Quadcopter Design for Medicine Transportation

in the Peruvian Amazon Rainforest

Juan A. Paredes, Juan Acevedo, Hector Mogrovejo, Jorge Villalta, Roberto Furukawa
Perú IEEE EMBS – Pontificia Universidad Católica del Perú, Lima, Perú
[email protected]


Abstract— Medicine transportation in the Peruvian Amazon This paper proposes a design for a UAV of the quadcopter type
Rainforest is difficult due to environmental hazards and lack of for medicine transportation in the before-mentioned area. Due
proper land transportation roads, which leads to insufficient and to its intrinsic ability to overcome terrestrial obstacles, its speed
improper health services in this area. In order to overcome the
and its simplicity in design, a quadcopter would be most
terrestrial impediment1s mentioned above and remedy this
situation, a quadcopter UAV system is proposed to be used for suitable for this kind of task. The design takes into account the
aerial transportation of medicine. This paper will explain the transportation of low-weight cargo and a structure impermeable
mechanical, electrical and control design considerations taken into to water (in case rain suddenly starts pouring).
account for the implementation of the proposed system.
II. SYSTEM OVERVIEW
Keywords— Aerospace Technology, Aerospace Control Systems,
Unmanned Aerial Vehicle, FEM Analysis, Sensor Fusion. The proposed system is a quadcopter suitable for long-range
transportation of medicine. Figure 1 displays the basic
components of said system. The flight controller regulates the
I. MOTIVATION rotation speed of the rotors via Electronic Speed Controllers
(ESC) in order to control the thrust and torque that these exert
T HE UAV industry has seen increased technological
developments in recent years, thus creating more accessible
flying autonomous vehicles. With it, companies around the
on the system and, consequently, its mobility. The desired
attitude and position is determined by instructions send by a
user via wireless communications or preprogrammed before
world identified commercial applications for said technology, take-off.
namely as low-cost transportation of low-weight cargo
vehicles. Companies such as Google, Amazon and DHL have
developed their own prototypes with the goal of creating
commercial networks based around UAV transportation. Other
companies have focused on more humanitarian applications.
For example, Matternet, in association with Doctors Without
Borders, experimented with delivery drones in Papua New
Guinea [1] with the purpose of delivering blood samples across
this region for faster tuberculosis analysis.

In light of this situation, it is proposed that a UAV would

alleviate the current medicine transportation situation in the
Peruvian Amazon Rainforest. In this area, there exist
geographical barriers that consist of “the difficulty that entails Fig. 1. An overview of the UAV proposed system. The
getting access to health services, due to distance, transportation communications, energy distribution and control feedback systems are
and seasonal geographical isolation”, as stated in the “Informe displayed and will be explained in the next sections of the document.
Defensorial Nº 134” [2]. Muddy roads and the great distances
between town make it difficult to establish an efficient The quadcopter system is also composed by an array of sensors
terrestrial medicine distribution network. Moreover, according which allow the flight controller to receive information
to the results of the Second Census of Native Communities in regarding the state of the system. Also, a couple of simple
the Amazon Region conducted in 2007 [3], in an interval of 12 electrical components are included, such as a bicolor LED to
months, approximately 20,679 patients were diagnosed with signal the state of the flight and low battery condition, and a
parasites, 11,232 with diarrhea, 8,026 with lack of nutrition and push button that may be pressed to let the system know that the
medicine has been retrieved. The LiPo batteries supply the
anemia, and 7,202 with malaria which demonstrates the need of
entire system with electrical energy. A power distribution board
a better transportation system.
is employed to help with this task.

