See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/317380270

Design, Development and Flight Testing of a Novel Quadrotor Convertiplane

Unmanned Air Vehicle

Conference Paper · May 2017

DOI: 10.4050/F-0073-2017-12294

CITATIONS READS
8 3,029

5 authors, including:

Abhishek Abhishek M. Rama Krishna

Indian Institute of Technology Kanpur Indian Institute of Technology Kanpur
69 PUBLICATIONS 734 CITATIONS 3 PUBLICATIONS 8 CITATIONS

SEE PROFILE SEE PROFILE

Sourav Sinha Joydeep Bhowmik

Virginia Tech Indian Institute of Engineering Science and Technology, Shibpur
14 PUBLICATIONS 17 CITATIONS 10 PUBLICATIONS 20 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Abhishek Abhishek on 07 June 2017.

The user has requested enhancement of the downloaded file.

 Design, Development and Flight Testing of a Novel Quadrotor Convertiplane
Unmanned Air Vehicle
Abhishek A. M. Rama Krishna
Assistant Professor Undergraduate Student

Sourav Sinha Joydeep Bhowmik Debopam Das

Undergraduate Student Graduate Student Professor
Indian Institute of Technology Kanpur
Kanpur - 208016, India

ABSTRACT
This paper describes the design, development and flight testing of a small novel quadrotor convertiplane Unmanned Air
Vehicle (UAV) which has four propellers driven by individual brushless DC motors mounted at the tip of tandem wings.
The segment of the wing under the propeller downwash is tilted with the propeller to improve the hovering efficiency
of the UAV. Tandem low and high wing design is chosen to mitigate the interference losses due to the influence of
fore wing and propeller downwash on the aft wing and the propellers. The motor and propeller suitable for the design
are chosen through systematic performance measurements carried out using a hover test stand and wind tunnel tests.
The effect of tilting the wing segment below the rotor is studied by modifying the hover stand and including the wing
in the setup. A PID controller is developed, first in simulation and then implemented on a PixHawk autopilot board
to stabilize the prototype during hover. The performance of the controller is demonstrated through flight test of the
prototype. For the current configuration and size, the current design, with the wing in propeller downwash tilting with
the rotor, offers 50% higher power loading in hover in comparison to a conventional quad tiltrotor with no wing tilting.
It is observed that the propeller looses 40% of its thrust due to the download penalty on the wing when no part of
the wing is tilted with the propeller. The tilting of the wing portion equal to the radius of the propeller increases the
power loading of the propeller making it nearly same as the isolated rotor. The current design is estimated to consume
50% less power during airplane mode in comparison to the quadrotor mode thereby establishing the advantage of the
proposed convertiplane vehicle over regular quadrotor UAVs.

