Penny-Sized Ionocraft Flies With No Moving Parts

Image: UC Berkeley Insect-scale flying robots are usually designed to mimic biological insects, because biological insects are masters of efficient small-scale flying. These flapping-wing micro air vehicles (FMAVs) approach the size of real insects, and we’ve seen some impressive demonstrations of bee-size robots that can take off, hover, and even go for a swim. Making…

By , in world on .
A drone powered by electrohydrodynamic thrust is the smallest flying robot ever made
Image: UC Berkeley

Insect-scale flying robots are usually designed to mimic biological insects, because biological insects are masters of efficient small-scale flying. These flapping-wing micro air vehicles (FMAVs) approach the size of real insects, and we’ve seen some impressive demonstrations of bee-size robots that can take off, hover, and even go for a swim. Making a tiny robot with flapping wings that can move in all of the degrees of freedom necessary to keep it controllable is tricky, though, requiring complicated mechanical transmissions and complicated software as well.

It’s understandable why the biomimetic approach is the favored one—insects have had a couple hundred million years to work out all the kinks, and the other ways in which we’ve figured out how to get robots to fly under their own power (namely, propeller-based systems) don’t scale down to small sizes very well. But there’s another way to fly, and unlike wings or airfoils, it’s something that animals haven’t managed to come up with: electrohydrodynamic thrust, which requires no moving parts, just electricity.

Electrohydrodynamic (EHD) thrusters, sometimes called ion thrusters,* use a high strength electric field to generate a plasma of ionized air. The ions (mostly positively charged nitrogen molecules) are drawn toward a negatively charged grid, and along the way, they smack into neutral air molecules and impart momentum to them, which is where the EHD thrust comes from.

Schematic showing the cross-section of an emitter wire and collector grid electrode pair. Thrust is produced when positive ions, generated in the corona plasma region near the sharp tips of the emitter electrode, drift in the applied electric field and collide with neutral air molecules and impart momentum.
Image: UC Berkeley

This schematic shows the cross section of an emitter wire and collector grid electrode pair. Thrust is produced when positive ions, generated in the corona plasma region near the sharp tips of the emitter electrode, drift in the applied electric field and collide with neutral air molecules and impart momentum.

This isn’t a particularly new concept; the general phenomenon has been understood for a few hundred years, and for a while, folks thought that it might be possible to use it in manned aircraft, although you’d need a crazy big structure of emitters and collector grids to lift a person.

There was a mouse named Orville that took flight in an improbably large EHD aircraft back in 2003 (plenty of cute pictures here), but as far as we know, that’s about as fancy as large EHD-powered aircraft ever got. Going the other direction, though, could be where this technology becomes practical.

Flying microrobot based on ionocraft effect
Photo: UC Berkeley
UC Berkeley’s ionocraft is approximately 2 by 2 centimeters and has a mass of 30 milligrams.

This is the ionocraft, currently under development at UC Berkeley. It’s tiny— just 2 cm x 2 cm, weighing 30 mg, plus a 37-mg IMU (although power is supplied through a tether). At small scales, the lack of moving parts becomes a real asset, because you don’t have to worry about figuring out how to scale mechanical things like transmissions down beyond a point where it will be really frustrating at best and impossible at worst. Including its IMU payload, the ionocraft was able to take off and hover with an input of 2,000 volts at just under 0.35 mA:

Magical, right? No moving parts, completely silent, and it flies! Large scale EHD thrusters might be impractical, but scaling things down actually makes EHD thrusters better, since the electrostatic forces are scale invariant. That means smaller thrusters have a better thrust-to-weight ratio, as well as lower voltage requirements. And at small scales, an advantage that the ionocraft has over similarly sized FMAVs is that you can design a controller with a quadrotor as a starting point, as the ionocraft uses four thruster grids in a similar configuration. Since it doesn’t have rotating propellers, it can’t take advantage of angular momentum changes to yaw in place, but it turns out that “quick, repeated sequences of pitch and then roll” can result in a yawing motion, as long as you have a little bit of wiggle room. 

As with other micro air vehicles that we’ve seen, the big question now is whether we can look forward to autonomous operation with a useful payload. At the moment, the ionocraft is carrying a payload that weighs more than it does, but it also needs a total of seven external wires for power, data, and grounding. The UC Berkeley researchers are very explicit that they think autonomy is achievable, though:

Although initial demonstrations of controlled flight will certainly be tethered and in an indoor laboratory setting, we envision the future ionocraft as a fully autonomous robot capable of outdoor flight. The path towards autonomy requires making strides in a number of interdependent categories, including: system development to allow for incorporation of more onboard computation, control, and sensing electronics; further engineering refinement of the EHD actuator to provide increased thrust density and efficiency at a lower operating voltage; and further development of both the simulation environment and the bench-scale flight testing setup to begin development of more robust controllers with high level functionality.

The researchers suggest that 100 mg could be a reasonable payload for an ionocraft with an IMU, control ASIC, driver circuit, and optical flow sensor that could be capable of controlled autonomous flight with tethers only for power. The need for small-scale power is not unique to the ionocraft, and it’ll require “incorporation of high energy density research-grade batteries” that may not exist yet, plus some improvements to the actuator performance. The good news is that there’s a lot of room for improvements, and the researchers seem optimistic that they’ll be able to squeeze significantly more oomph out of their EHD thruster design. 

For more details, we spoke with lead author Daniel Drew via email.

