Weighing just 80 milligrams, the previous version of RoboBee was so light that it had trouble breaking the surface tension of the water, in order to go underwater. This issue was addressed by basically just shutting off its power as it hovered above the water – that power came from an external source by way of a wire tether – and allowing it to plummet down and crash through the surface.

From there, it swam through the water column simply by flapping its wings (albeit at a much slower rate than it did when flying, as water is 1,000 times denser than air). When it came time to return to the air, however, it couldn't simply repeat the crashing action in an upwards direction. Instead, it remained trapped at the surface, held in place by the surface tension.

Having the wings out of the water still isn't enough, though. Once the RoboBee surfaces, electrolytic plates in the chamber convert collected water into combustible oxyhydrogen gas. A sparker inside the chamber then ignites the gas, propelling the robot up off of the water. Unfortunately it can't resume sustained flight once it gets back in the air – at least not yet – although it is designed to stabilize itself, so that it will land on its feet when it comes back down.

"By modifying the vehicle design, we are now able to lift more than three times the payload of the previous RoboBee," says Yufeng Chen, first author of a paper on the research. "This additional payload capacity allowed us to carry the additional devices including the gas chamber, the electrolytic plates, sparker, and buoyant outriggers, bringing the total weight of the hybrid robot to 175 milligrams, about 90 mg heavier than previous designs. We hope that our work investigating tradeoffs like weight and surface tension can inspire future multi-functional microrobots – ones that can move on complex terrains and perform a variety of tasks."