INTRODUCTION tail-sitter (Refs. 9, 10), quad tiltrotor (Ref. 11), and quad-
tiltwing (Ref. 12) concepts.
The tremendous growth in Small Unmanned Aerial Systems
(SUASs) technology has enabled its application to wide range Each of these VTOL concepts have their own merits and de-
of military as well as civilian applications such as surveil- merits. For example, the monoplane tail sitter concept al-
lance, surveying, crop monitoring, precision agriculture. Sev- though simple to make, has poor control authority during
eral of these applications require the SUAs to have hovering hover due to use of fixed pitch propellers and conventional
and vertical flight capabilities. Various configurations aiming aircraft controls which are less effective when working in ro-
for hovering flight, ranging from flapping wing, rotary wing tor downwash. The transition from hover to forward flight and
to cycloidal rotor concepts have been tried and studied by var- back is quite challenging. The quad tiltrotor design has good
ious researchers (Refs. 1–5) in great detail. Some of these de- control authority during hover and transition and its flight con-
signs include configurations such as the Micro Coaxial Rotor- trol is simpler than that for monoplane tail-sitter. But, the rotor
craft (MICOR) developed at University of Maryland (Ref. 2), wing interaction in hover and the aerodynamic interaction be-
muFly (Ref. 3), a prop assisted mono blade RoboFly or the tween fore and aft wings result in loss of efficiency. It is inter-
Samara Micro Air Vehicle (Ref. 4), Cycloidal rotor Micro Air esting to note that the idea of quad-tiltrotor is several decades
Vehicle (Ref. 5) and quadrotor (Ref. 6). Beyond these, several old when Curtis-Wright X-19 was built as full scale quad-
innovative hybrid Vertical Take Off and Landing (VTOL) ca- tiltrotor VTOL aircraft in early 1960s. The X-19 had high-
pable fixed wing type concepts have also been attempted, such mounted tandem wings with two propellers mounted on each
as monoplane wing tail-sitter (Refs. 7, 8), biplane quadrotor wing which could be rotated through 90◦ allowing the aircraft
to take off and land like a helicopter. Later, Bell developed its
Presented at the AHS International 73rd Annual Forum & model D-322 as a quad tiltrotor concept in 1979. They dis-
Technology Display, Fort Worth, Texas, USA, May 9–11, 2017. closed a Quad TiltRotor (QTR) design in 1999 for heavy lift
Copyright c 2017 by AHS International, Inc. All rights reserved. applications based on the design of V-22 Osprey (Ref. 13).
1
An experimental study was performed at University of Mary- servo to change the role of the propellers from lifter to thruster
land and it was observed that the effect of download on the as shown in Fig. 1(b).
wing during hover Out of Ground Effect (HOGE) was of the
order of 10% of the total thrust (Ref. 14). This was further
confirmed through Computational Fluid Dynamic (CFD) cal-
culations done by Vinit et al. (Ref. 15) for full scale aircraft.
It was observed that during flight In Ground Effect (IGE) the
high download converts to a strong upload (9% of thrust) due
to strong upload on fuselage and rear wing allowing rolling
takeoffs with significantly increased payload. A CFD anal-
ysis of the aerodynamic interactions present in QTR design
by Sheng and Narramore (Ref. 16) concluded that the flow
field experienced by the aft rotor was characterized by a swirl
generated by the front nacelle, which resulted in a velocity
deficit on the rear rotor disk and changed the rotor-loading
pattern. Further, the downwash generated by the front wing
reduced the angle of attack of the rear wing and thus reduced
the lift produced by the rear wing. The interaction losses dur- (a) Hover mode
ing hover for quad tiltrotor can be eliminated by making the
wings tilt with rotors, which results in the quad-tiltwing de-
sign of (Ref. 12). However, the transition from hover to for-
ward flight and back to hover becomes more challenging due
to the lack of wing lift during the tilting of the rotor-wing
combination as the wing is at very high angles of attack at the
beginning of the transition.
To address some of the limitations of the quad tiltrotor and
quad tiltwing a novel quadrotor convertiplane design is pro-
posed. This paper describes the design, development and
flight testing of this novel quadrotor convertiplane unmanned
air vehicle (UAV) which has not been studied in literature.
The objective is to carry out systematic design and construc-
tion of the vehicle. First, the propeller selection is carried (b) Transition from hover to forward flight
out by experimentally measuring the power loading of various
different propellers. Next, the chosen propeller is paired with
various different motors and the motor with best efficiency is
selected for the vehicle. Next, the effect of tilting of the wing
section on the hover performance is studied experimentally.
The wing design is carried out by choosing suitable stalling
speed for the vehicle. Finally, the vehicle is fabricated, sim-
ple PID based controller is developed and implemented on a
PixHawk autopilot board and hovering flight is demonstrated.

VEHICLE CONFIGURATION
The proposed novel convertiplane design shown in Fig. 1 uses
four rotors or propellers, each powered using an individual
brushless DC motor. It also has two identical wings in tan- (c) Forward flight mode
dem configuration with forward wing being a low wing and
the aft wing a high wing (see Fig. 1(a)). The propellers are Fig. 1. Schematic of the quadrotor convertiplane during
mounted at the tip of the wing and their corresponding sense hover and forward flight modes
of rotation is shown in Fig. 1(a). In helicopter or hover mode,
the propeller axis of rotation is kept vertical to enable ver- During the transition process, the wing segment outboard of
tical takeoff and landing like a conventional quadrotor. The the rotor downwash (which remains fixed at all times) starts
control of the vehicle is achieved by varying the RPM of indi- generating lift as the vehicle gains forward speed and starts
vidual propellers, identical to control of conventional battery to contribute towards supporting the weight of the vehicle,
powered quadrotors. The propeller and the wing segment in thereby aiding in the transition process. Once, the transition is
the downwash of the propeller is tilted by 90 degrees using a complete, the propeller thrust is reduced to balance the drag of
2
 Fig. 3. Experimental setup used for measurement of pro-
peller and wing interaction

