Conference PaperPDF Available

Liftoff of a 190 mg Laser-Powered Aerial Vehicle: The Lightest Wireless Robot to Fly

Authors:
Liftoff of a 190 mg Laser-Powered Aerial Vehicle:
The Lightest Wireless Robot to Fly
Johannes James1, Vikram Iyer3, Yogesh Chukewad1, Shyamnath Gollakota2, Sawyer B. Fuller1
Abstract To date, insect scale aerial robots have required
wire tethers for providing power due to the challenges of
integrating the required high-voltage power electronics within
their severely constrained weight budgets. In this paper we
present a significant milestone in the achievement of flight
autonomy: the first wireless liftoff of a 190 mg aerial vehicle.
Our robot is remotely powered using a 976 nm laser and
integrates a complete power electronics package weighing a
total of 104 mg, using commercially available components
and fabricated using a fast-turnaround laser based circuit
fabrication technique. The onboard electronics include a light-
weight boost converter capable of producing high voltage bias
and drive signals of over 200 V at up to 170 Hz and regulated
by a microcontroller performing feedback control. We present
our system design and analysis, detailed description of our
fabrication method, and results from flight experiments.
I. INTRODUCTION
Honeybee-sized insect-scale aerial robots (100 mg) are
well suited to a variety of applications benefiting from their
small scale including environmental monitoring, agricultural
support, and search and rescue. Since they were originally
proposed as “gnat robots” in 1989 by Brooks and Flynn [1]
and attempted in earnest by the Berkeley Micro Robotic
Fly project starting in the early 2000s [2], progress toward
truly autonomous insect scale robots has seen important
milestones. These include the first lift greater than weight of
a 100 mg robot [3], subsequent controlled flight [4], sensor
integration [5], and expanded capabilities such as landing [6].
However, in the decade that has followed first liftoff, not one
of these 100 mg robots has been able to fly without tiny wires
to power and control it.
Realizing wireless flight requires solving three key chal-
lenges that arise from the small scale:
Insect scale at <200 mg discourages traditional forms
of propulsion such as a propeller driven by electro-
magnetic coils because unfavorable physics scaling [7].
Instead, flapping wings driven by piezo-electric actu-
ators are more efficient [8]. While piezo-driven robots
have been successfully used for flight, they require high
potentials over 200 V [3]. Generating the necessary
voltage signals has so far required large electronic
components with a prohibitive weight relative to insect-
scale aerial payload capacity (e.g. [9] at 675 mg).
Wireless flight requires an energy source to power the
electronic and mechanical components. To date, the
1Department of Mechanical Engineering, 2Paul G. Allen School of
Computer Science and Engineering, 3Department of Electrical Engineering,
University of Washington, Seattle, WA 98195. jmjames@uw.edu
Fig. 1: (Top) The 190 mg RoboFly and power system
before liftoff. (Bottom) After the laser is powered on, power
reaches the robot through photovoltaic cell at top. Onboard
electronics generate the waveform to drive the wings, causing
the robot to lift off. After liftoff the robot is no longer in
contact with its reflection on the surface below.
smallest high-drain (>10 C) batteries available are too
heavy at 350 mg (GM300910, PowerStream Technol-
ogy, Orem, Utah). The only currently viable alternative
is a battery-free design.
Finally, all the required digital processing has to be
performed on the aircraft. Onboard computation that
operates within the size, weight and power (SWaP)
requirements is not only necessary to control the elec-
tronics and piezo driver, but also a basic requirement
for truly autonomous insect robots capable of sensing
and more complex functionality.
This paper demonstrates the lightest wireless robotic flight
to date by showing the liftoff of a sub-200 mg aerial
vehicle. To achieve this we introduce three key technical
innovations. First, we present a novel ultra-lightweight and
fast-turnaround circuit fabrication technique with which we
Fig. 2: Circuit schematic showing the complete power electronics system. The boost converter produces a high voltage bias
(red) and the driver uses this to produce high voltage sinusoid. The boost converter and driver are controlled by PWM
signals (blue) from the microcontroller.
create the first sub 100 mg boost converter and piezo driver
that is integrated into an aerial robot. Second, we present a
battery-free design by demonstrating the first wireless power
transmission to an insect scale aerial robot at ranges of
over 1m using photovoltaic cells and lasers. Third, we
demonstrate the first insect-scale aerial vehicle with onboard
computation by integrating a light-weight microcontroller
that we use to control the boost converter and piezo driver.
Finally, we integrate all of these electronic components
onboard an insect robot, constituting a 104 mg package,
which is less than the weight of a typical toothpick. We use
them to perform the first physically untethered flights of an
insect-scale robot, weighing 190 mg altogether.
II. SYSTEM OVERVIEW
Our bio-inspired insect robot consists of a dual wing flap-
ping design driven by two piezoelectric actuators. Our full
system capable of wireless takeoff begins with a laser source,
which delivers a constant source of power wirelessly to the
robot. A photovoltaic cell then converts the optical power
to electrical power. The power provided by the laser is used
to run the boost converter, driver, and microcontroller which
produce sinusoidal voltage outputs capable of simultaneously
driving the two piezo-electric actuators.
In the rest of the paper, we first describe our flight-weight
power electronics which include our ultra light-weight boost
converter and driver design, followed by a description of
our rapid fabrication methods for producing the flight weight
circuit and integrating an onboard microcontroller. Next we
describe our laser system and the design choices for wireless
power transfer. Finally, we present implementation details for
our robofly followed by flight results.
III. FLIGHT-WEIGHT POWER ELE CT RO NIC S
The oscillating motion of the bimorph piezo beam actu-
ators that flap the wings must be driven by a sinusoidal,
high-voltage signal. This must be in the range of 200300 V
to maximize the power density produced by the actuator. The
need for low weight and high efficiency strongly influence
our design.
Efficiency would be improved if the actuator and wing
assembly operated at both electrical and mechanical reso-
nance. However, because the capacitance of the actuators
is approximately 5nF, an inductance needed to achieve
electrical resonance at the flapping frequencies would be
prohibitively heavy. Therefore, a design goal is that the
sine wave should operate at a user-programmable frequency
near the mechanical natural frequency of the actuator-wing
system, 140170 Hz [10].
We employ a design geared to bimorph actuators consist-
ing of a constant, high-voltage bias signal and a sinusoidally
varying signal channel, following the approach of [11]. An
example signal appropriate for this configuration is shown at
the top of Fig. 3.
