Researchers Using AI To Build Robotic Bees

An anonymous reader writes "British researchers at the Universities of Sussex and Sheffield are developing a computer model of a bee's brain that they hope can help scientists better understand the brains of more-complex animals, such as humans, and perhaps power artificial intelligence systems for bee-like robots. Called 'Green Brain,' the project is trying to advance the science of AI beyond systems that just follow a predetermined set of rules, and into an area where AI systems can actually act autonomously and respond to sensory signals."

Comments

Promising...

[raydobbs]

...every good project has to start somewhere - and it will be interesting to see what this kind of AI modeling will accomplish. Perhaps we can learn more about bees, and how to keep them doing their busy work throughout our world without mass murdering them. ...that being said... the day they crack the secrets of modelling the human female's brain... there is where the real money will be made.

Re:Promising...

[TheLink]

To me it seems that they should figure out how the single celled stuff (like amoebae and neutrophils) think before they go on to more complex stuff.

From what I see they seem smarter than most people (including many scientists) assume:
https://www.youtube.com/watch?v=I_xh-bkiv_c
https://www.youtube.com/watch?v=pvOz4V699gk&feature=related
http://www.brianjford.com/a-08-12-infocus_cell-intelligence.pdf

Have our AIs reached the level of intelligence of an Euglypha amoeba, which builds a pretty decent shell for itself: http://starcentral.mbl.edu/microscope/portal.php?pagetitle=assetfactsheet&imageid=26590

It's quite an elaborate shell - with holes in the front and back. The number of "teeth" in the shell apparently is not determined strictly genetically either see 3) in:
http://what-when-how.com/molecular-biology/maternal-genetic-effects-molecular-biology/
Note that it builds a new similar shell when reproducing.

It may be that a single neuron is actually not that stupid and it's because you need redundancy and the ability to control a large creature/"machine" that you have to have many of them and a multicellular body.

Curiosity.

[Ostracus]

A certain Mars Rover comes to mind.

Bee Brain

[busyqth]

for (x in recognize_nearby_objects())
{
if (x.type == FLOWER) {
nuzzle_flower(x);
}
else if (x.type == HUMAN) {
sting(x);
}
else if (x.type == SMOKE) {
sleep(1);
}
else {
buzz();
}
}

Re:Bee Brain

[pitchpipe]

// The following section is essentially complete. Someone
// just needs to flesh out a few of the functions
// (probably just take an intern an afternoon)

for (x in recognize_nearby_objects())
{
if (x.type == FLOWER) {
nuzzle_flower(x);
}
else if (x.type == HUMAN) {
sting(x);
}
else if (x.type == SMOKE) {
sleep(1);
}
else {
buzz();
}
}

Scale this up

[teaserX]

These don't have to be limited to just RoboBees. The algorithm could be used for more than just pollination. Think about it. Build anything of the appropriate size to autonomously go out and collect $RESOURCE, return with a load, refuel itself and go back out. Some cursory self-defense, like hazard evasion, would be nice. Throw in some networked communication to help with discovery of sources and you have a very efficient way to accumulate stuff.

Re:Scale this up

[wierd_w]

Logistically, if the swarm could not manufacture new units, and or, collect and repair damaged/errant units, the system has serious vulnerabilities.

take for instance, the human greed factor.

If there is a huge swarm of autonomous robots out scouring riverbed sandbars for teensy gold nuggets, or some other discrete but scattered and valuable resource, how long do you think it would be before unscrupulous people tried to trick the bees into dropping the cargo off at a "new" dropoff point?

Where there is profit, there will always be dirty dealing and crime. Look at the internet for instance, with something seemingly as harmless as email. Then along came the spammer.

Autocollecting robot swarms would be a smorgasboard for whitecollar criminals.

Tonight on SyFy...

[Zephyn]

Irwin Allen and James Cameron proudly present:

THE SWARMINATOR!

Excellent!

[PRMan]

Maybe these ones will be resistant to Monsanto products...

The problem with programming a brain..

[andydread]

may possibly be the approach many of these very smart researchers use. Perhaps the focus should be on developing some kind of artifical nevous system with the abitlity to learn on its own rather than trying to program for the dynamics of real world interaction. Perhaps the folks over at Boston Dynamics may be on to something? Not sure what its learning/memory capabilites are but it sure seems to behave like it has some kind of nervous system.

Monthy Python - Eric the Half a Bee

[Master Moose]

Is this a wretched demibee,
Half asleep upon my knee
Some freak from a menagerie?
NO! It's Eric the half-a-bee!