Received: 27 February 2017 Revised: 8 January 2018 Accepted: 25 January 2018

DOI: 10.1002/rob.21777

SYSTEMS ARTICLE

Loon Copter: Implementation of a hybrid unmanned

aquatic–aerial quadcopter with active buoyancy control

Hamzeh Alzu'bi Iyad Mansour Osamah Rawashdeh

Department of Electrical and Computer

Engineering, Oakland University, Rochester, Abstract
Michigan, USA Aquatic–aerial unmanned vehicles recently became the focus of many researchers due to their
Correspondence various possible applications. Achieving a fully operational vehicle that is capable of aerial, water-
Hamzeh Alzu'bi, Department of Electrical and
surface, and underwater operations is a significant challenge considering the vehicle's air–water–
Computer Engineering, Oakland University,
Rochester, Michigan, USA. air transition, propulsion system, and stability underwater. We present in this paper an unconven-
Email: hmalzubi@oakland.edu tional unmanned hybrid aquatic–aerial quadcopter with active buoyancy control that is capable
Funding information of aerial flight and water-surface operation, as well as subaquatic diving. We report on the first
This work was sponsored in part by the National successful prototype of the vehicle, named the Loon Copter, to provide initial evaluation results
Science Foundation (NSF) through award no.
EEC-1263133, which is available to undergradu-
of its performance in both mediums. The Loon Copter uses a single set of motors and propellers
ate students participating in a 10-week Research for both air and underwater maneuvering. It utilizes a ballast system to control vehicle buoyancy
Experience for Undergraduates (REU) program and depth underwater, as well as to perform seamless air-to-water and water-to-air transitions.
at Oakland University in Rochester, Michigan.
A closed loop control algorithm is utilized for the vehicle's aerial and water-surface stability and
maneuver, whereas an open loop control algorithm is used for underwater maneuver. The experi-
mental results show a fully operational prototype with six degrees of freedom underwater, stable
flight, operation capabilities on water surface, and agile maneuvering underwater.

KEYWORDS
buoyancy, control, propellers, unmanned aerial vehicle, unmanned aquatic-aerial vehicle,
unmanned underwater vehicle

1 INTRODUCTION does not allow for simultaneously inspection of both the submersed
and the nonsubmersed portions of the structure. A hybrid vehicle
For the last two decades, unmanned aerial vehicles (UAVs) and would be able to accomplish both more efficiently and at a lower cost.
unmanned underwater vehicles (UUVs) have been the focus of many (Drews et al., 2009)
researchers (Bouabdallah, 2007; Budiyono, 2009; Ferreira, Matos, An aquatic unmanned aerial vehicle is a hybrid aircraft that has
Cruz, & Pinto, 2010; Iscold, Pereira, & Torres, 2010; Ranganathan, capabilities to operate in the air and in the water. (Siddall & Kovač,
Thondiyath, & Kumar 2015; Ridao, Batlle, & Carreras 2001; Schmid, 2014) It possesses the advantages of UAVs and UUVs. The con-
Lutz, Tomić, Mair, & Hirschmüller, 2014). The popularity of these types cept of manned hybrid aquatic–aerial aircraft was proposed in 1934.
of vehicles will continuously increase due to their expanding avail- There have been few manned prototypes; none operated success-
ability and the cost efficiency of their production. These two kinds of fully in either the air or the water due to technical difficulties,
vehicles were designed to perform well in their own specific environ- including the compatibility of the structure in the two different
ments. However, some situations require a vehicle to operate well in mediums, the crew cabin, and the life support system. (Yang et al.,
both aerial and underwater environments, such as mapping of remote 2015)
regions, and inspection of submerse structures, such as petroleum This article will further discuss the design and implementation of
platforms, ship hulls, and gas pipelines, as well as several military the vertical takeoff and landing (VTOL) hybrid aquatic–aerial Loon
applications. (Drews, Neto, & Campos, 2009; Yang, Wang, Liang, Yao, Copter. As the name implies, this vehicle possesses capabilities of fly-
& Liu, 2015) To fulfill this type of mission, the traditional UAV or ing in the air, taking off from the water, landing on the water, maneu-
UUV needs multivehicle collaboration. (Yang et al., 2015) An auxil- vering on the water's surface (WS), and navigating underwater. A novel
iary vessel is currently used to transport the underwater vehicle to buoyancy control setup allows the Loon Copter to dive, surface, and
the goal location so it may dive to perform the inspection. However, loiter on the surface or underwater without the use of its propellers
this kind of operation is very expensive and inefficient. Moreover, it (Figure 1).

J Field Robotics. 2018;1–15. wileyonlinelibrary.com/journal/rob 

c 2018 Wiley Periodicals, Inc. 1
2 ALZU'BI ET AL .

F I G U R E 1 The Loon Copter functions in three modes of operation: the aerial mode (a), in which the vehicle operates as a conventional quad-
copter; the surfacing mode (b), where the vehicle transits on the surface of the water; and the diving mode (c), where the vehicle propels
underwater

