that front?

The difference is subtle because it means we have either a 'mere' replacement for computer chips, or potentially much more - a sentient clump of cells which want the 'best' for the robot it's controlling.

Re:Sentient cells?

[golden age villain]

I was wondering the same and honestly it seems fishy to me. There is no such thing as a negative electrical stimulation for neurons. Granted there is inhibition by GABA and some other neuromodulators. So unless they drop something on the tissue to induce some sort of learning, I simply don't see why any coherent behavior would emerge since there is no "motivation" to behave in one way rather than the other. From the wall-avoidance behavior video, my guess is that the sensors continuously feed the network until they detect a surface and then stop. In that case, the behavior would be hard-coded in the sensors and not in the network.

Re:Sentient cells?

[RadioElectric]

"Reward" is an interesting word to use. In whole brains there are entire systems of neurons which control motivation and reward (dopamine, endorphins, etc.). "Reward" at the level of a single neuron means nothing. There are ways of encouraging a particular input/output association (LTP), which I guess is as close as you'd get at the level of a single neuron, but there doesn't seem to be much info on what Warwick et al. actually DID here.

Re:Sentient cells?

[Anonymous Coward]

What about knowledge or behaviour inherent to a species? It's not like the rat brain started from zero like a "start from scratch" computer-learning program would.

Re:Robot Controlled by Rat Brain

[MichaelSmith]

Mainly the neuron control helps the robot to avoid walls.

So there must be messaging back into the rat. So the robot is to some extent controlling the rat brain.

Human brains?

[MichaelSmith]

Say I have a terminal illness. It some of my brain cells can be kept alive, and given a robot body to motor around in, maybe its worth a go.

Randomly wandering robot = Science?

[vadeskoc]

I looked at the videos, but mostly what I saw was a robot semi-randomly driving around. Did they do some kind of experiment to prove they had done something more than set loose a stochastic system with wheels? I tried to follow up on some of the references, but after the second not-so-reputable journal with some kind of barrier to entry, I gave up. If I had done experiments in this vein, I would be yelling as loudly as possible about what tests I did to ensure this actually proves something. You know, so people wouldn't think I was just a crack-pot looking for attention. Doesn't help either that this is the same douche-bag that stuck a chip in his arm and claimed he was a "cyborg". In addition to not feeding trolls, can we avoid feeding media whores in future too?