A. Circuit design
Commercially available piezo driver ICs (e.g. Texas In-
struments DRV2700) cannot produce the required voltage.
Commercially-available integrated solutions such as the
PD100 (Piezo Drive, Callaghan, Australia) are too heavy
at 500 mg. While a monolithic SoC design is the typical
solution for reducing the size and weight of electronics, we
instead focus on a simple switched mode design built with
off-the-shelf components.
This approach has a number of advantages. First, it allows
for greater design flexibility and rapid prototyping which is
important considering that insect robots are still an active
area of research with frequent design changes. For example,
optimizing an IC for a particular actuator design precludes
further improvement in that domain or later incorporation
of additional features such as energy recovery mechanisms.
Second, designing a single SoC solution that integrates
the high voltage actuator drive circuitry presents a design
tradeoff. Digital circuits for processing can take advantage
of device scaling to operate at low power and reduce size,
however these devices cannot tolerant high voltages needed
Fig. 3: (Top) Example target waveforms for boost converter
output (yellow) and sinusoidal driver output signal(black);
(Bottom) Example of driver pulse train varying by pulse
frequency (PFM)
for the drive electronics. We instead choose to use a com-
mercially available ARM microcontroller (STM32F051) to
implement the timing and control which allows us to leverage
the plethora of commercial products which are thoroughly
tested and highly optimized for low power and size.
A schematic of our boost converter is shown in Fig. 2.
The switching mode boost converter switches electrical cur-
rent through a coupled inductor with a high turns ratio at
frequencies above 100 kHz [11]. The switch control signal
is generated by the pulse width modulation (PWM) output of
a microcontroller and connected to the gate of the MOSFET
M1 in Fig. 2. Current through the primary winding of the
coupled inductors stores energy in the magnetic field which
is transferred to the secondary winding. Brief high voltage
pulses on the output of the secondary winding after the
MOSFET switches rapidly to nonconducting state are recti-
fied through a fast diode. The diode’s output charges a high
voltage capacitor for storage and this output is connected to
the load. The load in this case is the driver circuit, which
linearly regulates the center node of the bimorph actuator in
order to drive sinusoidal displacement. The bias and driver
waveforms seen in Fig. 4 are connected directly to the top
piezo surface and the carbon fiber layer of both the bimorph
actuators respectively.
The driver circuit is designed to source or sink current at
a sinusoidal rate to the center node of the bimorph actuator
(a) Driver with 100 nF capacitor
(b) Driver with 660 pF capacitor
Fig. 4: Waveform output by onboard driver effecting modest
amplitude sinusoid of controllable frequency by high-side
and low-side control signals with a 100 nF capacitor on the
bias rail in 4a and a 660 pF capacitor in 4b.
in simultaneous drive configuration. Transistors Q1 and Q2
are configured as a two stage amplifier designed to source
current from the bias rail, implementing the “high side” to
increase the sinusoid to its maximum voltage. Transistor Q3
generates the “low side” of the waveform by sinking current
from the center node of each actuator to ground.
We use bipolar junction transistors (BJTs) as opposed
to FETS standard in current aerial microrobotic research
[12] simply due to the ability to tolerate higher voltages
than FETs in somewhat smaller commercially available SMT
packages, and simple gate biasing design for both linear
operation and the rapid pulsing required for potential use of
inductive energy recovery schemes [11]. Since the required
sinusoid is at a low frequency compared to the clocks
of microcontrollers, we generated the sinusoid using pulse
width modulation, as depicted in Fig. 3. Since the micro-
controller clock frequency is a significant factor in its power
consumption, the frequency of the pulse width modulation
was selected to generate a sufficiently smooth sinusoid with
adequate PWM resolution, without excessively high internal
oscillator frequency. Future work will investigate whether the
DACs built into microcontrollers, PSoCs, FPGAs, or even
passive oscillator circuits could generate the waveform in a
way that improved efficiency.
Because actuator displacement depends on the voltage
difference between the piezo layers and the center carbon
fiber layer, our goal is to maintain a constant high voltage
bias while the sinusoid varies over time. Dynamic common
mode control of the bias rail as in [12] is desirable but must
be conducted carefully so as to achieve the correct sinusoidal
driver output which is effectively equal to Vbias Vsignal.
The complications of common mode control are evident
in Fig. 4b Because the actuator load varies dynamically
during flight and the input power source may be unstable,
we design a feedback controller to help regulate a constant
bias voltage. We use a simple resistive voltage divider to
reduce the bias voltage to within the 3 V operating range of
the microcontroller and use its ADC to digitize the value.
Based on the ADC reading we adjust the duty cycle using a
basic proportional controller in addition to feedforward terms
anticipating dynamic load increases of the driver throughout
the low frequency actuator cycle.
B. Circuit Performance
Fig. 4 shows the high voltage output waveforms generated
by open-loop boost converter and driver for different storage
capacitor values. As expected, a large 100 nF capacitor as
shown in Fig. 4a produces a very consistent bias output
thereby reducing the need for feedback control to maxi-
mize wing displacement. In contrast, the waveform with
a smaller 660 pF capacitor varies noticeably, challenging
feedback control and complicating the driving of consistent
waveforms. While a large capacitor reduces bias variation it
comes at a cost of 16 mg of weight. These waveforms also
demonstrate two other important results. First, in both cases
the circuit produces an amplitude greater than 170 V which
we verified experimentally is the minimum required to lift
our MAV. Second, the sinusoid waveform is smooth and does
not have abrupt discontinuities that could potentially stress
and damage the actuator.
IV. CIRCUIT FABRICATION METHOD
We fabricated the circuit described above with the smallest
commercially available packages for each component with
required characteristics. Although coupled inductors such
as available in the Coilcraft LPR3015 series are similar to
the needs of the boost converter, full control of component
characteristics within a lower weight budget was obtained
by custom manufacture in-house without great difficulty.
Therefore, a coupled inductor with the required ratio of
turns, inductance, winding series resistance, coupling, and
total weight was custom built for the application and guided
by simulation in order to optimize for operating conditions.
We fabricated a custom inductor by winding 43 AWG wire
around the ferrite core removed from an LPD3015 inductor
for the primary side, and 46 AWG wire for the secondary.