There is currently some active research on hybrid aquatic–aerial the surface of turbulent water due to its static neutral buoyancy. The
vehicles, such as the hybrid aerial underwater vehicle, designed by Loon Copter is different in that it is capable of actively controlling its
the MIT Lincoln Lab. This design features a fixed-wing unmanned buoyancy using a ballast system. This allows it to naturally loiter on
vehicle capable of flying, plunge diving into water, and navigating the surface of water and then allows for submersion without the need
underwater by folding its wings. The vehicle can only transition from for propellers to pull the vehicle underwater. The Loon Copter's ballast
air to water, not from water to air. (Fabian, Feng, Swartz, Thurmer, & system allows the vehicle to suspend in a horizontal position underwa-
Wang, 2012) In addition, the Flimmer (Flying Swimmer) Program at ter at a required depth and makes it capable of controlling the depth
the Naval Research Laboratory (NRL) produced a flying submarine, without use of propellers.
which also featured a fixed wing airplane. (Edwards, 2014) Both vehi- Interest in operating drones in and around water is also exempli-
cles lack the ability to transition smoothly between air and water, as the fied through the Corrosion Resistant Aerial Covert Unmanned Nauti-
transition at high speed creates a high impact on the vehicle's frame. cal System (CRACUNS) project. CRACUNS’ vehicle is designed to be
This increase the complexity and cost of designing the vehicle's struc- launched by an UUV, diver, or a fixed position underwater and to fly
ture so that it can handle the impact force. On the other hand, the like an autonomous UAV. (Hopkins University Applied Physics Labo-
Loon Copter utilizes a ballast system to perform a seamless transi- ratory, 2016) The project is focused on the construction and testing
tion between the two media with simple airframe design. Some work of corrosion resistant drones. The designers fabricated a lightweight,
involving multirotor-type drones is also underway. Drews, Neto, and submersible, composite airframe able to withstand the water pressure
Campos (2014) published a paper that modeled and simulated a mul- experienced while submerged. The vehicle's airframe is designed to
tirotor type drone that uses two sets of propellers for a multirotor sink in water, and it uses the propellers to control depth and main-
hybrid aquatic–aerial vehicle; one set for aerial flight and the other tain stability underwater, which would add power penalty on the rotors
set for underwater maneuver. The vehicle will carry constantly the similar to other vehicles previously discussed. Also, it lacks the ability
unused system, which would increase the power consumption, and to park on the surface of water. On the other hand, the Loon Copter is
the inactive propellers may add drag penalties. Lock, Vaidyanathan, designed with active buoyancy control, and it can control its buoyancy
and Burgess (2013) and Izraelevitz and Triantafyllou (2014) studied a and maintain its stability without the need for the rotors. Results from
bioinspired flapping foil with in-line motion to ensure efficient thrust the CRACUNS project are shaping to provide valuable references for
production in different media. Both demonstrated the efficacy of this constructing robust, corrosion resistant hybrid drones.
strategy in tunnel tests. Tan et al. investigated a single propeller with In 2014, we set out to design a hybrid drone, which we have now
a gear box for both media; however, this increases the mechanical dubbed the Loon Copter. To achieve the Loon Copter's objectives of
complexity and it risks power losses penalty due to the gear box fric- operations in aerial flight, surface, and underwater, we set out with the
tion. In contrast, the Loon Coper is a hybrid multirotor platform that following main requirements:
uses a single set of propellers with direct drive in air and underwater,
thereby reducing the vehicle's mechanical complexity and extra weight (i) Proper vehicle waterproofing.
penalty. (ii) Air and underwater stability and capability of buoyancy adjust-
The “Naviator” Drone (Maia, Soni, & Diez-Garias, 2015) was intro- ments for a seamless transition between air and water, as well as
duced at about the same time as the Loon Copter and is most closely to control depth underwater.
related. The drone is an octacoaxial multirotor hybrid aquatic–aerial (iii) Minimal load on the vehicle's power-plant underwater.
vehicle. The design makes use of four coaxial rotors with fixed pitch
propellers. It is designed to be naturally buoyant, and uses the rotors to This paper provides a description of the design and discusses
go underwater and to control the depth. All eight propellers are used results and possible future improvements and modifications. Sec-
to maintain or change depth as well as to maneuver underwater. One tion 2 begins by providing a high-level overview of the quadcopter.
disadvantage of this approach is that it consumes significant power to Section 3 introduces the vehicle's technique for air–water–air transi-
maintain the vehicle's stability and depth underwater and to stay on tioning and achieving stability underwater. In Section 4, we evaluate
ALZU'BI ET AL . 3

F I G U R E 2 Loon Copter external and internal components: external components (left) consist of water-proof hull, cameras, four brushless motors
with four propellers, and plastic frame. Internal components (right) consist of lithium polymer batteries, water pump, water cylinder and two-way
valve, as well as four speed controllers, flight controller, and pump controller which are not shown

FIGURE 3 Loon Copter first 2014 prototype (left) and current 2016 prototype (right)

the aerial power-plant for use underwater. Lastly, Section 5 explores unmanned aquatic–aerial quadcopter. The field test∗ of the vehicle's
the vehicle's electronics and control systems. We then discuss the per- first prototype demonstrated promising results of our novel hybrid
formance of the vehicle and conclude the paper. VTOL drone, which was able to fly, surface, and dive underwater.
After experimentally analyzing the vehicle's performance, it was obvi-
ous that the transition between air and water needed enhancement
2 HYBRID AQUATIC–AERIAL to improve the vehicle's stability underwater. As shown in Figure 3,
QUADCOPTER OVERVIEW the ballast system of the first prototype was mounted on the hull
externally, and the water cylinders (consisting of two syringes) were
The Loon Copter consists of the following main components, as shown mounted horizontally on the vehicle's airframe directly under the pro-
in Figure 2 a water-proof hull, four sealed-bearing motors with four pellers. This lack of symmetry of the vehicle's frame made the vehicle
fixed-pitch propellers, two lithium polymer (LiPo) batteries, a plastic difficult to control underwater. Based on the first prototype results, the
frame, a ballast system, four speed controllers, and an avionics system. current prototype was developed to enhance the Loon Copter's design
Two cameras are also attached as a payload for surveillance, one for and controllability underwater. As shown in Figure 3, the current pro-
air and one for underwater. The vehicle's frame is constructed out of totype has a cleaner frame compared with the first prototype, which
a plastic hull for cost efficiency and ease of manufacturing. The rotors was achieved by moving the ballast system inside the hull. This also had
consist of waterproof brushless motors and carbon fiber fixed-pitch a major effect on the vehicle's stability underwater since the buoyancy
propellers. The depth control system is comprised of a water pump, force depends on the shape of the vehicle. The detailed operation of
a two-way valve, a water cylinder, and a water pump controller. The the ballast system is discussed in Section 3.2. Experimental tests† show
avionics system is a Multiwii pro flight controller. As a power source, that the current design produced a fully operational prototype.‡
two LiPo—3600 mAh (14.8 V) batteries are used to power the motors
and the avionics. ∗ Hybrid aquatic–aerial quadcopter first prototype test: http://youtu.be/5pV32tLSv8Y
Figure 3 shows the vehicle's first prototype, successfully tested in † Test of the Loon Copter current prototype: https://www.youtube.com/watch?v=STRijGJ5qtI
2014, and the current prototype of the Loon Copter. The first pro- ‡ Loon Copter won the 2016 UAE Drones for Good competition in Dubai. The UAE Drone for
totype was designed to assess the feasibility of developing a hybrid Good Award: http://www.dronesforgood.ae/
4 ALZU'BI ET AL .