This results in a disc loading of approximately 269 N/m2 (5.66

Fig. 2. Experimental setup used for thrust and power mea- lb/ft2 ) at 0.5 kg thrust. The design of various systems are car-
surement ried out through systematic experimentation and analysis in a
step by step manner described in the following sub-sections.
the vehicle and the weight is completely supported by the two
wings. This mode is called airplane or forward flight mode, Experimental Setup for Performance Measurement in
as shown in Fig. 1(c). The tilting of the wing segment be- Hover
low the propeller minimizes the hover losses due to aerody-
namic download and interference of the rotor wake and the The propeller selection is carried out by comparing the per-
wing. The rest of the wing remains straight throughout the formance of nine different off-the-shelf propellers. For this,
flight. Since, only a segment of the wing tilts with the pro- a hover test stand is constructed using Ati Nano 43 six-
peller, the actuator loads required for tilting are expected to component load cell as shown in Fig. 2. The setup is used
be smaller compared to a full tiltwing vehicle. The choice of to measure the thrust, rotor aerodynamic torque, rotor RPM
low and high tandem wing configuration reduces the induced and motor electrical power to identify the propeller with best
losses encountered by the wing in forward flight. The fore power-loading for the desired range of thrust. The load cell
wing is placed below the fuselage and the aft wing is placed acquires data at 1 kHz allowing for the measurement of the
above the fuselage. The aircraft also has conventional fixed time resolved vibrations due to minute mass imbalance in the
wing control surfaces (not shown in the schematic) to enable propeller. The first harmonic obtained from the FFT analysis
conventional fixed wing type control during forward flight in of the acquired data gives an accurate estimate of RPM of the
airplane mode. The control of the vehicle during the transition motor. This was verified from the set RPM in the motor speed
process may be achieved by using a combination of the RPM controller and also using an optical tachometer. During hover
regulation of motors as well as fixed wing controls (once they test as well as the wind tunnel test the data is acquired for 5
start becoming effective). seconds for each test point and averaged to get the net thrust
and torque values. The propellers tested include HK 5040,
HK 6050, GF 5045 (two bladed), GF 5045 (three bladed), GF
APPROACH 5550, GF 6030, GF 6040, Lumenier 5035 and KK 6040. Af-
The main objective of this paper is to carry out systematic de- ter identifying the propeller with best power loading for 0.4
sign, fabrication and flight testing of an innovative new small kg to 0.6 kg thrust. The chosen propeller is paired with eight
sized quadrotor convertiplane with all up weight of approxi- different motors namely: Tornado (2300 KV), Emax (2300
mately 1.5-2.0 kg. The design process requires careful selec- KV), RCX (2350 KV), Tiger F40 (2500 KV), Cobra (2300
tion of propeller and motor, wing sizing and fabrication of the KV), Tiger F40 (2300 KV), Tiger F60 (2450 KV) and Lume-
vehicle. First, a baseline motor and propeller combination is nier (2350 KV) to find the motor with highest efficiency (ratio
chosen based on desired thrust of approximately 0.5 kg dur- of aerodynamic power to electrical power).
ing hover mode. Based on past experience, propeller of 15.25 The effect of rotor-wing interaction and the performance en-
cm (6 inch) diameter is chosen as the baseline propeller size. hancement in hover due to tilting of the wing segment below
3
the propeller is studied experimentally by setting up a modi- section of size 1524 mm × 914.4 mm × 609 mm with a max-
fied hover stand in which the segment of wing in the down- imum operating speed of up to 35 m/s. The setup is shown
wash of the propeller can be changed. The setup allows for in Fig. 4. Turbulence level is observed to be low (0.5% for
changing the vertical separation between the propeller and the present section). The propeller diameter is only 15.24cm
wing element as well and is shown in Fig 3. To ensure that (6 inch) thereby eliminating the possibility of any wall effects.
there is no change in propeller performance due to ground ef- The propellers have been mounted on the load cell from which
fect, it is rotated in the appropriate direction to generate thrust the thrust and torque is acquired. The tunnel speed is varied
in the downward direction with the rotor downwash pointing from 2 m/s to 25 m/s with propeller RPM maintained at three
upwards. The wing segment with same chord as that of the different angular speeds of 5000, 8000 and 13000 RPM using
actual UAV is fabricated and put in the wake of the propeller. a RPM regulated motor speed controller. The propellers are
The tip of the wing is positioned at 1.75 cm from the axis tested in tractor configuration so that the inflow is undisturbed.
of rotation of the propeller to replicate the positioning of the
motor and propeller in actual vehicle. The experiment is con- Control Strategy
ducted by varying the segment of wing oriented parallel to the
downwash. The span of the wing segment in the downwash The convertiplane UAV has three fundamental flight modes:
being tilted is changed from 50% of radius of the propeller (R)
to 125% of the propeller radius in steps of 25% of R. The re- 1. the quadrotor like hover and forward flight modes: in this
sults are compared with no tilt (pure tiltrotor) case to identify the vehicle is controlled purely through RPM regulation
the hover performance recovery by tilting of wing segment. of individual motors in different combination to tracking
The results from this study are used to decide upon the span roll, pitch and yaw motion of the quadrotor.
of the wing segment to be tilted with the propeller for best
hover efficiency. 2. transition mode: this involves change from quadrotor
type control to aircraft type control and vice verse.
3. aircraft mode: the control of the vehicle is completely
done through use of fixed wing control surfaces such as
aileron, elevator and rudder. It may be possible to aug-
ment the control action in this mode by further regulating
the motor RPMs, but it is not attempted at this point to
keep the control logic simple.

