Robotic Insect Eyes Destined for Next-Gen Micro Drones

This flexible new camera can see even better than bugs do

3 min read
Robotic Insect Eyes Destined for Next-Gen Micro Drones

Just a few weeks ago, IEEE Spectrum wrote about an artificial compound insect eye that was developed by a group of researchers based in the United States. Not to be outdone, a group from EPFL in Switzerland has announced their own artificial compound insect eye, and we got a hands-on a few weeks ago in Lausanne.

Generally, we like to make camera systems that work like our eyeballs do. And that's fine. But in a lot of ways, human eyeballs are terrible. The most successful class of animals ever, the arthropods, have gotten along just fine with compound eyes for a very long time, and the most sophisticated eyes of any animal are of the compound variety (belonging to our friend the mantis shrimp).

So obviously, compound eyes have something going for them, which is why researchers in general (and roboticists specifically) are so keen on developing their own versions. The eye to come out of EPFL this week is unique because it offers a huge insect-like field of view, very fast performance under all sorts of lighting conditions, and most notably, it's mechanically flexible: at just 1 mm thin, you can bend it into different shapes.

We got a look at this thing a couple weeks ago while we were visiting Dario Floreano's lab at EPFL, and it's totally cool:

It's amazing how the use of this flexible substrate enables sensors that are not just bio-inspired, but in fact end up nearly identical to the types of compound eyes that you find on everything from flies to trilobites. Here's a figure from the paper showing a comparison:

Image C shows the eye from an extinct species of trilobite, while Image D shows the eye from a fruit fly. Both the real and artificial eyes offer a horizontal field of view of 180 degrees, and they consist of a similar number of pixels. However, the artificial eye is significantly faster, operating at up to 300 hertz, while a fruit fly only updates at 100 hertz. Take that, nature!

It's important to note that these aren't the sort of cameras that you'd want to use to take pictures. What they're best at is sensing movement, or to be more specific, sensing changes in the intensity of light generated by motion. It doesn't sound like much (and it doesn't look like much, either), but it's how bugs navigate and avoid obstacles, and as anyone who's ever tried to swat a fly can attest to, it works rather well. It also works indoors, outdoors, in bright sun, and in shade (or even moonlight), and has no trouble adapting to abrupt transitions between any of these states, which is something that conventional cameras are lousy at.

Here's a video of the system in action, which gives a good idea of what an optic flow sensor "sees" and why it works so well for insects:

Clearly, there are many advantages that these kinds of sensors can bring to robotics, especially in the context of lightweight aerial platforms. And there's no reason to stop at just one of 'em, either: put a couple CurvACE sensors together and all of a sudden you have a 360 degree panoramic sensor system the size of a couple of quarters:

Freaky. And awesome.

Going forward, we expect to see this tech integrated into robots, and there's also the potential for it to show up in lots of other applications. If it can be made inexpensively enough, we could end up with something like "imaging tape" that could be integrated into smart clothing, to provide a self-contained way of detecting distances to objects. Why? Because why not, that's why. But seriously, flexible sensors like these have the potential to enable all sorts of new applications that have been impossible until now, and we're excited to see what happens.

"Miniature Curved Artificial Compound Eyes," by Dario Floreano, Ramon Pericet-Camara, Stéphane Viollet, Franck Ruffier, Andreas Brückner, Robert Leitel, Wolfgang Buss, Mohsine Menouni, Fabien Expert, Raphaël Juston, Michal Karol Dobrzynski, Geraud L’Eplattenier, Fabian Recktenwald, Hanspeter A. Mallot, and Nicolas Franceschini, was published today in Proceedings of National Academy of Sciences.

Special thanks to Dario Floreano and Ramon Pericet-Camara for showing us these eyes at EPFL.

[ CurvACE ]

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman

“I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

This article is part of our special report on AI, “The Great AI Reckoning.”

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less