The three operation modes of the drone were tested in indoor i) An air compressor-based ballast system.
and outdoor environmental conditions. A calm swimming pool was ii) A hydraulic pump-based ballast system.
used for water-surface and underwater tests. In the outdoor test, the
iii) A piston tank ballast system.
average wind speed was about 13 kph. This speed barely affected
iv) A direct thrust system.
the condition of the pool; however, small waves were created due to
water-surface operation. The vehicle flight tests were performed in
The first three concepts are classified as static diving systems,
hover mode for both indoor and outdoor conditions, and the lowest
whereas the latter concept is a dynamic diving system. (Wang, Chen,
batteries voltage reached was 14.6 V to avoid damaging the LiPo bat-
Marburg, Chase, & Hann, 2008)
teries. The vehicle's average outdoor flight time was about 10.5 and
The Loon Copter's depth and transition control system design
12 min for indoor flight. Obviously, the difference in flight time was due
employs a combination of a hydraulic pump-based ballast system and
to wind effects. The average underwater operation time was 22 min,
piston tank ballast system to achieve a seamless air–water and water–
with an average underwater speed of 0.5 m/s. Moreover, the average
air transition and to reduce power load on the vehicle rotors. This static
water-surface operation time for both indoor and outdoor tests was
diving system allows for control of the vehicle's depth underwater, as
about 20 min with an average speed of 1.0 m/s. The total operation time
well as enables the vehicle to flip 90◦ with minimal power consump-
of the Loon Copter was about 15 min for the three operation modes,
tion compared with dynamic diving. This system was created because
with about 8 min of flight time. Using the ballast system, the vehicle
of its simple design, easiness to control, and flexibility to distribute
could suspend for about 11 h underwater without propellers. The max-
the system components on the vehicle frame. As shown in Figure 4,
imum communication link underwater between the vehicle and the
the Loon Copter's ballast system consists of the vehicle's hull, a water
remote controller was 3 m in depth and 10 m horizontally. The Loon
cylinder, a movable piston, a two-way valve, and a water pump. The
Copter's air–water transition time was about 25 s, and the water–air
hull of the vehicle is used to mount the ballast system components
transition time was about 13 s. The current prototype weight with pay-
as well as to hold the compressed air. In addition, the movable piston
load is 2.7 kg.
is used to hold the water in the water cylinder during diving, which
improves the vehicle's stability. The water cylinder capacity is 150 ml,
about 100 ml is used to submerge the Loon Copter underwater and the
3 LOON COPTER TRANSITION AND rest is utilized to control depth. The two-way valve is needed to hold

STABILITY UNDERWATER the water in the water cylinder, as the compressed air inside the hull
will attempt to push the water out of the cylinder when it is under-
The vehicle's transition between air and water, as well as stability and water. The Loon Copter utilizes the ballast system without the vehi-
depth control while underwater, must also be examined. The following cle's rotors to perform the two-media transition and to control depth.
subsections will consider the details of aerial to underwater control. A data logger and power sensor were used to measure the power con-
sumption of the Loon Copter's ballast system, as it performed transi-
tion and maintained the vehicle underwater, versus its rotors. The bal-
3.1 Buoyancy and depth control
last system consumes 6 W, whereas the rotors consume 450 W, which
UUVs can dive in one of two ways: static and dynamic. The static diving is a significant amount of power to perform transition and to maintain
system utilizes a ballast system to dive and surface the vehicle without the vehicle underwater. Even though the ballast system costs the Loon
any forward movement, whereas dynamic diving uses control surfaces Copter about 1 min of flight time, it saves a significant amount of power
or rotors to force the vehicle to dive underwater. There are four con- to suspend the vehicle and to control depth underwater compared with
cepts within these two diving systems: the vehicle's rotors.

FIGURE 4 Loon Copter's hull side view showing ballast system

ALZU'BI ET AL . 5

F I G U R E 5 Air–water transition water flow (left): the water pump moves the water through the valve to the water cylinder until the vehicle
submerges underwater. Water–air transition water flow (right): to restore the vehicle to the water surface, the water pump moves the water from
the water cylinder through the valve