The on-board autopilot on the UAV needs to maintain stable

flight and successful trajectory tracking under all these flight
modes. Due to the unconventional and complex flight modes
of the tandem wing quad-tiltrotor, controller must be designed
for quadrotor, transition and aircraft modes separately. In the
hover mode all the four propellers and its adjacent wing sec-
tions are vertical and the control of the vehicle is identical to
that of a conventional RPM controlled quadrotor. The quadro-
tor type hover mode is simple and the system behaves as an
Fig. 4. Experimental setup used for measurement of per- under-actuated system. Therefore, the angular and transla-
formance of propellers in forward flight tional dynamics are handled by using separate PID control
loops. The inner loop works as an attitude controller which
takes attitudes and angular rates data from Inertial Measure-
ment Unit (IMU) to control the Euler angles of the vehicle by
Experimental Setup for Performance Measurement in
regulating the RPM of the motors. The outer loop (also PID)
Forward Flight
then generates desired attitude commands to control roll and
Being a convertiplane VTOL vehicle the current design is ex- pitch angle of the vehicle to track desired trajectory. During
pected to spend significant amount of its flight time in airplane this, the UAV needs to produce required translational acceler-
mode. Therefore, it is essential to study the forward flight ations in order to track the desired trajectory.
performance of some of the best propellers from the hover The propellers are tilted to initiate the transition to forward
tests. Apart from confirming the propeller selection done dur- flight / aircraft mode. The objective of the controller would
ing hover tests, the wind tunnel tests also enable estimation be to accelerate the aircraft above the stalling speed to make
of power consumption in forward flight, which can be used the fixed wing controls functional by the time the propeller
for estimating the endurance of the vehicle in airplane mode. shaft rotates by 90 degrees. During the transition phase, the
During this test, the thrust, RPM and aerodynamic torque of controller must regulate the RPM of all the motors to ensure
the propellers are measured at various propeller advance ra- that the component of weight supported by the propeller thrust
tios. The test is performed in a low speed tunnel with a test and the contribution from wing lift balances the entire weight
4
of the aircraft. In the aircraft mode, the aircraft flight con-
trol can be achieved by using conventional control surfaces Table 1. Characteristics of the propellers tested
(elevators, ailerons and rudder) and with careful placement of Name (No. of Blades) Diameter Pitch
aircraft center of mass, the aircraft can be made statically sta- HK 5040 (2) 12.7 cm (5 in) 10.16 cm (4 in)
ble, allowing for manual remote controlled flight as well. The HK 6050 (2) 15.25 cm (6 in) 12.7 cm (5 in)
current paper only focuses on the quadrotor mode for demon- GF 5045 (2) 12.7 cm (5 in) 11.43 cm (4.5 in)
stration of stable hovering flight. The controller design for GF 5045 (3) 12.7 cm (5 in) 11.42 cm (4.5 in)
transition flight and aircraft mode would be undertaken in fu- GF 5550 (2) 12.7 cm (5 in) 12.7 cm (5 in)
ture work. GF 6030 (2) 15.25 cm (6 in) 7.62 cm (3 in)
GF 6040 (2) 15.25 cm (6 in) 10.16 cm (4 in)
DESIGN AND INTEGRATION Lumenier 5035 (2) 12.7 cm (5 in) 8.89 cm (3.5 in)
KK 6040 (2) 15.25 cm (6 in) 10.16 cm (4 in)
The key design parameters of the UAV are selected through a
series of experiments and the result of these experiments are
systematically discussed in this section. Fig. 6(b). HK 6050 propeller consumes the maximum power
among all propellers, but also generated highest thrust.