The wire used was selected due to availability, resistivity,
insulation characteristics, and ease of winding for good
magnetic coupling. Total component weight of the simple
flyback transformer is 21 mg.
In addition to the size and weight of the components
themselves, the circuit board and copper traces on which they
are mounted contribute weight as well. Traditional PCB ma-
terials such as copper-clad FR4 have a density of 2.6 g/cm3
which is not feasible for insect-scale applications. We instead
developed a new rapid prototyping process for fabricating
ultra light-weight circuits requiring no chemical etching. This
process is an alternative to existing copper ablation flexible
PCB fabrication techniques which are expensive and require
care to ablate only the copper while leaving the substrate
intact. Our process is inspired by the laser micro-machining
methods used to fabricate the other parts of the insect robot,
and uses the same equipment.
Fig. 5 outlines our fabrication process. We begin by
cleaning both sides of a sheet of 25 µm copper foil with
isopropanol and placing it on a low-tack adhesive (Gelpak
X8, Hayward, CA). Next, we use the same UV DPSS laser
micromachining system used for fabricating the actuators and
body of the insect robot to cut out the desired copper traces.
The 20 µm spot size of the laser has enough resolution for
even the finest pitch electronic components.
After cutting, a low-power cleaning raster is performed
to achieve better adhesion. We then peel the excess copper
that can be peeled off of the Gelpak leaving only the desired
pattern. Next, we place a piece of readily available 25 µm
thick Kapton tape onto the copper and lift the traces off
of the Gelpak. The result is a flexible circuit marginally
thicker than 50 µm but still approximately 5-7 mg/cm2for a
typical circuit design. We select Kapton tape as the substrate
material due to its ability to withstand high temperatures
needed for soldering. Thinner Kapton tape and copper sheet
can be obtained, but is not as readily available. Such circuits
are vulnerable to contamination at sites of exposed adhesive
and are generally not as durable, but have in praxis survived
repeated rework and handling in research applications.
The final step is to populate the circuit with components.
While this can easily be done with a normal soldering
iron for most components, the lightest weight microcon-
trollers available in wafer level chip scale (WLCSP) packages
present a challenge. Because our circuit board only has
a single side and used no soldermask or insulating layer
to minimize weight, traces will short the contacts on the
interior parts of the chip without care to avoid this. A simple
method of addressing this is to precut holes in the kapton
tape at the desired solder ball sites and align those holes
to the circuit in the adhesion step. The chip is then aligned
and placed on the reverse side and soldered at the desired
contact points through the precut holes. Alternatively the
same micro-machining method can be used to to cut an
additional insulating layer of kapton that can be placed as
a mask over the chip and allow the use of normal reflow
soldering methods. The power electronics unit (PEU) of our
insect robot was constructed in this fashion and the results
at different points in the process can be seen in Fig.6.
V. LASER POWER TRANSFER
Achieving wireless liftoff requires powering all the above
components. Our robot requires 200-300 mW of power for
liftoff and requires a total of 25 mA of current.
The required energy density and peak current draw are
beyond the capabilities of existing battery technologies, but
a potential alternative is to use super-capacitors. A 7.5 mF
capacitor for example has a maximum voltage of 2.6 V [13].
This voltage however is insufficient to run the boost converter
even in simulation. A series parallel combination of 4 super-
capacitors could theoretically provide power for 250 ms of
flight at which point the capacitors would discharge from
Fig. 5: Steps of the circuit fabrication process beginning with laser micro-machining, followed by removing the excess
copper and adhering the desired traces to Kapton tape to produce an ultra-light weight flexible PCB.
Fig. 6: PEU at several stages in the circuit fabrication
process. Top left: bare unpopulated circuit. Top right: cir-
cuit populated with components including coupled inductor.
Bottom: Assembled PEU with boost converter and driver,
ready to mount to robot.
the total 5.2 V to below 4 V at which the boost converter
stops functioning in empirical evaluation. Perhaps even more
important than their inability to support sustained flight
though is their combined weight of 96 mg which is greater
than the weight of the entire boost converter.
A. Optical Wireless Power Transfer
Since on-board energy storage cannot meet the require-
ments for flight, we look to wireless power technologies
instead. A practical wireless power solution for an insect
scale robot must meet two criteria: 1) it must be able to
deliver the 250 mW of power required for flight, and 2)
it should have an operating range that allows for flight.
Near field magnetic induction can provide efficient power
delivery and have been demonstrated for walking robots [9],
however its range is fundamentally constrained to tens of
centimeters. Far-field microwave approaches (e.g., Wi-Fi)
can operate at longer ranges but suffers from efficiencies
less than 1% due to RF path loss [14]. We instead select an
optical approach as lasers provide a collimated beam with
high power density that can be harvested by photovoltaic
(PV) cells with conversion efficiencies of over 20%.
Our laser power delivery system consists of a 976 nm laser
source and a photovoltaic (PV) cell. For our laser source we
use the MHGoPower LSM-010 976 nm laser source capable
of providing 10 W of optical power. We connect the fiber
output to a collimator (Thor Labs F220FC-980) to produce a
beam in free space. An ideal laser should produce a perfect
collimated beam that does not diverge in space, however the
internal focusing optics of this laser and the use of multi-
mode fiber causes measurable beam divergence in space.
Unlike typical PV cells designed to harvest broad spec-
trum solar energy, our system should be optimized for
a single wavelength and high power densities. We there-
fore select a vertical multi-junction PV cell (MH GoPower
5S0303.4) [15] which consists of serially interconnected p-
n junctions bonded together to form a small PV array with
low series resistance that performs well under high intensity
light [15]. The PV cell measures 2.88 mm x 2.95 mm and
weighs 8 mg with an additional 5 mg of wires. This is well
within the size and weight constraints of our Robofly. We find
experimentally that at intensities up to 20 W/cm2the cells
achieve maximum power output when operating at 8.8 V
with efficiencies of up to 40% for short pulses.
While a power source for liftoff only requires a limited
range, we note that laser power beaming can be extended to
longer ranges. For example, our laser can deliver sufficient
power to the Robofly up to ranges of 1.23 m indoors. This
range is not fundamentally limited, but rather determined by
the beam divergence and output power of our specific laser
source. At ranges beyond 1.23 m, our beam expands to a
point where insufficient power is available over the small area
of cell. Thus, we can in principle achieve tens of meters of
range using commercially available lasers with higher output
power or a more collimated beam.