Figure 5 shows the water flow direction when the vehicle transi- will level horizontally. On the other hand, if one of the centers shifts
tions between the two media. While the vehicle is on the WS, the water from the line of action, a rolling moment will be created that will tilt
cylinder and the hull will have a certain amount of air at ambient pres- the vehicle. Figure 6b captures the effect of the rolling moment. When
sure. The vehicle's transition will begin as water is pumped into the water begins to pump to the cylinder, the Cg will shift to the left, and
water cylinder from the water inlet. As the water cylinder is filled with by continuing to pump the water, Cg will shift enough that it will make
water, the air inside the cylinder will move to the vehicle hull, which will the Loon Copter tilt ∼90◦ relative to WS and dive. The final state after
become pressurized by the addition of water. When the water cylinder transition is completed is shown in Figure 6c.
is filled, the Loon Copter will start to tilt 90◦ and submerge simultane- When the vehicle is on the surface of the water, the metacenter will
ously. Similarly, the vehicle's water–air transition will occur when the be above Cb and Cg . If some disturbance attempts to move the vehi-
water cylinder is emptied, which will cause the Loon Copter to float to cle from the equilibrium position, a moment will be created around M
the surface of the water and revert back to its upright position. to restore the vehicle to its original position. When the Loon Copter is
completely submerged Cb and M will be at a common point. After the
Loon Copter reaches the 90◦ position underwater, the Cb and Cg will
3.2 Center of buoyancy and center of gravity control lie on the same line of action.
As shown in Figure 4, the ballast system was designed with the
This subsection discusses the relationship between the center of grav-
water cylinder placed alongside the hull of the Loon Copter, which aids
ity and the center of buoyancy of the Loon Copter and its effect on sta-
in the transition and stability of the vehicle underwater. As the water
bility. Buoyancy is the upward force asserted on an immersed or float-
cylinder fills with water, the center of gravity becomes below the cen-
ing body by the supporting fluid. This conception of the term conveys
ter line of the airframe. If some underwater turbulence disrupts the
the idea that volume, alone, determines buoyancy and that the upward
vehicle from its equilibrium condition, a moment will be created around
force exerted on the immersed or floating body equals the weight of
M to restore the vehicle to its original position. When the Loon Copter
the fluid that it displaces. (United States Navy, 2008) Given this prop-
is in the air, the center of gravity is located just below the propellers,
erty, tilting the vehicle underwater while maintaining its stability is
like in a conventional aerial quadcopter, which allows for stability while
considered a major challenge due to the location of the center of grav-
the vehicle is in flight.
ity relative to that of the center of buoyancy. The stability of the vehi-
cle depends on the relative lines of action of the two centers. Thus,
the center of buoyancy must be precisely above the center of gravity;
otherwise, the vehicle will be unbalanced in terms of roll and/or pitch 4 LOON COPTER POWER PLANT
moment. To reduce the power requirements of the Loon Copter sta- EVALUATION
bility and underwater maneuvering, the Loon Copter's ballast system
is used to control the depth of the vehicle and to tilt the airframe of This section will evaluate the preliminary feasibility of using aerial,
the vehicle such that its rotors are perpendicular to the surface of the fixed-pitch propellers underwater. Various tests were performed both
water. in air and water using a test stand designed to study the thrust to
The vehicle's primary stability underwater depends mainly on the power ratio of commercially available, counter-rotating, multirotor
center of buoyancy and the center of gravity. The Loon Copter's air– propellers.
water transition is illustrated in Figure 6. Figure 6a shows the vehicle's As shown in Figure 1, the Loon Copter makes use of one set of
landing on the WS. As notated in this figure, there are three centers on propellers to both fly and move underwater. This objective ulti-
the line of action: the metacenter (M), the center of buoyancy (Cb ), and mately reduces the complexity and weight of the vehicle. On most
the center of gravity (Cg ). The metacenter is the point of intersection aerial/aquatic vehicles, propellers are designed to operate efficiently in
of a vertical line with the center of buoyancy of the Loon Copter body, various mediums with different densities and viscosities, and therefore
with the vertical line through the new center of buoyancy when the have distinct properties. Thus, in the context of combining propeller
body tilts. If Cb and Cg are on the same line of action, the Loon Copter usage for operation in air and underwater, cavitation, or the formation
6 ALZU'BI ET AL .

F I G U R E 6 Loon Copter air–water transition. Subfigures (a) shows the Loon Copter's landing with its rotors are parallel to the water's surface
(WS), (b) illustrates the vehicle's air–water transition, and (c) presents the final state of the vehicle, where the Loon Copter suspends horizontally
underwater

of vapor bubbles in water near a moving blade, becomes a challenge. (revolution per minute − rpm), D is the propeller diameter (m), CT is
Such bubbles bursting may result in shockwaves being sent back to the thrust coefficient, and CP is the power coefficient. These two
the propellers, causing the propellers to erode. This phenomenon, equations describe the essential parameters that are needed to
though, usually occurs at high speeds. Nevertheless, from a theoretical calculate the thrust and power of a propeller in both air and water
standpoint, air and water propellers will behave similarly, essentially mediums. Based on these two equations, a test-stand was developed
in the interest of operating air propellers at low Mach numbers (where to experimentally calculate the thrust and power in order to study
compressibility effects are negligible) and water propellers operating the propeller's performance. For each test, CT , CP , 𝜌, and D were
without cavitation. (Kerwin, 1986) fixed constants and the variable 𝜔 was adjusted. The total thrust to
The main parameters that describe thrust and power of a propeller electrical power consumption was calculated based on experimental
are shown in the following equations (Drews et al., 2014): measurements.
The propeller test stand, shown in Figure 7, was built using com-
T = CT 𝜌𝜔2 D4 (1) mercially available components. It is comprised of a data logger to
collect test data, a thrust meter (scale) to measure thrust, a rotor, an
3
P = CP 𝜌𝜔 D 5
(2) Arduino board, an rpm sensor, a power sensor, and a thrust measure-
ment mechanism. The rotor is a brushless motor (model dji 2212/920
where T is the thrust (Newtons), P is the propulsion power KV) with a variable pitch propeller (model varioprop 8″ by 3″ - 4.75″ )
(Watts), 𝜌 is the environment density (kg∕m3 ), 𝜔 is rotation speed as the Arduino board was programmed to run the brushless motor
ALZU'BI ET AL . 7

