Artificial intelligence may answer some key questions about insect flight

Anyone who has tried to capture a fly or other insect can't help but marvel at their aeronautic prowess. Their reflexes are lightning-fast, and they seem to avoid obstacles before they are even perceptible. The brain of a fly or a honey bee is as little as a millionth the size of a human brain, with as few as a hundred thousand neurons compared to our hundred billion. How can such small computing power lead to such effective flight?

Yet flying insects also exhibit curious behavior. They land when flying into a headwind, and gain elevation with a tailwind. Some honey bees will land on the reflective surface of a flat pond and drown. A team of researchers using insect-sized robot microhelicopters believes it now can explain both the amazing flying ability of insects and their seemingly bizarre anomalies:

What these authors call an "optic flow regulator" is a reflex that keeps the optic flow, and thus the speed/altitude ratio, at a constant value. If the insect changes speed, this reflex will make it change altitude so that ratio remains constant. Adjusting the speed/altitude ratio means that the insect has no need to measure either its speed or its altitude.

If there is a strong headwind, its forward speed will be reduced. Thus its optic flow regulator will constantly force it to reduce altitude so that the optic flow always remains at the reference value. The insect has to make a forced landing against the wind, but a safe landing, because it takes place at a vertical speed of zero. Reactions of this type to a headwind have been described countless times in insects and even in birds. They are also observed on the microhelicopter each time it faces a laboratory-produced headwind, reinforcing the hypothesis that flying creatures have an optic flow regulator.

"Optic flow" is just the apparent speed of the ground below you: if you keep this value constant than you'll be flying faster at higher altitudes and slower at lower altitudes, thus sparing yourself from damaging crashes. But when the wind comes into play, it changes the apparent speed of the ground, and so insects -- and their robot counterparts -- automatically change altitude.

But why do honey bees crash into ponds? The reflective surface of ponds makes it difficult to measure optic flow, and so their internal guidance system fails.

Tags
Categories

More like this

This synopsis intrigues me. A tangential and somewhat unrelated question--I wonder how mosquitoes in particular, have peculiar evasive flight patterns that almost seem to have them disappear before my very eyes, only to reappear somewhere else? I've often wondered if it had anything to do with my own visual system and some inherent flaw (do I need new glasses...) or perceptual loophole the insect seems able to capitalize on by flying in a particular manner? Hopf, 2004 suggests some attentional over spatial concepts. I don't know if his work is related to this can't-kill-a-mosquito question though. Bottom line--do you think there might be a perception/processing explanation behind the informal but accurate "no-see-ums" label for these pesky, often successfully evasive insects!
http://www.jneurosci.org/cgi/content/full/24/8/1822
:)

I'm no expert on insect behavior, but my recollection is that most bugs have a few preprogrammed algorithms for escaping based on the direction the attack is coming from.

The large object (your hand trying to slap the mosquito) may also temporarily distract your visual system from the smaller object, allowing it to "disappear."

An aside -- I don't think of mosquitos as no-see-ums. I usually reserve that word for smaller insects -- gnats and the like. Do others call mosquitos no-see-ums too?

Certainly in the hiking community no-see-ums are much smaller than mosquitos. No-see-um netting is advertised as being small enough to keep out those gnats you refer to.

Dave Munger: I agree, "no-see-um" should be reserved for biters sized at a millimeter or less. (If it's a couple of millimeters, I'd call it a gnat; "midges", IIRC, are slightly larger than that -- maybe three to five mm.)

Mosquitos certainly have generic evasive patterns, and they can sense when the skin they've landed on tenses up. (I.e., the victim noticed!) What I don't know is how well they can sense air currents, which is how cockroaches and houseflies pull their swat-dodging tricks. Also, given their shorter signal paths, they probably have faster reactions than you do!

By David Harmon (not verified) on 16 Feb 2007 #permalink