Flying microrobot takes steps toward full autonomy (w/ video)

September 20, 2011 by Lisa Zyga, feature

This screenshot from the video below shows the flying microrobot performing vertical flight with closed-loop control. Image credit: Néstor O. Pérez-Arancibia, et al. ©2011 IOP Publishing Ltd
( -- With the goal of designing an insect-inspired flying microrobot capable of sustained autonomous flight, researchers have demonstrated for the first time a microrobot that achieves vertical flight using closed-loop control. The researchers predict that the approach they use for controlling flight on this one axis could also be used for controlling flight on all three axes.

The team of researchers, Dr. Néstor Pérez-Arancibia, Kevin Ma, Dr. Kevin Galloway, Jack Greenberg, and Prof. Robert Wood from the Harvard Microrobotics Lab at Harvard University, has published their study on the first controlled vertical flight of a biologically inspired microrobot in a recent issue of Bioinspiration & Biomimetics. The methods they used could provide a key step toward developing completely autonomous flying microrobots.

“Basically, a fully autonomous flying microrobot would do similar things to what natural bees and flies can do: take off, land, and navigate through difficult environments,” Pérez-Arancibia told “In the long term, I can also envision microrobots that can adapt to their environments, coordinate with other robots to accomplish difficult tasks, and interact with natural insects (this would be very cool, I think).”

As the researchers explained in their study, designing a microrobot with total autonomy is a complex problem for which aerodynamics, sensing, actuators, and other factors must be considered simultaneously. To tackle the problem, the researchers focused on just one degree of freedom: altitude.

The hovering microrobot can reject disturbances such as compressed air jets from a small hose. The ruler shown in the video is a rough visual reference; precise altitude is measured with the laser displacement sensor. Video credit: Néstor O. Pérez-Arancibia, et al.

After designing and fabricating a 56-mg flapping-wing microrobot, the researchers attached the microrobot to a double-cantilever beam that could move only in the vertical direction. When the robot flapped its wings, the flapping induced inertial and aerodynamic forces that caused the wings to passively rotate. In turn, the passive rotation created a non-zero angle of attack during the wing stroke, which produced lift. In general, the faster the wings flapped (i.e., the higher the frequency and/or the amplitude of their stroke angle), the greater the lift force.

As Pérez-Arancibia explained, to design the controller – the set of rules used to generate input to the system – the researchers used two separate experimental setups. In the first experimental setup, the researchers gathered relevant information about the robot’s dynamics by performing static experiments (i.e., the robot flaps, but it does not move). Two sensors measured the actuator output and the force produced by the flapping wings. With this information, the researchers could determine the controller’s general structure. In order to fine-tune this structure, the researchers then computed additional parameters by performing experiments in which the robot moves up and down, as shown in the video. This controller resulted in the first demonstration of the closed-loop control of an insect-scale robot.

“The term ‘closed-loop’ implies that feedback is used to generate the input to the robotic system,” Pérez-Arancibia explained. “In this particular case, the robotʼs altitude is measured using an external laser position sensor and then this information is used by the controller (a set of rules) to generate the control signal, which is the voltage input to the system that makes the wings flap. Note that altitude is the variable that we control (we make it follow the trajectory we desire, using a feedback controller).”

Using this controller design process, the researchers demonstrated that the microrobot could perform tasks such as hovering in one place and following a trajectory. Also, when the researchers caused a disturbance by blowing air from a hose at the microrobot, the microrobot was able to withstand the disturbance. In addition, the 56-mg is capable of generating lift forces of up to 3.6 times its own weight, meaning it could carry a payload including steering components, sensors, and power sources. These characteristics could make the flying microrobots appealing for a variety of applications.

“It is reasonable to assume that they will be cheap if manufactured at a large scale,” Pérez-Arancibia said. “Therefore, they will be excellent for going into, exploring, and sending information out from areas inaccessible or too dangerous to humans. I am thinking of buildings on fire, contaminated areas (by dangerous chemicals, radiation, or even pathogens), collapsed structures, etc. They will be excellent to do field biological research. Imagine the use of 1000 or 2000 of these robots for exploring a tree in the middle of the Amazonian forest, for example. Imagine a group of microrobots flying alongside Monarch butterflies on their migration route from Canada to Mexico, etc. Another application often mentioned is artificial pollination.”