0.26 N/W underwater. Subfigure (e) illustrates that pitch angle 3.75″
was the best, with maximum thrust to power ratio of 0.085 N/W for
aerial tests. It is obvious that using an aerial propeller with low pitch
angle is more efficient for underwater operation and that a higher pitch
angle is better for aerial flight. We considered different propulsion sys-
tem designs to use small pitch angle underwater and high pitch angle
in air, such as a variable pitch propeller, or the possibility of using two
separate propellers, one for air and the other for underwater. Also,
Tan, Siddall, and Kovac (2017) suggested a single aerial propeller with
a gear box to reduce the power load on the rotor, and to avoid damag-
ing the brushless motor. However, his results are questionable as they
were based on simulation. All those alternatives would add significant
mechanical complexity and weight penalty on the rotors, which makes
this challenging to implement. We chose to use a single fixed pitch pro-
peller with direct drive to make the design simple and to reduce the
extra weight penalty. Moreover, the experimental results show that
the power consumption at low speed without cavitation underwater
is comparable to the power consumption in aerial flight. For example,
at 3.75″ pitch angle, the aerial power consumption is 10.57 W and
thrust is 0.87 N at about 4000 rpm, whereas the underwater power
consumption is 8.14 W and thrust is 1.33 N at about 200 rpm. Based
FIGURE 7 CAD model of the propeller test stand
on the test results shown in Figure 8, it can be deduced that the thrust,
power consumption and efficiency curves have similar patterns in both
at different speeds. The data points of power, thrust, and rpm were air and water. In addition, it was found that the aerial propeller was
collected at each speed. The propeller was tested underwater in a efficient at a low rpm without cavitation. Based on these test results
rectangular plastic tank, 0.75 m long by 0.55 m wide by 0.50 m deep. and Equations (1) and (2), we conclude that the behavior of the aerial
The tip of the propeller was 20 cm under the WS. fixed pitch propeller underwater at low speed is similar to its behavior
As shown in Figure 7, the thrust measurement mechanism was in the air, and it can provide enough thrust for movement and maneu-
designed to measure the rotor's thrust. This mechanism contains a ver underwater. Therefore, the 12″ × 3.8″ aerial fixed pitch propellers
bearing and an aluminum arm. The rotor is attached to one end of the prove to be the most advantageous option for the Loon Copter vehi-
arm, while the scale is on the other end with both of them equidistant cle. Since each of the vehicle's rotors could generate 12 N thrust at
to the bearing. When the rotor pulls the arm, the arm rotates around 800 rpm, and the ballast system is used to suspend the vehicle under-
the bearing and creates force on the scale. This force is the thrust water the generated amount of thrust is sufficient to move the vehicle
produced from the rotor. underwater.
Figure 8 shows the test results of the variprop propeller starting The chosen power plant (motor–propeller combination) is quite
at a 3″ pitch angle and adjusting through 4.75″ pitch angle in 0.25″ common for drones of the Loon Copter's size and weight, and there-
increments. Depicted on the left-hand side of Figure 8 are the pro- fore resulted in minimal penalty in terms of flight time in air. The power
peller's measurements in the air test, whereas the right-hand side dis- plant was also extensively tested underwater. The current limit of the
plays the measurements for the tests underwater. As expected, it was Loon Copter's rotor is 40 A, and the maximum current drawn underwa-
found that, as the rpm of the propeller increase, the thrust and the ter at 800 rpm is 7 A. Therefore, no mechanical fatigue, temperature, or
power consumption also increase. Furthermore, the thrust to power motor controller overloading issues were found when the speed was
ratio increased until it reached a maximum rpm (180 rpm for underwa- limited to 800 rpm underwater. At higher speed, the brushless motor
ter and 4000 rpm for air on average) for each pitch angle. As shown in experienced cogging due to overload, and thus the speed controller of
subfigures (a)–(d), there is a proportional relationship between thrust, the motor shut down, which is the reason for limiting the maximum
power consumption and pitch angle in both media. Also, from subfig- rotational speed of the rotors. Piccoli and Yim (2016) suggested an
ures (e) and (f), we notice that the thrust to power ratio in aerial tests anticogging algorithm to reduce the effect of cogging torque on the
is proportional to pitch angle; on the other hand, the thrust to power brushless motor by improving the software of the motor speed con-
ratio has a contrary relationship with pitch angle in underwater tests. troller, which may be used in future development. Nevertheless, each
In subfigures (b), (d), and (f), the aerial rotor generates the cavitation of the vehicle's rotors could generate about 12 N thrust at 800 rpm
phenomenon at about 200 rpm rotor speed, and the experimental data and thus limiting the speed of the motor reduced the effect of this limi-
shows that the power consumption increased significantly, which nega- tation. Using the chosen power plant provided adequate thrust to fully
tively affected the thrust to power ratio. Subfigure (f) shows that pitch maneuver the vehicle when submersed. Lateral speed underwater may
angle 3.0″ was the best among the angles in terms of the force gen- be restricted due to the limited thrust and drag of the vehicle; how-
erated to power consumed, with maximum thrust to power ratio of ever, it was deemed reasonable considering the vehicle's capacity to
8 ALZU'BI ET AL .

F I G U R E 8 Propeller's data analysis in the air (left) and in the water (right). Subfigures (a) and (b) represent the thrust versus rotor speed, (c) and
(d) show the power consumption versus rotor speed, and (e) and (f) show the thrust to power ratio in air and underwater, respectively

surface and cover larger distances in air before diving again at a desired ATmega 2560 microcontroller and an Inertial Measurement Unit
location. (IMU).
ii) A low frequency, Futaba 72 MHz, RC transmitter, and receiver is
used to communicate with the Loon Copter underwater.
5 LOON COPTER ELECTRONICS AND iii) A water pump speed controller, Viper Marine 10, is used to control
PERFORMANCE the ballast system.

iv) Four brushless speed controllers, Castle edge 50, are used to con-
The Loon Copter's avionics system is comprised of the following:
trol the waterproof brushless motors.