978-1-5090-1341-8/16/$31.00 ©2016 IEEE

A couple of requirements are needed for the system to perform Considering a payload of about 2 kg for the medicine, the
optimally in the given conditions. First of all, weight selected devices were the U8 Pro Motor - KV170 (which
distribution along the structure of the quadcopter should be present an additional benefit by being waterproof), 26” x 8.5”
symmetrical so that stabilization procedures may be simplified. carbon fiber propellers and the AIR 40 ESC, all products of T-
In addition, the system should be impermeable to water so that MOTOR.
it can sustain itself in case rain starts pouring in. It should be
noted that the objective here is not to fly in rainy conditions but For the remaining lift and hover time calculations, a battery
to persevere through them in case they happen spontaneously. with 20 Ah capacity and 2.378 kg mass was considered, as well
Other requirements will be reviewed as they become relevant as a safety factor of 0.85, a throttle of 65% and an Ir of 1 A. The
through this document. corresponding current consumption and thrust were obtained
from the tables provided by T-Motor [4]; a portion of said tables
III. THROTTLE CALCULATION is displayed in Table 1. Using this parameters, a hover time of
The main three components of the quadcopter propulsion 31.29 minutes was calculated as well as a remaining lift of
system are the ESCs, the brushless motors and the propellers. 1.836 kg. This last constraint will be addressed in the
A combination of these three elements will determine the mechanical design section.
payload the quadcopter will be able to lift and the amount of
TABLE I
time it may be capable of hovering above the ground. The THRUST DATA TABLE PROVIDED BY T-MOTOR
battery capacity is also an important factor to take into account
in these calculations. Throttle Current (A) Thrust (g)
50% 4.4 1420
Taking into account a battery with capacity E in mAh and an 65% 7.9 2220
energy safety factor of SF, the real battery capacity E R may be 75% 10.6 2780
calculated in the following fashion:
IV. MECHANICAL DESIGN
ER = E × SF. (1) The proposed mechanical design aims to be as light as possible
in order to meet the mass constraint set in the throttle
Normally, companies that sell aeronautic components provide calculation, be symmetrical in order to simplify the control
tables that display the resulting thrust (T) of applying a certain requirements and rigid enough to prevent either material failure
quantity of amperage (I) to a specific motor and propeller or (in the case of the arms) bending. Said quadcopter is
combination. Such thrust is normally displayed in terms of presented in figure 2.
grams (g). In order to lower battery consumption, a low thrust
should be used in these calculations. A safety factor should also
be applied to this thrust value as well, which will be assumed to
be the same as the energy one. This way, real thrust (T R) can be
calculated like this:

TR = T × SF. (2)

In order to calculate remaining lift (L), one must subtract the

weight of the batteries (mB), the rotor weight (mr) and the
payload weight (mP) from the total thrust. In the case of a
quadcopter, the formula would look like this: Fig. 2. Proposed frame design for the Quadcopter.

L = 4TR – mB – mr – mP. (3) A. Main Hub

The remaining lift represents the maximum amount of weight The main hub is composed by two carbon fiber plates with a
in grams that the frame may weight, which, in this case, will be thickness of 3 mm assembled by four boom clamps; these
addressed in the mechanical design section. clamps will also keep the carbon fiber tubes assembled to the
In order to estimate the total hover time (th), one must get the main hub through a tight grip. Allen bolts are used to perform
motor current (Im) correspondent to the thrust used before and this assembly. The clamps and booms can be bought, but the
the current consumed by the rest of electronic components (I r) plates will have to be machined, preferably with a CNC mill
(in preliminary calculations, this last value must be assumed). and carbide or PCD milling drills. Figure 3 displays this
The calculation to be performed is the following: structure.

ER The frame is designed to contain the electronic components

th = , (4) needed to control the drone, such as sensors and controllers. For
4Im + Ir
that purpose, a top protection structure made of ABS
in which the time is given in hours. (Acrylonitrile Butadiene Styrene) with a thickness of 1 mm will
protect the sensors from rain, dust and wind forces. This
component could be manufactured by a 3D printer, although
further procedure will have to be performed to the printed D. Motor Mounts
structure in order to guarantee impermeability, such as painting
to coat the surface of the structure. The mounting is composed by 3 mm carbon fiber plates
assembled to each arm by two boom clamps. Each plate should
be designed to fit the U8 Motor’s mounting holes. It is
encouraged to drill a circular hole in the middle to allow the U8
rotor to freely rotate and to remove as much material as possible
to reduce the overall weight.

