Bomb sniffing locust

Imagine a world where cyborg locusts are trained to detect bombs. Sounds strange? One lab in St. Louis Missouri is busy making this a reality.

TRANSCRIPT

Imagine a world where cyborg locusts are trained to detect bombs.

Sound strange?

One lab in St. Louis, Missouri, is busy making this a reality.

Let's take a look.

So, here's the star of the show.

It's a locust -- a grasshopper, really -- wearing a backpack -- a tiny computer wired into an even tinier brain.

Welcome to the world of biorobotics.

At Washington University in St. Louis, a team of scientists and engineers, headed by Barani Raman, is working to create a bomb-sniffing machine that uses the antennas of living locusts as the detector.

It can actually amplify the signal that they're recording.

As a grad student, Dr. Raman built a smelling machine, a mechanical nose, but realized there was no way he would ever get close to what nature had already come up with.

So, turned out it was a very humbling experience.

Even an insect, like locust with its antennae, has hundreds of thousands of sensors on the antennae, and there are so many different varieties of them.

So, in terms of the complexity that the biology has, it beats the engineering, hands down.

With funding for the Office of Naval Research, they hope one day to be able to send a swarm -- yes, a swarm -- of bomb detectors wherever needed.

Great idea, really complicated to do.

First of all, you have to train the locust to recognize a particular odor.

Then you have to equip it with a backpack computer, and then figure out how to steer the locust where you want it to go.

Turns out that training a locust to detect a smell, that's the easy part.

The researchers won't exactly say what chemical they're targeting, but let's say it's TNT.

So, what is going on here is, there are a bunch of six locusts.

Each one is getting trained.

So in order to give them this, they are being starved for a day, so they are very motivated to learn this task.

The locusts are strapped in, facing a tube that will send puffs of the odor.

Their mouthparts are painted green so researchers can easily see them when they open up.

And it only takes a few sessions with hungry locusts.

They get a puff followed by a piece of grass.

Puff, grass, puff, grass.

Pretty soon, they get a puff, and they're opening their mouths in anticipation.

So, they get a puff...

They get a puff.

They've been trained on this puff already.

See that?

Yeah.

Yeah, so that definitely is.

Here's the instant replay.

And they keep responding to that smell even when they're not being fed.

So we can actually train them like Pavlov's dog, right?

So, you ring the bell, feed the dog.

Same thing you can do with a locust, too.

Training to recognize the smell?

Check.

But here's what they're really watching now -- not the moving mouthparts on the locust.

They need to identify what the odor response looks like inside the locust's brain.

So, the popping sound that you're hearing are the neurons of the brain talking with each other.

Ah.

These are the wires that get implanted in the locusts' brain.

Say it again -- you've implanted electrodes in the locust's brain.

Arrays of electrodes.

Not just one wire but a bunch of wires.

And this is where they start turning an insect into a cyborg, part animal, part machine, that can translate, amplify, and transmit a bomb alert in the field.

Enter the guy from the Computer Science and Engineering Department.

Because my area is in electronics, integrated circuit design, microchip design.

We can only train biology to pick up interesting signatures.

But then the filtering mechanism has to be done on the electronic side.

There's a lot going on in this little backpack.

Right now, yes.

But the ultimate goal for this project would be to shrink everything into a very tiny, 3 millimeter by 3 millimeter dot.

So it will not even look as big as this.

It will be close... It will be as small, or as big, as something like the locust's head.

But then the challenge is, of course -- which is true with any electronics -- is how do we power them?

So that's where also my expertise also came in is because my area is to, how do make such systems without running into batteries?

Right.

So, that's also one of the parts of this project is, how do you make this whole electronics run without any -- by harvesting energy from the ambient source?

So, let's say they solve all the challenges of detection and transmission, and the locust is ready to go to work.

How do you keep it from just flying away?

There has to be some method of remote control, of turning the locust into a drone.

For that, they turn to a professor of engineering and materials science, who brought in a vial of gold nanoparticles, which are most commonly used in cancer treatment.

What these gold nanorods are good at is they can absorb a lot of light and convert the light into heat.

Here's the thing about locusts -- they naturally turn away from heat.

So they will tattoo a bit of that heat-absorbing gold on each of the locust's wings.

Heat the right side, the locusts goes left, heat the left, it turns right.

You can apply the heat with a light, and in the lab, it works, but they're going to have to do this in mid-flight.

But at some point later, what we are going to do is actually have the light right on the locust itself.

We are going to have tiny LEDs on the locust itself, and you just turn on and off these LEDs to start heating these tattoos and start steering the locust.

If that steering system works, they may have a working prototype -- a bomb-sniffing, data-processing, remote-controlled cyborg locust -- by the end of 2017.

Locusts only live a few months, so the training and the wiring continues on a regular basis.

But they're small.

They're cheap.

They're easy to train.

And they don't spend any time thinking, analyzing, considering options.

It's those antennas.

When it comes to odors, they can find a needle in a haystack.

Actually, the truth is, they are engineering marvels.

For the amount of footprint they have, the type of sensors they have, the type of energy consumptions they have, we cannot make machines that actually match them.

We need machines that match them.

So if you cannot build one, why not hijack one?

That's the philosophy here.

And we have tools right now to hijack.

All these tools are there, independently, if you look at them.

They are pretty standard tools of the whole field.

It is the integration of these multidisciplinary tools together that makes this particular project possible, and this is the perfect time to do that.