To select the propeller with best performance, the variation of

power loading (thrust/aerodynamic power) is plotted against
thrust and is shown in Fig. 6(c). It is observed that the HK
6050 and GF 6030 have best power loading for thrust range
of 400 to 600 grams. But since, HK 6050 is able to generate a
maximum thrust of approximately 900 grams, it is chosen as
the propeller for the current design.

Motor Selection

The small brushless DC motors are known to have wide vari-

ation in efficiency depending upon the loading conditions.
To study the efficiency of the motors the HK 6050 propeller
selected above is paired with eight different motors and the
thrust, electrical power and aerodynamic power for the com-
bination is measured using the hover stand. To characterize
the electrical efficiency of the motor-propeller combination,
the electrical power loading (thrust/electrical power) is plot-
Fig. 5. Candidate propellers tested for quadrotor conver-
ted against thrust and is shown in Fig. 7(a). Lumenier 2350
tiplane prototype
KV motor is observed to have highest power loading in the de-
sired thrust range. Figure 7(b) compares the electrical power
consumed by different motors to generate the given amount
Propeller Selection of thrust. This information is useful in deciding the battery
capacity required to achieve desired level of hover endurance.
The propeller selection is carried out by comparing the perfor- The electrical efficiency defined as the ratio of aerodynamic
mance of nine different propellers shown in Fig. 5. The data power to electrical power, for all the motors tested is shown
for these propellers is given in Table 1. During the experiment in Fig. 7(c). Lumenier 2350 KV motor is observed to have
all the nine propellers were tested using the same motor and consistent efficiency of approximately 70% during the entire
the thrust, torque, RPM and electrical power was measured. operating range and is selected for the current design.
The experimental variation of thrust with rotor RPM for all
the propellers is shown in Fig. 6(a). The RPM is varied from
3000 to 23000 approximately. It is observed that the desired
Effect of Tilting the Wing Segment
thrust of 400-500 grams required from each rotor for hovering
flight can be generated by most of the propellers tested. How-
ever, Hobby King HK 6050 propeller gives the highest value The key feature of the current design is the tilting of the wing
of thrust at a given RPM among the nine propellers tested. The segment in the downwash of the propeller with the propeller
aerodynamic power variation is obtained by multiplying the axis. The motor and propeller selected above are now tested in
measured torque with angular velocity. The variation of aero- a modified hover stand which includes the wing in the down-
dynamic power with RPM for all the propellers is shown in wash of the propeller. The effect of propeller-wing interaction
5
 HK 5040
1000 GF 6040 16
Lumenier 5035
KK 6040 Tornado 2300 kv
14