E. Medicine Package

The bottom part of the drone is an ABS plastic box with a 2 mm

thickness held together by two toggle latches. The box is used
to carry medicine and two Li-Po 10 Ah batteries on its top. It is
assembled to the main hub by aluminum spacers. In order to get
its contents, the latches have to be opened so that the lower part
Fig. 3. Central hub without upper case. Its purpose is to assemble all of the box becomes unfastened while the top part remains
the rotors through its clamps and to house sensible electronics, such as attached.
the flight controller and the sensors.
F. Total Mass and Symmetry
B. Arms
A 3D model of the proposed quadcopter was created in
Each of the four arms consists of a carbon fiber tube with 20 Autodesk Inventor to determine the mass and inertia moments
mm diameter, 2mm of wall thickness and 500 mm of length, of the frame. The estimated frame mass is 1.6 kg, less than the
enough to prevent the propellers from hitting the main hub. established limit. Also, since the resulting products of inertia
These are assembled to the main hub through boom clamps and are low compared to the principal moments of inertia, it can be
are used to mount the landing gear and the rotors. concluded that the frame is symmetrical enough not to
complicate the design of a control algorithm. These inertia
In theory, the arms will sustain a maximum force of 38.06 N moments are shown in Table 2.
due to the thrust of the chosen rotor combination. Through an
FEM simulation of these forces, it was determined that, under TABLE II
the effect of that force, the arms will have a safety factor of 7.42, QUADCOPTER FRAME INERTIA MOMENTS (IN KG.M2)
meaning the arms won’t suffer from material failure in these Ixx Iyy Izz Ixy Ixz Iyz
conditions. 5.3832 5.3831 9.4120 -1.1832 -1.1363 1.1860
× 10-2 × 10-2 × 10-2 × 10-3 × 10-5 × 10-5
C. Landing Gear
G. FEM Analysis
Landing legs are mounted to each arm, each one having two
clamps, two carbon fiber plates and three aluminum standoffs In order to verify that the proposed mechanical designed won’t
as part of their structure. Landing plates are 2 mm wide and break during a flight, FEM analyses were performed on the 3D
have certain parts removed to lower the weight of the structure model of the frame under critical conditions (maximum rotor
and maintain the necessary rigidity for landing operations. thrust) for testing, considering the carbon fiber as a brittle
material. The FoS along one of the arms is shown in figure 4.
The Factor of Safety will determine whether or not the structure
will be resistant enough. The results from the analyses are
displayed in Table 2. Due to the FoS having a value greater than
3 on all the components, it can be concluded that the quadcopter
will resist the greatest thrust the rotors are capable of.

TABLE III
FACTOR OF SAFETY UNDER CRITICAL CONDITIONS
Component Material FoS
Arm Carbon Fiber 3.93
Top central plate Carbon Fiber 4.45
Fig 4. Safety factor of frame arms after performing FEM simulation. Lower central plate Carbon Fiber 3.16
The side of the tube under the clamps is under the most pressure during Standoff connector Aluminum 6061 10.44
a flight. Top box package Plastic ABS 4.01
 V. ELECTRICAL SYSTEM distance to the floor and perform landing operations more
In this section, the main electrical components will be precisely. Most of these systems send information through
mentioned as well as the way in which they are all connected PWM signals.
and how they contribute to the functioning of the quadcopter
system. The block diagram displayed in Figure 5 may be used In order to control the quadcopter and use the information
as a reference for the rest of this section since it summarizes its obtained by the sensors, the controller must send signals to
contents. control the generated lift. For that purpose, PWM signals will
be sent to ESCs, which in turn will regulate DC voltage to
control brushless motors.

In order to maintain communications with a ground station, a

wireless module must be installed in the quadcopter, which will
establish serial communication with the controller to send
telemetry messages. Signal strength will be determined by the
chosen signal frequency, module signal gain and antenna gain.

Another important system is the voltage sensor, which is

connected to the quadcopter and can measure how much energy
is left in it. If a LiPo battery is discharged too much it may
become permanently damaged, so this system is used to prevent
this and look for a recharge station when the battery level
becomes low enough. The controller reads this value through
an analog input.
Fig. 5. Block Diagram of Quadcopter Electrical System. The main
The notification mechanisms are used for notifying and getting
electrical connections are shown here in a simplified manner
information from the operators as stated in the system overview
The most important part of the electrical system would be the section. They send signals to the controller through its GPIO.
flight controller, which is in charge of recollecting
environmental information, sending control signals, handling The Arduino Due will power most of these components, while
communications among other tasks. Due to its flexibility, it and motors will be directly energized by the batteries through
amount of input and output connection options and processing a Power Distribution Board. This board should be designed so
power, an Arduino Due was chosen for this assignment. that it can handle the required amount of current that all the
electrical components will require.
Several sensors will be needed by the controller for it to be
capable of establishing a reliable feedback loop. Four main VI. CONTROL SYSTEM
sensors are needed for this task, which are the following: In order to study the dynamic model, it’s important to have a
simple representation of the quadcopter. Therefore, the system
 IMU: This sensor includes an accelerometer, a gyroscope shall be represented by a central spherical body coupled to four
and a magnetometer and it allows the controller to get cylindrical arms; at the end of each of these, the rotor will be
information about the attitude and heading of the quadcopter. attached, which will allow the actuation of the system. This is
Most IMUs are connected through an I2C bus. illustrated in figure 6.