While lasers are capable of powering the Robofly, their
use raises other practical questions as well. First, to maintain
flight, the laser must track the Robofly. Although tracking is
beyond the scope of this work, potential solutions include
using motion capture systems as demonstrated in [4] for
robot control to track the position of the Robofly and direct
the laser appropriately using a device like a galvo mirror.
Additionally, we can simplify this problem by placing the
laser on a ceiling or floor requiring it to move along only
2 axes. Another alternative to vision based approaches is to
use optical feedback from a device like a retroreflector. By
placing a light weight retroreflector on the Robofly, we can
use an additional laser to verify alignment to the robot. Such
tracking systems could be attached to fixed or moving chase
vehicles acting as laser power base stations.
Second, 976 nm laser radiation at the levels required for
flight are above safe exposure limits. While the area within
the laser may not be safe, we can exploit the fact that laser’s
power is highly focused, therefore guaranteeing that the
unsafe areas are limited to the beam itself [16]. By using one
of the tracking methods proposed above, we could recognize
humans before they enter the beam to immediately turn off
the laser source, thus complying with the exposure limits.
Component Weight (mg)
DC-DC Converter & Driver Subtotal 73.7
Coupled Inductor 21
MOSFET 9.2
VsCapacitor 2.6
Diode 1.5
Driver Transistors 17.5
Cu traces 6.3
Circuit Substrate 10.0
Assorted Resistors 0.4
Solder & Conductors 3.0
Carbon Fiber Frame 2.2
MCU Assembly 17.5
PV Cell & Leads 13
Robofly without PEU 73
Misc. Glue & Wiring 13
Total Robot Weight 190
TABLE I: Total weight of all robot components including
body, power electronics, microcontroller, and PV cell.
VI. IMPLEMENTATION AND EVALUATION
We begin by describing the mechanical structure of our
robofly MAV followed by a detailed description of our setup
for flight experiments and discussion of results.
A. Robofly Design and Fabrication
The basic principle of the University of Washington (UW)
Robofly design [17] inherits from [3]: a bimorph piezo
cantilever actuator amplifies the small field-induced strains
into relatively larger motions at the tip. This is further ge-
ometrically amplified by a transmission structure consisting
of flexure joints to attain a 90stroke amplitude. The
wing’s angle of attack is allowed to rotate passively around
a torsional spring consisting of a flexure at the base of the
wing, resulting in a simple mechanism that produces insect-
like wing kinematics. The airframe consists of a single folded
structure made from laser micro-machined unidirectional car-
bon fiber composite bonded to polyimide flexural material.
Slight changes to transmission/flexure geometry have greatly
increased lift [10].
In addition to simplifying fabrication relative to [4], by
employing only a single part for the airframe, our Robofly
has a different arrangement of piezo actuators, that are
oriented horizontally. This facilitates easy integration of
electronics directly below, as shown in Fig. 7. The position
of the electronics package was chosen to facilitate assembly
and rework while avoiding adverse impacts on thrust and
stability. The PV cell is positioned above the robot to
achieve a direct line-of-sight path to the laser source. Without
position tracking, liftoff will move the cell out of the laser
beam and cut off power to the fly. We therefore assume a
small flight altitude and position the cell at a height of 20 mm
above the fly body. To drive the piezo bimorph to produce
wing-flapping oscillations requires a roughly constant high-
voltage bias signal, a ground signal, and an oscillating drive
signal that is roughly sinusoidal, as shown in the top half of
Fig. 3.
B. Setup and Takeoff Results
To demonstrate wireless liftoff capability, we position the
fly at a distance of 1m from the collimator output. With the
beam divergence of the laser, this results in a 13 mm spot
size at the PV cell which is more than sufficient to cover
it. We design a 0.6×0.75 ×0.6m enclosure and use a
series of 2 mirrors to achieve the 1m distance and to align
the beam on the cell. We program the microcontroller to
flap both wings continuously at maximum possible amplitude
at a frequency of 170 Hz using a single driver circuit to
maximize lift. Because the fly is dynamically unstable and
our goal is simply to demonstrate liftoff, we attach a carbon
fiber rod across the base of the fly in order to minimize
risk of structural damage during repeated experiments. We
perform experiments by placing a digital camera inside the
laser enclosure recording at a 240 Hz and apply short pulses
from our laser power source.
Prior to performing flight experiments with the electronics
attached, we verify that the fly is capable of liftoff when
driven by a 190 Vpp sinusoid at 170 Hz with a 130 mg
toothpick attached as a dummy load. Table. I shows the
final weight of the power electronics amounts to 104 mg,
which is well within this weight budget. We also verify the
functionality of the full system by measuring the output of
the electronics prior to final attachment on the fly while
powered by the PV cell and driving the actual fly. We
measure that the output of the PV cell is capable of providing
over 250 mW to supply the power demands of the boost
converter, drive circuit and microcontroller. Operating the
boost converter at 150 kHz with a 6% duty cycle yields
unloaded bias voltages of over 250 V as shown in Fig. 4. In
this configuration, despite variation in the bias rail due to use
of a smaller storage capacitor, the voltage difference between
the bias and sinusoid is more than 170 VPP at 170 Hz.
As seen in Fig. 1 and the supplementary video [18], we
demonstrate a completely wireless RoboFly liftoff using only
onboard electronics and wireless power transfer. We note that
the altitude of the flight could be easily improved in future
experiments. Specifically, the prototype fly shown in Fig. 7
includes a variety of fabrication errors and repairs which may
have made flight even more difficult. Additionally, lighter
components for the boost converter such as a sub milligram
single chip voltage regulator to replace the multicomponent
Fig. 7: Full insect scale robotic fly placed on a US penny for
scale. The power electronics and microcontroller are below
the robot and the PV cell is 20 mm above it.
shunt regulator used for the microcontroller, a lighter 5 mg
MOSFET as well as laser micromachining to remove unpop-
ulated areas of the circuit substrate would easily improve the
payload margin. Performance could further be improved by
reducing the power consumption of the micro-controller us-
ing low power optimizations on the chip or using alternative
chips available in the same size or smaller thereby allowing
a greater fraction of the total laser power to be delivered to
the wings. These weight reductions could allow the use of
a larger storage capacitor which would improve the boost
converter and driver performance thereby increasing wing
stroke amplitude and lift.