IEEE Spectrum: Why haven’t we seen more robots utilizing this technology?

Daniel Drew: One reason is that, due to some unfortunate “bad science,” widespread understanding of the effect took much longer than it should have. Another is that there was no push for miniaturization of EHD in the realm of propulsion. Large electrode gaps require tens of thousands of volts, which typically requires very heavy DC-DC conversion electronics (e.g. for required inductance or dielectric strength). Combined with the low (for human-scale flight) aerial thrust of EHD, this made a system seem infeasible. Of course advances in power electronics have made successes possible, but still not yet a competitive technology at that scale. Our position is that EHD is particularly well suited for small (“micro”) robots due to the force being scale-invariant— that is, as a fundamentally electrostatic force, the output thrust can be held constant as system volume is decreased, resulting in much higher thrust-to-weight ratios. This has the added advantage of decreasing operation voltage and the required power electronics mass. 

Can you describe the process of building the ionocraft?

Although the ionocraft components are batch-fabricated, the robots themselves are assembled by hand. After a good deal of iteration I have arrived at an assembly process that takes me about 40 minutes per robot, with near 100% success rate. This relies on mechanical affordances and tolerances built into the components, a separately machined assembly jig, and the use of a vacuum wand. I found that the key for reliable assembly has been making sure that as much alignment as possible comes from the geometry of the components and that no complicated movements are required. The fact that I am almost entirely moving a vacuum wand in orthogonal axes during assembly means that I think the process could relatively easily be translated to a SMT-style pick-and-place robot. The skill aspect is largely in understanding how much force can be applied without fracturing the components (and in keeping a steady hand). 

UC Berkeley ionocraft being assembled
Photo: UC Berkeley
The ionocraft is built from 41 discrete components, each connected by a combination of mechanical slots and UV-curable epoxy.

How well would the ionocraft scale, either up or down? How small could it get, and what might potential applications be? 

I think at the centimeter scale we could have a fully controlled, autonomous ionocraft that performs as well as any biomimetic flier. I think at the meter scale the propulsion mechanism could be viable, but harder to motivate. I think one of the most interesting paths forward is actually in continued miniaturization. The mechanical simplicity of the mechanism makes it relatively easy to scale. There is no reliance on a lift generating airfoil, the performance of which is extremely Reynolds number dependent. In the future I think it’s possible to create millimeter scale EHD thrusters, based on ions emitted via field emission arrays, that operate under a hundred volts. The nearer term applications of such technology would be in wide-dispersal distributed sensing, but as we move towards full controllability at that scale we approach the dream of ubiquitous swarm interfaces/”programmable matter”.

What will have to happen for the ionocraft to fly untethered? 

Energy autonomy will rely on cutting-edge thin-film battery or thin-film supercapacitor technology to reach the required energy and power densities. Given published densities, it actually seems to make sense to throttle back our EHD thrust in the interest of increasing thrust efficiency (the former scales with E^2, the later inversely with E). At the current levels of actuator miniaturization we actually don’t have a great ability to make this trade-off; based on the ratio of emitter radius to electrode gap, at the time of corona discharge inception the drift field is set at a high minimum value. We need “sharper” emitters, which will decrease inception voltage and increase the potential thrust efficiency operating point. I believe that near term gains are possible with some processing improvements, but long term maybe we move towards something like ZnO nanowires, silicon nanowires, or CNTs as emission points.

What are some potential applications for a robot like this? Would it be good for planetary exploration, and if so, how would it work on a planet with a thicker atmosphere like Venus or Titan?

Insect-scale flying robots will one day be used in applications ranging from distributed air quality and chemical sensing, to precision agriculture, to search and rescue. If we move towards a model of robot swarms as future interstellar explorers (e.g. Starshot) then I think EHD would be an interested propulsion mechanism. It’s relatively simple to incorporate on chip and, maybe combined with some clever physical structures, could help with wide area dispersal even in the absence of full autonomy. 

In terms of what atmospheres make sense, a first order statement would be that, as you allude to, the thicker the atmosphere the better as ion mobility will decrease and yield a higher force output. There is not much literature that I have found on corona discharge based ionization in gas mixtures besides air, pure N2, or pure O2 at near STP. I would be very interested in assessing EHD in conditions approximating those found on other planets/moons— my silicon-based electrodes could at least withstand the temperatures of Venus! This seems like a great future collaboration.


We’re told that controlled flight using onboard sensors is likely achievable within the next few months, and that a complete system on a chip is currently under development. UC Berkeley will be collaborating with researchers at Dartmouth on DC-DC conversion, so that when batteries catch up to where aerial picorobots need them to be, the ionocraft will be able to get the high voltages it needs using onboard electronics. Putting all of this together means that we could see an autonomous ionocraft tethered only for power flying sometime in 2019.

“Toward Controlled Flight of the Ionocraft: A Flying Microrobot Using Electrohydrodynamic Thrust With Onboard Sensing and No Moving Parts,” by Daniel S. Drew, Nathan O. Lambert , Craig B. Schindler , and Kristofer S. J. Pister from UC Berkeley, was presented at IROS 2018 in Madrid, Spain.

* This is a bit of a misnomer, since it implies that most of the thrust is coming from the ions, which is not quite true—it’s the air molecules that provide the majority of the momentum transfer. You’ll find actual ion thrusters on spacecraft, which use ionized liquids or gases as reaction mass, and charged grids to accelerate them.

Recommended articles