 Altimeter: This sensor determines the current altitude over

sea level of the quadcopter from absolute pressure readings.
It would be advisable to use one capable of I2C
communication.

 GPS Module: This device will allow the quadcopter to get Fig. 6. Quadcopter Free Body Diagram
its geographical location and heading in the world, granted
that it doesn’t get obstructed by walls and receives signals Four control variables will be defined to control pitch (ϕ), roll
from at least four satellites. Serial communication is (θ) and yaw (ψ), which will be T, τϕ , τθ and τψ respectively.
commonly used to communicate with these devices.
T = F1 + F2 + F3 + F4 (5)
 Ultrasound sensor: This sensor uses ultrasonic waves in τϕ = L (−F2 + F4) (6)
order to measure the distance to the closest solid object. The τθ = L (−F1 + F3) (7)
quadcopter will use this sensor in order to measure its τψ = −M1 + M2 − M3 + M4 (8)
Considering the thrust and moments to be proportional to the
squared rotational velocity, the following can be established:

F = b × ω2 (9)
M = q × ω2 (10) Fig. 7. Proposed Control Feedback Loop

With these equations, a transformation matrix that relates the Controlling the altitude and the yaw required a single feedback
previous variables with the squared rotational velocities can be loop, although an offset must be included in the altitude
obtained, which will be used by the flight controller to send the controller for it to keep countering the quadcopter weight.
correct PWM signals to the ESCs and control the rotation of the Horizontal position control requires two control loops; the outer
motors. loop determines the required pitch/roll according to the position
2
T b b b b ω1 error and the inner loop determined the required torque
τϕ 0 −b L𝑖 0 b L𝑖 ω22 according to the pitch/roll error. This is illustrated in Figure 8.
[τ ] = [ ] . (11)
θ −b L𝑖 0 b L𝑖 0 ω23
τψ −q q −q q [ω2 ]
4

The Euler equations will be used to outline the dynamic

equations of the quadcopter. The documents [5], [6], [7] and [8]
were used as reference for this section. The equations are the
following:

Ẍ 0 sϕ∗ sψ+ cϕ∗ sθ∗cψ

𝑇
[Ÿ] = −g [0] + [− s ϕ ∗ c ψ + c ϕ ∗ s θ ∗ s ψ] (12)
𝑚
Z̈ 1 cϕ∗ cθ
Fig. 8. Proposed Control System
ϕ̇ 1 sin ϕ tan θ cos ϕ tan θ 𝑃
[ θ̇ ] = [0 cos ϕ − sin ϕ ] [𝑄 ] (13) The values for the simulation were taken from the information
sin ϕ cos ϕ gained from the 3D model of the structure and documentation
ψ̇ 0 𝑅
cos θ cos θ from the rotor. The model’s values are m = 5.978 kg, L = 0.5
(𝐼𝑦𝑦 −𝐼𝑧𝑧 ) τϕ
𝑄̇ 𝑅̇ m, Ixx = 0.0746 kg.m2, Iyy = 0.0746 kg.m2, Izz = 0.111929 kg.m2,
𝐼𝑥𝑥 𝐼𝑥𝑥
𝑃̇ (𝐼𝑧𝑧 −𝐼𝑥𝑥 ) τθ b = 5.1248*10-6 N/RPM2 and q = 2.8750*10-6 N.m/RPM2.
[𝑄̇ ] = 𝑃̇ 𝑅̇ + 𝐼𝑦𝑦 . (14)
𝐼𝑦𝑦
𝑅̇ (𝐼𝑥𝑥 −𝐼𝑦𝑦 )
τψ

[ 𝑃̇ 𝑄̇ ] [ 𝐼𝑧𝑧 ]
𝐼𝑧𝑧

In these equations, X, Y and Z represent the 3D location of the

quadcopter and PQR represent its angular velocity along the
body coordinate system.