VII. RELATED WORK
Light weight robotic flight. Our work traces its lineage to
the Berkeley Micromechanical Flying Insect project (MFI)
to produce a honeybee sized flying robot. [2]. The MFI
approach differed from a similar but parallel attempt, the
mesicopter, that took a more traditional approach of rotors
and electromagnetic motors [19], which achieved lift greater
than weight in a larger, 3 g, 3 cm quad-rotor design. Fearing’s
piezo-and-wing approach, however, allowed for a much
smaller package because electrostatic forces (piezo) scale
downward more favorably in terms of efficiency and power
density than magnetic forces (motor) [7]. The piezo approach
first yielded lift greater than weight in 100 mg robot
by Wood [3]. Since that time, subsequent advances derived
from that basic approach [4][6] have required both power
and computation to be supplied from offboard resources
and supplied through a tiny wire tether. A recent work
achieved successful lift-off along vertical guides [20] using
electromagnetic actuation of a very similar flapping-wing
insect-scale robot. Due to the dependence on high electrical
current to generate a magnetic field, this approach requires
roughly five times more power at 1.2mW.
All other known demonstrations of wirelessly powered
flight are either passive mechanical or magnetic designs
which preclude useful autonomous application, or are sub-
stantially heavier. Purely mechanical demonstrations include
a rubber-band powered butterfly with a comparable weight
to our vehicle at 390 mg [21], and a 100 mg, 1cm paper
cone powered by external subwoofer [22]. [23] demonstrates
a passive 5mg flying machine using anisotropic magnetic
structure in an alternating external magnetic field to flap
wings. All electrically-powered robots are at least an order
of magnitude heavier, including a 2.1g jellyfish robot [24],
the Delfy Micro at 3g [25], the Piccolissimo at 2.5g [26],
and the nano hummingbird at 19 g [27]. Because the power
required by a flying robot scales closely with its weight,
lighter robots may make use of a greater diversity of power
sources, some of which are too small to power larger robots.
Boost converter design. Prior works have proposed a va-
riety of boost converter designs for MAVs using bimorph
actuators. [28] proposed transformer based designs, however
following [29], works have focused on coupled inductor
designs. [30] and [12] have focused on combining custom
controller ICs with coupled inductor circuits implemented
off chip. However, control functionality for these converter
designs can be implemented on micro-controllers of similar
size and weight that are already heavily optimized for power
and performance. Further, the weight of the converter circuit
is dominated by the inductor. Thus, we focus on an off-the-
shelf design that allows for faster prototyping and flexibility.
Our topology and inductor fabrication method is similar to
[11] which uses a bobbin and custom cut E-cores of similar
size, however we do not implement energy recovery in order
to reduce weight. While our design leverages these proven
topologies, our circuit fabrication technique allows us to
reduce weight. Most importantly, in contrast to prior work
we demonstrate the first boost converter system and a laser
based wireless power system, fully integrated into an MAV
and demonstrate that it provides an output capable of liftoff.
Laser power beaming. Prior works have successfully
demonstrated wirelessly powered robots [9] utilizing near
field power transfer. However these robots are not capable of
flight and the power transfer technique is physically limited
to close operational range. In the realm of wirelessly pow-
ered aircraft, WiBotic is developing solutions for near field
charging of drones [31], however as previously explained the
weight constraints of our MAV make onboard energy storage
such as rechargeable batteries infeasible.
Battery-free solutions such as the NASA Armstrong
project demonstrates laser power transfer by manually di-
recting a 1kW laser at a 300 g fixed wing aircraft propelled
by a 6W motor [32]. More recently LaserMotive has demon-
strated a laser power transfer to a quad rotor aircraft [33].
While these systems demonstrate wireless power transfer
using lasers, we focus on aerial vehicles that are orders of
magnitude smaller. Additionally, due to their higher weight
capacity and use of standard propulsion methods such as mo-
tors, these aircraft can use standard electronics components
and circuit manufacturing techniques to drive their motors.
In contrast wirelessly powering an MAV requires the power
electronics introduced in this paper to take the raw output
of the PV cell and convert it to the drive signal required
by the actuators. Moreover, from a practical perspective
the relatively low power requirements of insect scale robots
allow the use of comparatively lower power lasers.
VIII. CONCLUSION
This paper presents a significant milestone towards the
achievement of flight autonomy for honeybee-sized insect-
scale robots, demonstrating a wireless takeoff. Specifically,
we present the lightest wireless robotic flight to date by
showing liftoff of a 190 mg robot. We demonstrate the first
power electronics package fully integrated into a functional
aerial robot and fabricated by a unique application of laser
micro-machining techniques for fast-turnaround circuit fab-
rication. We also successfully demonstrate optical wireless
power transfer sufficient to run the power electronics and
onboard microcontroller.
This work serves as a platform for enabling a multitude
of new research directions advancing MAVs closer to the
vision of autonomous flight. The integration of an onboard
microcontroller presents the opportunity to add sensing and
communication capabilities which is a necessity for enabling
onboard control and achieving extended stable flight. Laser
based power for insect scale aerial robots also opens multiple
research directions for extended flight including tracking,
extending range, and laser-based communication.
ACK NO WL E DG EME NTS
The authors would like to thank Noah Jafferis for insight-
ful discussions about how to increase lift.
REFERENCES
[1] R. A. Brooks and A. M. Flynn, “Fast, cheap, and out of control: A
robot invasion of the solar system,Journal of the British Interplan-
etary Society, vol. 42, pp. 478–485, 1989.
[2] R. Fearing, K. Chiang, M. Dickinson, D. Pick, M. Sitti, and J. Yan,
“Wing transmission for a micromechanical flying insect,” in Robotics
and Automation (ICRA), 2000 IEEE Int. Conf. IEEE, 2000.
[3] R. J. Wood, “The first takeoff of a biologically inspired at-scale robotic
insect,” IEEE Trans. Robotics, vol. 24, no. 2, pp. 341–347, 2008.
[4] K. Y. Ma, P. Chirarattananon, S. B. Fuller, and R. J. Wood, “Controlled
flight of a biologically inspired, insect-scale robot,” Science, vol. 340,
no. 6132, pp. 603–607, 2013.
[5] S. B. Fuller, M. Karpelson, A. Censi, K. Y. Ma, and R. J. Wood,
“Controlling free flight of a robotic fly using an onboard vision sensor
inspired by insect ocelli,” Journal of The Royal Society Interface,
vol. 11, no. 97, p. 20140281, 2014.
