The July 2022 issue of IEEE Spectrum is here!

Close bar

Harvard's Robot Bee Is Now Also a Submarine

Without any hardware modifications, the Harvard RoboBee learns to land in the water and go for a swim

3 min read
Harvard's Robot Bee Is Now Also a Submarine
Gone swimmin'.
Image: Wyss Institute/Harvard University

For the last several years, Harvard has been developing a robot bee. They’ve done some impressive work: their sub-paper-clip-sized, 100-milligram flapping-wing micro aerial vehicle is fully controllable down to a stable autonomous hover. It’s still tethered for power, and there’s no onboard autonomous control, but the robot flaps its wings and flies like an insect, which is awesome.

Tiny robotic bugs have lots of potential for search and rescue, surveillance, and exploration, but what’s been all the rage recently is adaptive multi-modal robotics: robots thatcan creativelyhandle a combinationof terrains, making them much more versatile. With some exceptions, robots are usually pretty bad at this, and with some exceptions, humans and animals are too. There are ground robots that can handle water, and a few flying robots that aren’t totally helpless on the ground, but so far, we haven’t seen much in the way of flying robots that are good swimmers. 

Yesterday at IROS, Harvard researchers presented a paper describing how they managed to get their robotic bee to swim, which I’m pretty sure is not a thing that even real bees are known for doing. With no hardware modifications at all, Harvard’s RoboBee can fly through the air, crash land in the water, and turn into a little submarine. You know what that means: nowhere is safe from robot bees.

imgHarvard’s RoboBee.Photo: Wyss Institute/Harvard University

Things to keep in mind about these videos: RoboBee is small enough to sit on the tip of your finger, and light enough that you’d barely feel it if it was. When it flies (or swims), it’s doing so under full control: a motion capture system tracks its position, and sends trajectory commands to the robot. This works in both air and water, and RoboBee’s method of entry (a pitch over, dive, crash, and sink) is deliberate. Also, the hovering at the beginning of the first video below looks a little bit wonky, but that’s because RoboBee still has some water on its wings from previous tests (a more stable hover is shown in that same video, after the diving sequence, and the second video below has more details about the experiments).

The key realization here is that swimming is actually a lot like flying: in both cases, you’re trying to propel yourself through a fluid by moving a wing (or fin) back and forth. To fly (and particularly to hover) you need to do this very quickly, but to swim, it’s a much more relaxed motion. It’s fundamentally the same motion, though, and you can achieve it with the same basic hardware. In the case of RoboBee, to fly in air it flaps its wings at 120 Hz, while to swim in water it flaps its wings at just 9 Hz. Otherwise, three axis torque control is very similar, meaning that the robot can be steered around in the water, too.


One unique problem that RoboBee has with the water entry is that it’s so small that the surface tension of the water is enough to keep it from submerging. This is part of the reason that it has to crash land in water [right] (it also needs to have its wings treated with a surfactant to help it sink). A fully loaded RoboBee (with a battery on board) might be heavy enough to avoid this problem, but at this point, it’s still an issue. Also still an issue is the whole water-air transition, which seems like it’s significantly more difficult than going from air to water, but we’ve been assured that the researchers will be tackling this in future work.

“Hybrid Aerial and Aquatic Locomotion in an At-Scale Robotic Insect,” by Yufeng Chen, E. Farrell Helbling, Nick Gravish, Kevin Ma, and Robert J. Wood from the Wyss Institute for Biologically Inspired Engineering at Harvard University was presented this week at IROS 2015 in Hamburg, Germany.

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