Driverless cars are already on the streets in some places.

Now Central New York scientists are moving from the roads to the skies designing and testing drones that can think for themselves.

Here's the story.

All drones are not equal.

There are, like, a photography drones that you can use for cinematography, and someone can easily pilot that one, and you can use it, like, with GPS.

And people are thinking, 'Okay.

We can buy few of those drones and use it for inspecting, like, my nuclear power plant or inspecting this right line or inspecting this farmland.'

No, that is not the case.

So Akrobotix, we are developing safe, reliable, autonomous unmanned systems in all domains -- space, air, ground and underwater.

It's purely mathematical.

We provide scientific solutions rather than engineering fix, and we want to get this trust, so missions are not going to be harmful.

We can trust this mission.

It's going to be really good, and it's going to help humanity.

A lot of people have an idea of what a drone is, and they think of it as this sort of remote-controlled thing.

You're trying to take it to the next step.

What's that next step?

So there is a lot of research ongoing where you have onboard autonomous unmanned aerial systems, which... At which point it's basically a flying robot.

So the overall mission goals are supplied by the human, but most of the details of planning the trajectory, executing the tasks is done onboard autonomously.

And that's the next level that a lot of research is going on in that regard.

Was the drone programmed to fly in a certain way and then you see if it actually flew the way you wanted it to?

Yes, exactly.

Yeah.

Okay.

So we have... So that drone is just like a black box.

It doesn't know anything, so all the command and control is from this one.

Okay.

We are trying to make drones smarter and more autonomous, and whatever you see in our lab, actually, are being used for this purpose.

There are eight Vicon cameras, which are motion-capture cameras.

They help us to know the current position and orientation of drones, and then we can send the required comments by our control, guidance and navigation codes and algorithms we have to drone, and drone will know, actually, where to go and how to go.

So to make sure I understand then, the drone has the sensors on it.

The cameras detect where the drone is.

The cameras send the signals to the computer.

Yes.

The computer sends the signal to the drone, so the drone know where it is.

Yes.

And this is all a feedback loop that's happening, and how specific is it?

Sort of, generally, you know the drone is sort of over here or sort of over there?

Down to 1 millimeter?

Yes.

So we know exactly where the drone is?

Yes.

Exactly, precisely.

Why is that so important?

Can't you just be a few feet off?

Just thinking big picture and as you look forward, what happens with unmanned drones when everything is successful, say 10 or 15 years from now?

Is the idea of delivering Amazon and pizza by drone... Could that be real?

I mean, the technology is there to deliver packages.

What it needs, actually, is the factor of what happens when something bad happens, like when a drone crashes.

Who is responsible?

Those are factors that have to be worked out, and those involve not just the technology developers like us but involve regulatory policy makers.

They involve federal agencies as well as local communities.

But isn't part of your research making sure they don't crash?

That's what it is about.

So the autonomous part, the safe autonomy, the reliable autonomy part is to ensure that it does not crash, but there are things that happen in nature that we are not in control of.

For example, what happens if you have a bad weather or sudden wind gust blows your craft off course, and that's actually the more challenging as well as the interesting research questions that we are looking at.