[6] M. A. Graule, P. Chirarattananon, S. B. Fuller, N. T. Jafferis, K. Y. Ma,
M. Spenko, R. Kornbluh, and R. J. Wood, “Perching and takeoff of a
robotic insect on overhangs using switchable electrostatic adhesion,
Science, vol. 352, no. 6288, pp. 978–982, 2016.
[7] W. S. N. Trimmer, “Microbots and micromechanical systems,Sensors
and Actuators, vol. 19, pp. 267–287, 1989.
[8] R. J. Wood, B. Finio, M. Karpelson, K. Ma, N. O. P´
erez-Arancibia,
P. S. Sreetharan, H. Tanaka, and J. P. Whitney, “Progress on “pico”
air vehicles,” Int. J. Robotics Research, no. 11, 2012.
[9] M. Karpelson, B. H. Waters, B. Goldberg, B. Mahoney, O. Ozcan,
A. Baisch, P. M. Meyitang, J. R. Smith, and R. J. Wood, “A wirelessly
powered, biologically inspired ambulatory microrobot,” in 2014 IEEE
International Conference on Robotics and Automation (ICRA), May
2014, pp. 2384–2391.
[10] N. T. Jafferis, M. A. Graule, and R. J. Wood, “Non-linear resonance
modeling and system design improvements for underactuated flapping-
wing vehicles,” in 2016 IEEE International Conference on Robotics
and Automation (ICRA). IEEE, 2016, pp. 3234–3241.
[11] M. Karpelson, G.-Y. Wei, and R. J. Wood, “Driving high voltage
piezoelectric actuators in microrobotic applications,” Sensors and
Actuators A: Physical, vol. 176, pp. 78 – 89, 2012.
[12] M. Lok, E. F. Helbling, X. Zhang, R. Wood, D. Brooks, and G. Y. Wei,
“A low mass power electronics unit to drive piezoelectric actuators for
flying microrobots,” IEEE Transactions on Power Electronics, vol. PP,
no. 99, pp. 1–1, 2017.
[13] S. I. Inc., “Low esr, low leak current chip-type edlc cpx3225a752d,
2016. [Online]. Available: http://www.sii.co.jp/en/me/datasheets/chip-
capacitor/cpx3225a752d/
[14] V. Talla, B. Kellogg, B. Ransford, S. Naderiparizi, S. Gollakota, and
J. R. Smith, “Powering the next billion devices with wi-fi,ACM
CoNEXT ’15.
[15] M. Perales, M.-h. Yang, C.-l. Wu, C.-w. Hsu, W.-s. Chao, K.-h. Chen,
and T. Zahuranec, “Characterization of high performance silicon-based
vmj pv cells for laser power transmission applications,” in Proc. of
SPIE Vol, vol. 9733, 2016, pp. 97 330U–1.
[16] V. Iyer, E. Bayati, R. Nandakumar, A. Majumdar, and S. Gollakota,
“Charging a smartphone across a room using lasers,” IMWUT 2017.
[17] Y. Chukewad, A. Singh, and S. B. Fuller, “A robot fly composed of
actuator units folded from a single laminate sheet capable of flight,
landing, and ground locomotion,” in Robotics and Automation (ICRA),
IEEE Int. Conf., 2018, (under review).
[18] “Autonomous Insect Robotics Laboratory,” 2018. [Online]. Available:
http://depts.washington.edu/airlab/ICRA2018-1996.html
[19] I. Kroo and P. Kunz, “Development of the mesicopter: A miniature
autonomous rotorcraft,” in American Helicopter Society (AHS) Vertical
Lift Aircraft Design Conference, San Francisco, CA, 2000.
[20] Y. Zou, W. Zhang, and Z. Zhang, “Liftoff of an electromagnetically
driven insect-inspired flapping-wing robot,IEEE Transactions on
Robotics, vol. 32, no. 5, pp. 1285–1289, Oct 2016.
[21] H. Tanaka and I. Shimoyama, “Forward flight of swallowtail butterfly
with simple flapping motion.” Bioinspiration & biomimetics, vol. 5,
no. 2, p. 026003, Jun 2010.
[22] B. Liu, L. Ristroph, A. Weathers, S. Childress, and J. Zhang, “Intrinsic
stability of a body hovering in an oscillating airflow,” Physical Review
Letters, vol. 108, no. 6, p. 068103, 2012.
[23] K. I. Arai, W. Sugawara, K. Ishiyama, T. Honda, and M. Yamaguchi,
“Fabrication of small flying machines using magnetic thin films,” IEEE
Transactions on Magnetics, vol. 31, no. 6, pp. 3758–3760, Nov 1995.
[24] L. Ristroph and S. Childress, “Stable hovering of a jellyfish-like flying
machine,” Journal of The Royal Society Interface, no. 92, 2014.
[25] G. De Croon, K. De Clercq, R. Ruijsink, B. Remes, and C. De Wagter,
“Design, aerodynamics, and vision-based control of the delfly,Inter-
national Journal of Micro Air Vehicles, vol. 1, no. 2, pp. 71–97, 2009.
[26] M. Piccoli and M. Yim, “Piccolissimo: The smallest micro aerial
vehicle,” in Robotics and Automation (ICRA), 2017 IEEE International
Conference on. IEEE, 2017, pp. 3328–3333.
[27] M. Keennon, K. Klingebiel, H. Won, and A. Andriukov, “Development
of the nano hummingbird: A tailless flapping wing micro air vehicle,”
in AIAA Aerospace Sciences Meeting. Reston, VA: AIAA, 9–12
January 2012, pp. 1–24.
[28] E. Steltz, M. Seeman, S. Avadhanula, and R. S. Fearing, “Power
electronics design choice for piezoelectric microrobots,” in Intelligent
Robots and Systems, 2006 IEEE/RSJ International Conference on.
IEEE, 2006, pp. 1322–1328.
[29] M. Karpelson, G.-Y. Wei, and R. J. Wood, “Milligram-scale high-
voltage power electronics for piezoelectric microrobots,” in Robotics
and Automation, 2009. ICRA’09. IEEE International Conference on.
IEEE, 2009, pp. 2217–2224.
[30] S. Chaput, M. Renaud, R. Meingan, and J. F. Pratte, “A 3.7 v to 200
v highly integrated dc-dc converter with 70.4mems applications,” in
2014 IEEE 12th International New Circuits and Systems Conference
(NEWCAS), June 2014, pp. 381–384.
