Small, insect-like robot can now latch on to overhangs

In 2013, a group at Harvard's Wyss Institute for Biologically Inspired Engineering brought miniaturization to the world of drones, creating a tiny robot that could fly using rapidly beating wings. Now, after adding a handful of team members from other institutions, Robert Woods' team is back with a paper that gives an update to the group's little flying machines—one that lets the robots hang upside down like bats.

This might at first seem like a frivolous addition (although doing something because it's pretty cool can be a major motivator for cutting-edge engineering). But the researchers have some pretty solid reasons for adding the feature. One of the main reasons you'd build a drone, tiny or otherwise, is to be able to look down on an area from a high vantage point. A big limitation of this approach is that getting to and staying in that high vantage point takes energy.

If there's a way to latch on to something and sit there, it could provide big energy savings, which could allow the drone to monitor an area for much longer than it would otherwise be able to. In the case of a miniature drones, the authors have latch points like "trees, buildings, or powerlines."

People have built robots that perch on surfaces, but the surface has generally needed to be prepared in advance or covered in something from a limited range of materials. The authors wanted to make a device that would work on the broad range of surfaces that you find in a normal (if human-dominated) environment. Checking previous literature on the subject, they found that people were able to build robots that could climb walls using an electrostatic attachment.

The principle is pretty simple. If you build up a positive charge on your robot's foot, it will draw electrons in the neighboring material toward it. The positive and negative charges would then attract each other, keeping everything in place.

The researchers tested this idea by running wires underneath a thin polymer pad. When the positive and negative wires were given a 1,000 volt difference, enough force was generated to potentially attach one of the lightweight drones. For a small device, that voltage difference could be generated with as little as seven microAmps, easily fitting within the drone's power budget. Leakage of charges between the wiring and the surface meant that the device required 6.9µW or less energy to remain stuck on the surface—that's over 1,000-fold less than flying requires, so this approach really does save energy.

But the adhesion was dependent upon a variety of environmental factors, such as the surface material and its cleanliness, the relative humidity, and whether dirt and dust had gotten on to the device. The authors figured that they should just add a version onto their mini-drone and try it out. They built a small, lightweight version of the device and attached it to the top of their robotic flying machine.

To add the ability to roost, the researchers created an algorithm modeled on how a bee slows down as it approaches a surface to land on. (The bee slows down so that whatever it's planning to land on appears to expand at a constant rate.) This tuning turned out to be pretty challenging, because the aerodynamics of the flapping wings interacted with the air flow near the surface in a way that caused the machine to gain lift as it approached—a boost of 40 percent.

Not confident that they could always control for these sudden changes, the researchers simply inserted a small bit of flexible foam between the robot and the disk that would attach to the surface, meaning that the robot didn't need to get its approach perfectly right in order for everything to work.

The researchers found that the robot could successfully land on surfaces like glass, plywood, and the leaf of a living plant. Once there, energy use dropped as expected. Once the charge difference was removed, the robot would easily detach and could restart normal flight. They repeated the process multiple times as their landing attachment picked up dust from the environment. While the attachment strength dropped by nearly 30 percent over time, the electrostatic grip was still strong enough to keep working.

If you're the paranoid type, you might be rushing to check your windows for mini-drones, but we're not anywhere close to the stage where these devices are ready for real-world use. For now, all the processing has to be done on computers that are connected to the drones with thin, flexible wires. The robots can also only fly in a pre-made "arena" where cameras can track their location and orientation in order to update flight commands.

Hopefully, if Woods' group ever gets its flying machines to work beyond the confines of this arena, it will release another paper letting us know first.