Electrical Power Loading (g/W)

800 GF 5045 - 2 Bladed Emax 2300 kv
GF 5045 - 3 Bladed RCX 2350 kv
GF 5550 12 F40 2500 kv
Thrust (grams)

GF 6030 Cobra 2300 kv

600 HK 6050 F40 2300 kv
10 F60 2450 kv
Lumenier 2350 kv

400 8

6
200

4
0
3,000 8,000 13,000 18,000 23,000 2
RPM 0 200 400 600 800 1000
(a) Thrust vs. RPM Thrust (grams)
(a) Electrical Power Loading vs. RPM
250
HK 5040
300
GF 6040
Lumenier 5035 Tornado 200 kv
200
Aerodynamic Power (Watts)

KK 6040 Emax 2300 kv

GF 5045 - 2 Bladed 250 RCX 2350 kv
GF 5045 - 3 Bladed F40 2500 kv
GF 5550 Cobra 2300 kv
150
GF 6030 200 F40 2300 kv
Power (watts)

HK 6050
F60 2450 kv
Lumenier 2350 kv
100 150

50 100

50
0
3,000 8,000 13,000 18,000 23,000
RPM 0
0 200 400 600 800
(b) Aerodynamic Power vs. RPM
Thrust (grams)
(b) Electrical Power vs. Thrust
25

1
HK 5040 Tornado 2300 kv
20 GF 6040 Emax 2300 kv
Lumenier 5035
Power Loading (g/W)

KK 6040 0.9 RCX 2350 kv

GF 5045 - 2 bladed F40 2500 kv
15 GF 5045 - 3 blade Cobra 2300 kv
GF 5550 F40 2300 kv
0.8
Efficiency

GF 6030 F60 2450 kv

HK 6050
10 Lumenier 2350 kv
0.7

5
0.6

0
0 200 400 600 800 1000 0.5
Thrust (grams) 0 200 400 600 800 1000
Thrust (grams)
(c) Aerodynamic Power Loading vs. Thrust
(c) Motor Efficiency vs. Thrust
Fig. 6. Measured thrust, aerodynamic power and power
loading for different propellers Fig. 7. Measured electrical power and motor efficiency for
different motors

6
 800 0.2
Isolated rotor
Tilt wing span = 0.5R
GF 6030
Tilt wing span= 0.75R
Tilt wing span= R GF 6040
600 0.15 KK 6040
Tilt wing span= 1.25R
Thrust (grams)

Full tilt wing HK 6050

No tilt

400

T
0.1

C
200 0.05

0
3,000 7,000 11,000 15,000 0
0.05 0.1 0.15 0.2 0.25 0.3
RPM J (advance ratio)
(a) Thrust vs. RPM (a) RPM = 5000

300 0.2
Isolated Rotor
Tilt wing span = 0.5*R GF 6030
250 GF 6040
Tilt wing span = 0.75*R
Electrical Power (Watts)

Tilt wing span = R 0.15 KK 6040

200 Tilt wing span = 1.25*R HK 6050
Full tilt wing
No tilt
150
CT

0.1

100

0.05
50

0
3,000 7,000 11,000 15,000 0
0.05 0.1 0.15 0.2 0.25 0.3
RPM J (advance ratio)
(b) Electrical Power vs. RPM (b) RPM = 8000

18 0.2

Isolated rotor
16
Electrical Power Loading (g/W)

Tilt wing span = 0.5R GF 6030

Tilt wing span = 0.75R GF 6040
14 Tilt wing span = R
0.15
KK 6040
Tilt wing span = 1.25R
12 Full tilt wing HK 6050
T

No tilt
C

10 0.1

6 0.05

2 0
0 200 400 600 800 0.05 0.1 0.15 0.2 0.25 0.3
Thrust (grams) J (advance ratio)
(c) Electrical Power Loading vs. Thrust (c) RPM = 13000