[31] “Wibotic,” 2017. [Online]. Available: http://www.wibotic.com/
[32] NASA, “Laser power for uavs,” 2004.
[33] T. J. Nugent, Jr. and J. T. Kare, “Laser power beaming for defense
and security applications,” pp. 804514–804 514–8, 2011. [Online].
Available: http://dx.doi.org/10.1117/12.886169
... As the size of the vehicle decreases, the efficiency of electromagnetic motors decreases substantially owing to increased friction and resistive losses 3,14 , resulting in low lift-to-power efficiencies. To achieve untethered and sustained flight for ultralight aerial vehicles, different types of actuator (such as piezoelectric ceramics 8,15,16 , dielectric elastomers 17,18 and linear electromagnetic actuators 19,20 ) have been used in previous studies to make lighter propulsion systems with higher efficiency, but large and bulky additional ground power systems (such as laser 21 , high-intensity light 8 and wireless radio frequency 22 ) are still needed to take off, which seriously hinders their practical applications. ...
... However, this resulted in a bulky structure, and there are limitations regarding application scenarios for these vehicles. To address these problems, different types of actuator (such as piezoelectric ceramic, dielectric elastomers and linear electromagnetic) with high efficiency and low power consumption at small sizes have been presented, and several even smaller MAVs achieve untethered flight using solar cells 8,21 , but they are powered by a high-power-density laser or triple sunlight (3 sun). In other words, attaining untethered, sustained flight fully powered by natural sunlight remains a considerable challenge for MAVs. ...
Article
Full-text available
Limited flight duration is a considerable obstacle to the widespread application of micro aerial vehicles (MAVs)1–3, especially for ultralightweight MAVs weighing less than 10 g, which, in general, have a flight endurance of no more than 10 min (refs. 1,4). Sunlight power5–7 is a potential alternative to improve the endurance of ultralight MAVs, but owing to the restricted payload capacity of the vehicle and low lift-to-power efficiency of traditional propulsion systems, previous studies have not achieved untethered sustained flight of MAVs fully powered by natural sunlight8,9. Here, to address these challenges, we introduce the CoulombFly, an electrostatic flyer consisting of an electrostatic-driven propulsion system with a high lift-to-power efficiency of 30.7 g W⁻¹ and an ultralight kilovolt power system with a low power consumption of 0.568 W, to realize solar-powered sustained flight of an MAV under natural sunlight conditions (920 W m⁻²). The vehicle’s total mass is only 4.21 g, within 1/600 of the existing lightest sunlight-powered aerial vehicle⁶.
... Although the above-mentioned vehicles have achieved stable flight motion, untethered flight motion is still difficult to achieve due to the limitations of onboard energy sources. Currently, there are some vehicles which achieve flight by using wireless energy, such as by using the three sources of sunlight [30,31], ground lasers [32], or microwaves [33]. Although wireless energy transfer can help achieve untethered flight of a vehicle, it requires a complex ground system to transmit a massive amount of energy, and its energy conversion efficiency is low. ...
Article
Unmanned Aerial Vehicles (UAVs) are ideal mobile chargers for energy-starved sensor nodes in wireless rechargeable sensor networks (WRSNs). Recently, the Power Access Docker (PAD) is developed to extend the operating range of the UAV in large-scale WRSNs by recharging the UAV in air. Existing research on WRSNs with PADs assume that PADs have unlimited energy, implying PADs are connected to consistent long-term energy supplies wherever they are deployed. However, this assumption may be achievable in urban WRSNs, but it is overly strict in suburban WRSNs, where only a few locations have reliable energy supplies. Consequently, the deployment of PADs in suburban WRSNs is constrained and cannot support UAV coverage of the entire surveillance area. Our work addresses this issue by introducing a vehicle-borne mobile PAD to large-scale suburban WRSNs. The mobile PAD may drive to places that the pre-deployed PADs cannot cover to recharge the UAV, allowing it to function across the whole area. To the best of our knowledge, this is the first work to consider the energy constraint of PAD deployment. To minimize the nodal charging delay, we investigate the problem of periodic charging with the collaboration of the UAV and the mobile PAD in the proposed new scenario. To address this problem, we propose the Active Rendezvous Charging Scheme (ARCS), which divides nodes into two types, plans the UAV’s travel path, and schedules the rendezvous between the UAV and the mobile PAD. Simulation results show that ARCS can effectively address the energy problem in large-scale suburban WRSNs.
Article
Full-text available
Ornithopters are bird‐like flapping‐wing robots. Only small ornithopters can hover, with long endurance at hummingbird size. Could larger ornithopters be improved further to hover longer? This paper reviews and examines the drive and power of hovering ornithopters, and elastic means of energy or thrust boosters. While the rotation of flexible wings enhance the thrust generation, two‐winged ornithopters did not scale up well because of higher disk loading. In comparison, the X‐winged or multiple‐V‐winged ornithopters enjoy a lower disk loading by beating multiple wings slower, at a smaller stroke angle or a longer span. Further, the clap‐and‐fling interaction of V and X‐wings boosts the thrust generation. Future works can explore the wing flexibility and morphology change to improve the hoverability and flight agility of ornithopters.
Preprint
Full-text available
Flying insects can perform rapid, sophisticated maneuvers like backflips, sharp banked turns, and in-flight collision recovery. To emulate these in aerial robots weighing less than a gram, known as flying insect robots (FIRs), a fast and responsive control system is essential. To date, these have largely been, at their core, elaborations of proportional-integral-derivative (PID)-type feedback control. Without exception, their gains have been painstakingly tuned by hand. Aggressive maneuvers have further required task-specific tuning. Optimal control has the potential to mitigate these issues, but has to date only been demonstrated using approxiate models and receding horizon controllers (RHC) that are too computationally demanding to be carried out onboard the robot. Here we used a more accurate stroke-averaged model of forces and torques to implement the first demonstration of optimal control on an FIR that is computationally efficient enough to be performed by a microprocessor carried onboard. We took force and torque measurements from a 150 mg FIR, the UW Robofly, using a custom-built sensitive force-torque sensor, and validated them using motion capture data in free flight. We demonstrated stable hovering (RMS error of about 4 cm) and trajectory tracking maneuvers at translational velocities up to 25 cm/s using an optimal linear quadratic regulator (LQR). These results were enabled by a more accurate model and lay the foundation for future work that uses our improved model and optimal controller in conjunction with recent advances in low-power receding horizon control to perform accurate aggressive maneuvers without iterative, task-specific tuning.