The diagram of the avionics system wiring is provided in Figure 9.

i) A Multiwii pro open source flight controller is used to control The Mutiwii controller receives the flight command signals from the RC
multirotor RC models. The Multiwii hardware consists of a single receiver. Then, this controller generates the control signals to stabilize
ALZU'BI ET AL . 9

FIGURE 9 Loon Copter electronics wiring diagram

FIGURE 10 Loon Copter ballast system wiring diagram

and maneuver the Loon Copter. In addition, the water pump controller direction of the two-way valve. This switch generates three different
receives the command signal directly from the RC receiver to control values of the PWM: 1 ms, 1.5 ms, and 2 ms. When the water pump con-
the ballast system. troller receives a 1 ms PWM signal, it opens the valve and rotates the
water pump in a clockwise (cw) direction to pump the water into the
water cylinder. When it receives a 1.5 ms PWM signal, it turns off the
5.1 Depth control electronics water pump and closes the two-way valve. Lastly, when the controller
receives a 2 ms PWM signal, it opens the valve and rotates the water
The wiring diagram of the ballast system's depth control is shown in
pump in a counterclockwise (ccw) direction to pump out the water.
Figure 10. A 12 V power supply is used to power the water pump speed
controller, which is used to control the water pump and a two-way
valve. Both the water pump and the two-way valve are wired to the
5.2 Attitude control and stability
same output of the water pump controller.
The ballast system is designed to move the Loon Copter in the The Loon Copter has two control modes: an aerial mode and an aquatic
heave direction underwater. When the vehicle completes the air-water mode. In both modes, the Loon Copter is controlled remotely using the
transition and submerges in water, the amount of water in the water RC transmitter. The transmitter consists of six channels: four channels
cylinder can be adjusted to control the depth and to keep the vehicle to command the vehicle throttle, pitch, roll, and yaw. The remaining
suspended at the required depth. As shown in this figure, the speed two channels are two switches, one of which is used to switch between
controller receives the pulse width modulation (PWM) from the RC the air and water mode, whereas the other switch is used to control the
receiver to control the two-way valve and the water pump simultane- ballast system. An open source PD control algorithm was implemented
ously. A three-position switch is used to generate the required PWM to stabilize the vehicle in the aerial mode. However, the PD gains were
that controls the direction and the speed of the water pump and the altered to achieve stability in air. In the aquatic mode, an open loop
10 ALZU'BI ET AL .

simultaneously decreasing the angular velocities of motors M2 and M3 .

Roll is performed by increasing the angular velocity of motors M2 and
M4 and simultaneously decreasing the angular velocities of motors
M1 and M3 . Heave can be achieved using the ballast system while the
rotors are stationary, or it can be coupled with pitch.

5.2.1 Loon copter flight control

Figure 13 shows the Loon Copter's avionics control system. The sys-
tem includes two main components: the attitude PD controllers and
the four motor command mixers. The PD controllers are roll, pitch,
and yaw. The aerial altitude control is also open loop and is controlled
directly from the RC transmitter. The command mixing blocks are for
the front right, front left, rear right, and rear left motors (M1 , M2 , M4 ,
and M3 , respectively). For instance, to command the vehicle to roll
FIGURE 11 Loon Copter axes definition in aerial mode. xB , yB , and right, the mixing blocks will reduce the rotational speed of M1 and M4 ,
zB represent the roll, pitch, and yaw, respectively, of the Loon Copter's and increase that of M2 and M3 . The feedback states are received from
body frame
the onboard inertial measurement unit. Commands from the remote
pilot include the desired pitch, roll, and yaw angles, in addition to the
desired altitude. Each PD controller calculates the difference between
the feedback and the reference and tries to eliminate errors. The out-
put of each controller is mixed in the four mixing blocks to produce
the desired PWM signals and to send them to the motor controllers in
order to achieve the desired speed for each of the motors. Figure 14
shows the Pitch PD controller as an example of one of the PD con-
trollers used in the avionics system. Figure 15 illustrates the experi-
mental results of the Loon Copter hover flight test. Subfigures (a), (b),
and (c) represent the Loon Copter's roll, pitch, and yaw attitude test
results, respectively. The PD controller has to maintain the vehicle's
attitude angles at zero in flight within ±2◦ error. As shown in subfig-
ure (c), the yaw angle has some drift due to the vehicle's magnetometer
sensor noise. However, the Loon Copter is stabilized despite these dis-
FIGURE 12 Loon Copter's body frame underwater. The vehicle's turbances.
roll, pitch, yaw, surge, and sway are achieved using the rotors, while the
heave is performed using the ballast system in the direction of zB
5.2.2 Loon copter surfacing
The air mode control algorithm is also used for the vehicle's operation
control algorithm is used underwater since the stability of the vehicle
on the surface of water. The Loon Copter is positively bouyant. When it
is achieved by the ballast system.
lands on water, it floats and can be remotely controlled to move on the
The vehicle's body frame, shown in Figure 11, is specified with xB ,
surface of water.
yB , and zB axes. These represent the roll, pitch, and yaw axes, respec-
As shown in Figure 16, the maximum pitch angle (𝛼) for the Loon
tively. The Loon Copter is a cross-type quadcopter, that is, to control
Copter surfacing mode is 15◦ relative to WS when the hieght (l) of the
pitch and roll axes, all the rotors contribute in generating the forces and
vehicle is about 13 cm above the WS. The vehicle's maximum surfacing
moments necessary to stabilize and control the vehicle.
speed at these configurations was about 1.5 m/s.
As shown in Figure 11, M1 and M2 are located in the front of the
vehicle, whereas M3 and M4 are located in the rear of the vehicle.
M1 and M3 rotate in a ccw direction, and M2 and M4 rotate in a cw 5.2.3 Loon copter transition and suspension underwater
direction. The experimental results of the Loon Copter's air–water and water–air
The vehicle has 6 degrees of freedom (DOF) underwater. It has 5 transitions are shown in Figure 17. An IMU sensor was used to measure
DOF roll, pitch, yaw, surge, and heave and one DOF as sway is coupled the vehicle's tilt angle relative to the WS. Depicted on the left-hand
with yaw. As shown in Figure 12, when the four motors (M1 , M2 , M3 , side of Figure 17 are the vehicle's tilt angle measurements for the air to
and M4 ) are actuated with the same angular velocity, the vehicle moves water transition test, whereas the right-hand side displays the water to
in the surge direction. Pitch is achieved by increasing the angular air transition test measurements.
velocity of motors M1 and M2 and simultaneously decreasing the The water–air transition takes less time compared with the air–
angular velocities of motors M3 and M4 . Similarly, yaw and sway are water transition. As shown in the left figure, the vehicle takes about
achieved by increasing the angular velocity of motors M1 and M4 and 8 s to start submerging underwater because some parts of the vehicle
ALZU'BI ET AL . 11