A PID controller will be proposed to control this dynamic

system. Since this controller requires a linearized plant to be
tunes, the dynamic equations stated before will be linearized
around the hover operation point, in which ϕ = 0°, θ = 0°,
ψ = ψo and T = m ∗ g.
τϕ
Ẍ = g θ (15) ϕ̈ = (18)
Ixx
τθ
Ÿ = −g ϕ (16) θ̈ = (19)
Iyy
T τψ
Z̈ = (17) ψ̈ = (20) Fig. 9. Step response of dynamic system with PID controller
m Izz
implemented. [Up-Left] Altitude control (Z). [Up-Right] Yaw control.
For this system, the feedback loop would require the sensors to [Down-Left] Pitch/Roll control. [Down-Right] Horizontal Position
Control (X/Y).
get the process variables, which would go through a state
estimation phase to estimate and filter the required variables The simulation results are displayed in Figure 9. Most systems
and compare these with the reference values to get the error. converge to their set points in an acceptable amount of time.
This error would be used as input for the controller, whose The horizontal position control shows an underdamped
outputs would have to be decoupled into rotor angular velocities behavior and takes longer to converge to the set point due to the
in order to affect the system. This is represented by the diagram nature of the controller with two loops. Through these
in Figure 7. simulations, it has been demonstrated that the proposed design
can be controlled through a simple control algorithm like the ACKNOWLEDGMENTS
PID controller. The team would like to thank professor Carlos Saito from
Pontificia Universidad Católica del Perú for his advice on UAV
VII. OBSERVATIONS design and component selection.
The present paper aims to present a design for a medicine
transport system. However, in order for the transportation REFERENCES
service to work other devices will be required. For example, a [1] J. Leber, «Doctors Without Borders Is Experimenting With Delivery Drones
monitoring center may be required to overview the condition of To Battle An Epidemic,» 16 Octubre 2014. [Online].
Available: http://www.fastcoexist.com/3037013/doctors-without-borders-is-
all the quadcopter in flight, as well as a communications experimenting-with-delivery-drones-to-battle-an-epidemic.
antennae distributed throughout the area of operation in order
to coordinate activities in a more efficient manner. More [2] Programa de Comunidades Nativas, «“La Salud de las Comunidades
importantly, to guarantee an efficient deployment, automatic Nativas: Un reto para el Estado - Informe Defensorial N° 134”,» Defensoría del
Pueblo, Lima - Perú, 2008.
recharge centers will have to be deployed in strategic locations
for the quadcopters to recharge when their battery is low on [3] Dirección Nacional de Censos y Encuestas, «II Censo de Comunidades
power. This could prove to be challenging due to the difficulty Indígenas de la Amazonía Peruana 2007 – Tomo 1,» Instituto Nacional de
of performing precision landings and the requirement of the Estadística e Informática (INEI), Lima-Perú, 2008.
station to be protected against rain. These problems will be [4] T-Motor, «U8 - Pro - Efficiency Type,» [Online]. Available:
assessed in future studies. http://www.rctigermotor.com/html/2013/Efficiency-Type_1219/176.html.

[5] C. Powers, D. Mellinger and V. Kumar, «Quadrotor Kinematics and

Regarding the proposed dynamic model, many factors weren’t Dynamics» in Handbook of Unmanned Aerial Vehicles, New York, Springer
considered which would influence the system under realistic Reference, 2015, pp. 307 - 328
conditions. For example, drag forces weren’t modeled since
these depend on the body geometry and current airspeed, which [6] T. Bresciani, «Modelling, Identification and Control of a Quadrotor
Helicopter» Department of Automatic Control, Lund University, Lund, 2008.
would prove to be complicated; also, the gyroscopic forces
occasioned by the motors weren’t considered since their [7] P. Pounds, R. Mahony and J. Gresham, «Towards Dynamically-Favourable
modelling required the knowledge of the motor inertia. Future Quad-Rotor Aerial Robots» Australian National University, Canberra.
research will aim to construct more realistic models.
[8] D. Hartman, K. Landis, M. Mehrer, S. Moreno, J. Kim «Quad-Sim:
Mathematical Model Documentation» [Online]. Available:
VIII. CONCLUSIONS http://www.mathworks.com/matlabcentral/fileexchange/48053-quad-sim
A quadcopter design for medicine transportation in the
Peruvian Amazon Rainforest was proposed in this paper. The
reasons behind its design were explained through the entire
document and the following conclusions were established:

 The proposed thrust system is capable of attaining a hover

time of 31.29 minutes with a frame with 1.836 kg of mass
with an 85% Factor of Safety.

 The proposed frame structure is capable of keeping a

stabilized flight while lifting 2 kg of cargo and housing the
entirety of the electrical components needed for the
quadcopter to fly autonomously. Also, a top protections was
designed to protect the electrical components from rain, dust
and other hazards.

 The proposed electrical system, that included a sensor array,

brushless motor speed controllers and other peripherals, will
allow the creation of a feedback loop to control the
quadcopter through the system controller.

 The proposed PID controller is capable of controlling the

designed quadcopter in ideal conditions, although better
results may be attained through further tuning.