Article
Full-text available
We demonstrate a novel laser-based wireless power delivery system that can charge mobile devices such as smartphones across a room. The key challenges in achieving this are multi-fold: delivering greater than a watt of power across the room, minimizing the exposure of the resulting high-power lasers to human tissue, and finally ensuring that the design meets the form-factor requirements of a smartphone and requires minimal instrumentation to the environment. This paper presents a novel, and to the best of our knowledge, the first design, implementation and evaluation of an end-to-end power delivery system that satisfies all the above requirements. Our results show that we can deliver more than 2 W at ranges of 4.3 m and 12.2 m for a smartphone (25 cm2) and table-top form factor (100 cm2) receiver respectively. Further, extensive characterization of our safety system shows that we can turn off our laser source much before a human moving at a maximum speed of 44 m/s can even enter the high-power laser beam area.
Conference Paper
Full-text available
We present the first power over Wi-Fi system that delivers power to low-power sensors and devices and works with existing Wi-Fi chipsets. Specifically, we show that a ubiquitous part of wireless communication infrastructure, the Wi-Fi router, can provide far field wireless power without significantly compromising the network's communication performance. Building on our design, we prototype battery-free temperature and camera sensors that we power with Wi-Fi at ranges of 20 and 17 feet respectively. We also demonstrate the ability to wirelessly trickle-charge nickel---metal hydride and lithium-ion coin-cell batteries at distances of up to 28 feet. We deploy our system in six homes in a metropolitan area and show that it can successfully deliver power via Wi-Fi under real-world network conditions without significantly degrading network performance.
Article
Full-text available
Making small robots stick Aerial views offer the chance to observe a wide range of terrain at once, but they come at the cost of needing to stay aloft. Graule et al. found that electrostatic forces could keep their insect-sized flying robot stuck to the underside of a range of surfaces (see the Perspective by Kovac). They mounted an electrostatically charged pad to the top of their robot, which could then reversibly stick to existing elevated perches—including a leaf—using less power than would be needed for sustained flight. Science , this issue p. 978 ; see also p. 895
Article
This paper presents a power electronics design for the piezoelectric actuators of an insect-scale flapping-wing robot, the RoboBee. The proposed design outputs four high voltage drive signals tailored for the two bimorph actuators of the RoboBee in an alternating drive configuration. It utilizes fully-integrated drive stage circuits with a novel highside gate driver to save chip area and meet the strict mass constraint of the RoboBee. Compared with previous integrated designs, it also boosts efficiency in delivering energy to the actuators and recovering unused energy by applying three power saving techniques, dynamic common mode adjustment, envelope tracking, and charge sharing. Using this design to energize four 15 nF capacitor loads with a 200 V and 100 Hz drive signal and tracking the control commands recorded from an actual flight experiment for the robot, we measure an average power consumption of 290mW.
Article
We present the first power over Wi-Fi system that delivers power to low-power sensors and devices and works with existing Wi-Fi chipsets. We show that a ubiquitous part of wireless communication infrastructure, the Wi-Fi router, can provide far field wireless power without significantly compromising the network's communication performance. Building on our design, we prototype battery-free temperature and camera sensors that we power with Wi-Fi at ranges of 20 and 17 ft, respectively. We also demonstrate the ability to wirelessly trickle-charge nickel--metal hydride and lithium-ion coin-cell batteries at distances of up to 28 ft. We deploy our system in six homes in a metropolitan area and show that it can successfully deliver power via Wi-Fi under real-world network conditions without significantly degrading network performance.
Article
We present the first electromagnetically driven, self-lifting, sub-100-mg, insect-inspired flapping-wing microaerial vehicle. This robot, with a weight of 80 mg and a wingspan of 3.5 cm, can produce sufficient thrust to lift off. The wing beat frequency is up to 80 Hz and the flapping amplitude is approximately ±70°. An electromagnetic actuator is employed to control the flapping amplitude and create passive wing rotation. To the best of our knowledge, this is the world's smallest electromagnetically driven flapping-wing robot that is capable of liftoff.
Conference Paper
Continuing improvements in the cost and power of laser diodes have been critical in launching the emerging fields of power over fiber (PoF), and laser power beaming. Laser power is transmitted either over fiber (for PoF), or through free space (power beaming), and is converted to electricity by photovoltaic cells designed to efficiently convert the laser light. MH GoPower’s vertical multi-junction (VMJ) PV cell, designed for high intensity photovoltaic applications, is fueling the emergence of this market, by enabling unparalleled photovoltaic receiver flexibility in voltage, cell size, and power output. Our research examined the use of the VMJ PV cell for laser power transmission applications. We fully characterized the performance of the VMJ PV cell under various laser conditions, including multiple near IR wavelengths and light intensities up to tens of watts per cm². Results indicated VMJ PV cell efficiency over 40% for 9xx nm wavelengths, at laser power densities near 30 W/cm². We also investigated the impact of the physical dimensions (length, width, and height) of the VMJ PV cell on its performance, showing similarly high performance across a wide range of cell dimensions. We then evaluated the VMJ PV cell performance within the power over fiber application, examining the cell’s effectiveness in receiver packages that deliver target voltage, intensity, and power levels. By designing and characterizing multiple receivers, we illustrated techniques for packaging the VMJ PV cell for achieving high performance (> 30%), high power (> 185 W), and target voltages for power over fiber applications.
Conference Paper
Laser-based pico projectors are readily available, however to push their integration one step further and include them in smartphones, tablets or car head up displays, new ways to generate the high voltage required by electrostatic MEMS actuators must be proposed to meet smartphones' small form factor and power requirements. In that context, this paper presents a highly-integrated coupled inductor DC-DC converter that can convert a 3.7 V battery voltage into a 150 V to 300 V output voltage suitable for biasing electrostatic MEMS in portable applications. The fabricated circuit occupies 4.21 mm2 in a 0.8 μm 300 V HVCMOS process. To achieve that small active area, a power switch that reduces by up to 79% the switch silicon area requirement has been developed. At 200 V output voltage, the DC-DC converter can deliver up to 500 mW with a peak efficiency of 70.4% which outperforms current highly integrated solutions.