FIGURE 13 Block diagram of aerial mode control algorithm

FIGURE 14 Pitch PD control block diagram

FIGURE 15 Loon Copter's hover flight. Subfigures (a), (b), and (c) illustrate the vehicle's pitch, roll, and yaw test results, respectively

are initially out of water. This requires the ballast system to pump in suspend at about 90◦ underwater. When the ballast system releases
enough water to make the density of the vehicle higher than the den- the water, the vehicle slowly restores its level position. As shown in the
sity of water. At 10 s, the vehicle starts tilting and submerging slowly; right figure, at 11 s, part of the vehicle exits the water, which affects the
when the vehicle becomes fully submerged at 15.8 s, it tilts quickly. The relative positions of the center of buoyancy and the center of gravity.
overshoot apparent in Figure 17 occurred because an extra amount of This makes the vehicle's water to air transition instantaneously unsta-
water was pumped into the water cylinder which makes the vehicle ble. At approximately 12 s, the vehicle's rotors were pulsed to restore
tilt to 106◦ relative to the WS. This was adjusted by releasing a small the transition stability, which makes the Loon Copter float on the
amount of water from the water cylinder, which makes the Loon Copter surface of the water to its level position. After the transition to air is
12 ALZU'BI ET AL .

FIGURE 16 Loon Copter maximum surfacing angle (𝛼) relative to water surface. The vehicle's maximum angle depends on the height (l)

FIGURE 17 The Loon Copter air–water transition (left) and water–air transition (right): vehicle's tilt angle in degrees versus time in seconds

completed, the vehicle is buoyant and the four propellers of the drone command the vehicle to roll right, the mixing blocks will reduce the
clear the water, allowing for a clean takeoff. rotational speed of M1 and M3 and increase that of M2 and M4 .
When the air–water transition complete, the Loon Copter naturally The experimental results of the Loon Copter's yaw step response
suspend underwater at a required depth using the ballast system. are shown in Figure 20. Depicted in the figure are the vehicle's yaw
Figure 18 illustrates the vehicle's suspension test results. Subfigures angle measurements. The vehicle has an overdamped response due to
(a), (b), and (c) represent the vehicle's pitch, roll, and yaw angles water resistance. As shown in the figure, the rising time is about 2 s
underwater, respectively. As shown in the subfigures reference pitch, with about one degree of steady state error.
roll, and yaw represent the required angles underwater. The IMU
measurements illustrate that there are small differences between
the required and the measured angles in a range of ±3◦ . Therefore,
the ballast system is capable to stabilize the vehicle's attitude angles 6 FUTURE WORK
underwater.
The Loon Copter is a successful proof-of-concept vehicle that high-
lighted multiple possible areas of future work. The range of wireless
5.2.4 Loon copter underwater maneuver communication underwater is an obvious challenge. This may be
While the Loon Copter is underwater, its rotors are perpendicular to addressed by investigating acoustic or optical communication tech-
the WS. Figure 19 is a block diagram illustrating the control algorithm. niques, or alternatively, by outfitting the drone with sensors such as
Because the ballast system ensures the vehicle's stability, an open loop sonar, magnetometer and pressure sensors, and control algorithms
control algorithm was developed. The inputs of the command mix- to make it autonomous for inspection of submerse structures, such
ing blocks are received directly from the RC transmitter without any as petroleum platforms, ship hulls, and gas pipelines. The current
PD control corrections. The command mixing blocks are for the right shape of the vehicle can be improved as well. The vehicle's airframe is
and left motors (M1 , M4 , and M2 , M3 , respectively). For instance, to constructed from off-the-shelf components, which added some
ALZU'BI ET AL . 13

FIGURE 18 Loon Copter's suspension underwater. Subfigures (a), (b), and (c) illustrate the vehicle's pitch, roll, and yaw angles test results,
respectively

FIGURE 19 Block diagram of aquatic mode control algorithm

penalty to the vehicle's weight, and making the vehicle waterproof speed and accuracy of transition. This will require software modifica-
increased the surface area of the vehicle and, as a result, increased tions and additional hardware such as an IMU sensor to keep track
the vehicle drag. Therefore, there is the opportunity to redesign the of the vehicle's tilt angle. To improve the overall performance of the
vehicle's structure to reduce its weight and drag in air and underwater, controller, we are developing a dynamic model to simulate the Loon
which could potentially reduce the vehicle's power consumption and Copter's aerial and aquatic operation, as well as the transition between
thus increase operation time. As for the control aspect, we will be air and water. This will include determine the transfer functions of the
developing an algorithm to control the depth of the vehicle underwa- vehicle's rotors, and the lift and drag coefficients in both media, and the
ter. This will involve additional pressure sensors and modifications in mathematical modeling of the ballast system. As for safety aspect, the
the software. Loon Copter needs a failsafe system to prevent the vehicle from sink-
Currently, the air–water and water–air transition is operated ing in case of water leak or communication loss between the vehicle
remotely. A closed loop controller will be required to improve the and the remote.
14 ALZU'BI ET AL .