As Pérez-Arancibia added, the researchers are making swift progress toward the ultimate goal of fully autonomous flying microrobots.

“At the Harvard Microrobotics Lab, as a group, we are and will be working on bringing you autonomous flying microrobots ASAP,” he said. “Papers coming soon will present experiments on altitude control using optical flow, on pitch control, on new robotic designs that include steering and control actuators, etc. Stay tuned!”

Explore further: Cracking flight's mysteries: It's a bird, it's a plane, it's a microrobot

More information: Néstor O. Pérez-Arancibia, et al. “First controlled vertical flight of a biologically inspired microrobot.” Bioinsp. Biomim. 6 (2011) 036009 (11pp). DOI:10.1088/1748-3182/6/3/036009


Related Stories

World's Lightest Micro-Flying Robot Built by Epson

August 19, 2004

Seiko Epson Corporation ("Epson") today announced that it has successfully developed a lighter and more advanced successor to the FR, the world's smallest and lightest micro-flying robot. Turning once again to its micromechatronics ...

Cyclogyro Flying Robot Improves its Angles of Attack

January 22, 2009

( -- In the past few decades, researchers have been investigating a variety of flying machines. Most studies have focused on improving the flying performance of standard flying mechanisms, rather than developing ...

Recommended for you

Matter waves and quantum splinters

March 25, 2019

Physicists in the United States, Austria and Brazil have shown that shaking ultracold Bose-Einstein condensates (BECs) can cause them to either divide into uniform segments or shatter into unpredictable splinters, depending ...

How tree diversity regulates invading forest pests

March 25, 2019

A national-scale study of U.S. forests found strong relationships between the diversity of native tree species and the number of nonnative pests that pose economic and ecological threats to the nation's forests.


Adjust slider to filter visible comments by rank

Display comments: newest first

5 / 5 (3) Sep 20, 2011
Sweet , we'll probably need help with crop pollinations in the future. I for one would love to see the equivalent of agricultural predator drones.
1.8 / 5 (4) Sep 20, 2011
so a feedback system component, the algorithm, exists, but is based on an input from a laser attached to its own hardware.

so that hardware is going to one day be useful and mounteable on the actual hoverdrone?
i don't think so.

i would have rather seen this guy focuss designing a micro-mounted position awareness system on the drone that demonstrates a capability for the drone to know roughly where it is , based on where it was and where it's going. ....and then design a feedback loop that perfects this system .

all you have now is a motion capability (distrubance rejection) based on off-drone hardware that stands a poor chance of being adapted for placement on the ultralight drone.

i just have a hard time , from a program level where one hypothetically directs the development of the entire drone , seeing why it is useful to develop motion capabilities, when the position knowledge capability is not there; the former is useless without the latter.
this is piecemeal devlepment
not rated yet Sep 20, 2011
progress in science and engineering often proceeds in small steps. This is one step.
1 / 5 (1) Sep 20, 2011
They did not know the transfer function of the system in regard to lift and now they have one accurate enough to control the altitude of the unit in the presence of external noise inputs. More than likely their next step will be to understand how to incorporate backward/forward motion and lift together along with attitude control. Then they will likely incorporate side to side motion and rotation. At that point they will have a functional flight control system that can accept vectors and distances as command inputs.

Once you have that, a higher level path control system that gets feedback from position sensors will calculate the positional error from the desired path in the form of an error vector that it will use to command the flight control system. This will give path control capability.

A higher level path planning system can then provide flight path data to the path control system to execute missions and/or incorporate positional/situational awareness/adaptation.
1 / 5 (3) Sep 20, 2011
I wonder, might it be more practical to build a control system for an insect? ...a "cyborg" insect, so to speak.

"Resistance is futile. You will be assimilated." LOL
5 / 5 (1) Sep 21, 2011
I wonder, might it be more practical to build a control system for an insect? ...a "cyborg" insect, so to speak.

Already done.
1 / 5 (2) Sep 26, 2011
Some similar and plagiaristic research at Northwestern University:
Jack Peters
not rated yet Sep 30, 2011
After spending some time with the author's publications, I believe there is no new designed controller. Rather, its a new application to the microrobots. The feedforward controller (called Adaptive Law in ref 10) was developed at UCLA (Jiang 1995). It seems author used the scheme over and over with different applications.

The controller plays a key role in this application. Did the author give UCLA credit?

J Peters

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.