FIGURE 20 Loon Copter's yaw step response underwater

7 CONCLUSION Budiyono, A. (2009). Advances in unmanned underwater vehicles tech-

nologies: Modeling, control and guidance perspectives. Indian Journal of
Marine Sciences, 38(3), 282–295.
This publication explored the design, control, and implementation of
Drews, P., Neto, A., & Campos, M. (2009). A survey on aerial sub-
our novel hybrid aquatic–aerial quadrotor. The design employs only
mersible vehicles. IEEE/OES Oceans International Conference, Bremen,
four brushless DC motors with four counter-rotating fixed pitch pro- Germany.
pellers. The aerial and underwater maneuver is achieved by manipulat-
Drews, P., Neto, A., & Campos, M. (2014). Hybrid unmanned aerial underwa-
ing the rotational speeds of the propellers at each rotor. While in aerial ter vehicle: Modeling and simulation. Paper presented at the IEEE/RSJ
mode, the Loon Copter takes flight as a conventional aerial quadcopter International Conference on Intelligent Robots and Systems, Chicago,
and uses four rotors to control the roll, pitch, yaw, and altitude of the USA.

drone. While in aquatic mode, the vehicle utilizes the rotors for under- Edwards, D. (2014). Updates on NRL'S autonomy research. Spectra the Mag-
azine of the Navy's Corporate Laboratory, 4, 6–8.
water maneuver and the ballast system for depth control. The Loon
Copter achieves a novel seamless air–water and water–air transition Fabian, A., Feng, Y., Swartz, E., Thurmer, D., & Wang, R. (2012). Hybrid aerial
underwater vehicle (MIT Lincoln Lab). SCOPE Projects, Retrieved from
using a ballast system. This improves the vehicle's stability underwa-
https://www.olin.edu/collaborate/scope/projects/2011_12/mit_ll
ter and reduces the power consumption. Experimental results demon-
Ferreira, B., Matos, A., Cruz, N., & Pinto, M. (2010). Modeling and control of
strate that the proposed design produced a fully operational proto- the MARES autonomous underwater vehicle. Marine Technology Society
type with stable flight, operability on a WS, and maneuver capabilities Journal, 44(2), 19–36.
underwater. Hopkins University Applied Physics Laboratory. (2016). New
UAV can launch from underwater for aerial missions [Press
release]; Retrieved from https://www.jhuapl.edu/newscenter/
ACKNOWLEDGMENTS
pressreleases/2016/160317.asp
The authors would also like to thank Ahmad Kafrouny, Ghassan Abed,
Iscold, P., Pereira, G., & Torres, L. (2010). Development of a hand-launched
and Benjamin Nold for their help in testing the Loon Copter. We also small uav for ground reconnaissance. IEEE Transactions on Aerospace and
appreciate help from personnel of the Campus Recreation Center at Electronic Systems, 46(1), 335–348.
Oakland University. Izraelevitz, J., & Triantafyllou, M. (2014). Adding in-line motion and model-
based optimization offers exceptional force control authority in flapping
ORCID foils. Journal of Fluid Mechanics, 742, 5–34.
Kerwin, J. (1986). Marine propellers. Annual Review of Fluid Mechanics, 18(1),
Hamzeh Alzu'bi http://orcid.org/0000-0002-7692-7190
367–403.
Lock, R., Vaidyanathan, R., & Burgess, S. (2013). Impact of marine locomo-
tion constraints on a bio-inspired aerial-aquatic wing: Experimental per-
REFERENCES formance verification. Journal of Mechanisms Robotics, 6(1), 11001.
Bouabdallah, S. (2007). Design and control of quadrotors with appli-
Maia, M., Soni, P., & Diez-Garias, F. (2015). Demonstration of an aerial
cation to autonomous flying. Ph.D. thesis, EPFL Research Institu-
and submersible vehicle capable of flight and underwater nav-
tion in Lausanne, SwitzerlandRetrieved from Retrieved from https://
igation with seamless air-water transition. arXiv, Retrieved from
infoscience.epfl.ch/record/95939
https://arxiv.org/abs/1507.01932
ALZU'BI ET AL . 15

Piccoli, M., & Yim, M. (2016). Anticogging: Torque ripple suppression, mod- United States Navy. (2008). The Submarine. Los Angeles, CA: Periscope Film.
eling, and parameter selection. The International Journal of Robotics Wang, W., Chen, X., Marburg, A., Chase, J., & Hann, C. (2008). A low-cost
Research, 35(1–3), 148–160. unmanned underwater vehicle prototype for shallow water tasks. Paper
Ranganathan, T., Thondiyath, A., & Kumar, S. (2015). Design and Analysis presented at the IEEE/ASME International Conference on Mechatronic
of an Underwater Quadrotor - AQUAD. Paper presented at the IEEE and Embedded Systems and Applications, Beijing, China.
Underwater Technology (UT) Conference, Chennai, India. Yang, X., Wang, T., Liang, J., Yao, G., & Liu, M. (2015). Survey on the
Ridao, P., Batlle, J., & Carreras, M. (2001). Model identification of a low- novel hybrid aquatic–aerial amphibious aircraft: Aquatic unmanned
speed UUV. Paper presented at IFAC Conference on Control Applica- aerial vehicle (Aqua UAV). Progress in Aerospace Sciences, 74, 131–
tions in Marine Systems, Glasgow, Scotland. 151.
Schmid, K., Lutz, P., Tomić, T., Mair, E., & Hirschmüller, H. (2014).
Autonomous vision-based micro air vehicle for indoor and outdoor nav-
igation. Journal of Field Robotics, 31(4), 537–570.
Siddall, R., & Kovač, M. (2014). Launching the AquaMAV: Bioinspired design How to cite this article: Alzu'bi H, Mansour I, Rawashdeh O.
for aerial-aquatic robotic platforms. Bioinspiration & Biomimetics, 9(3),
Loon Copter: Implementation of a hybrid unmanned aquatic–
031001.
aerial quadcopter with active buoyancy control. J Field Robotics.
Tan, Y., Siddall, R., & Kovac, M. (2017). Efficient aerial–aquatic locomotion
2018;1–15. https://doi.org/10.1002/rob.21777
with a single propulsion system. IEEE Robotics and Automation Letters,
2(3), 1304–1311.