Fig. 8. Effect of tilting of wing under rotor downwash Fig. 9. Thrust coefficient CT vs. Advance Ratio of the pro-
on measured thrust, electrical power and electrical power peller at different operating RPM
loading

7
 resulting in thrust loss is studied by changing the span of the
1 wing segment present in the propeller downwash. The airfoil
GF 6030 (Experiment)
GF 6030 (Best fit) and the wing chord used in the experiment is same as that used
GF 6040 (Experiment) on actual vehicle. The performance of the rotor in presence of
0.8 GF 6040 (Best fit)
KK 6040 (Experiment) the wing in the downwash is compared with that of the iso-
KK 6040 (Best fit) lated rotor to quantify the losses. The following six different
HK 6050 (Experiment)
Efficiency ( η)

0.6 HK 6050 (Best fit) cases are examined and compared with the isolated rotor case:
1) span of the wing segment parallel to the downwash is equal
to 50% of propeller radius (R), 2) span of the wing segment
0.4 parallel to the downwash is equal to 75% of propeller radius,
3) span of the wing segment parallel to the downwash is equal
0.2
to 100% of propeller radius, 4) span of the wing segment par-
allel to the downwash is equal to 125% of propeller radius,
5) entire wing is aligned parallel to downwash (referred to
0 as full tilt wing) and 6) entire wing is aligned normal to the
0.05 0.1 0.15 0.2 0.25
downwash (similar to a pure tiltrotor and is referred to as no
J (advance ratio)
tilt case).
(a) RPM = 5000
Figure 8(a) compares the variation of thrust with RPM for all
1
the cases listed above. It is observed that if the wing segment
GF 6030 (Experiment) parallel to the downwash has span of at least 75% of the ra-
GF 6030 (Best fit)
GF 6040 (Experiment) dius of the propeller, the loss in thrust is less significant (of the
0.8 GF6040 (Best fit) order of 2–3%). This is due to the elimination of the down-
KK 6040 (Experiment)
KK 6040 (Best fit) load penalty of the rotor wake on the wing. For the current
HK 6050 (Experiment) configuration, if span of wing segment parallel to downwash
Efficiency ( η)

0.6 HK 6050 (Best fit)

is reduced to 50% of propeller radius the loss in thrust is 18%
to 22%. In the case of the vehicle being a pure tiltrotor and
0.4 no portion of the wing is aligned parallel to the propeller axis,
at the current scale, the loss in thrust is approximately 40%
for the same RPM. The reduction in power loading due to the
0.2 presence of the wing is 50%, when compared to the isolated
rotor. The maximum thrust magnitude achievable with the
motor used in the test dropped from 787 grams to 420 grams.
0
0.05 0.1 0.15 0.2 0.25 0.3 The variation of electrical power consumed by the motor with
J (advance ratio) RPM for all the cases is shown in Fig. 8(b). The difference
(b) RPM = 8000 in power consumed by the motor for all the cases is less sig-
nificant. The significant drop in electrical power loading of
1 the propeller-motor combination in the tiltrotor configuration
GF 6030 (Experiment)
GF 6030 (Best fit)
can be observed in Fig. 8(c). Based on these results, it is de-
GF 6040 (Experiment) cided to keep the wing segment of span equal to 100% of the
GF 6040 (Best fit)
0.8
KK 6040 (Experiment)
propeller radius aligned with the propeller axis to ensure that
KK 6040 (Best fit) there is practically no loss in performance compared to an iso-
HK 6050 (Experiment)
lated propeller during hovering flight.
Efficiency ( η)

HK 6050 (Best fit)

0.6

Forward Flight Performance

0.4
The performance of the best four propellers (GF 6030, GF
6040, KK 6040 and HK 6050) from the hover test is studied in
0.2
forward flight by testing it in the wind tunnel. The measured
non-dimensional thrust (CT = T /(ρ n2 D4 )) variation for these
0 propellers is plotted against advance ratio (J = V /nD), where
0.05 0.1 0.15 0.2 0.25 0.3 T is measured propeller thrust, ρ is density of air, n is num-
J (advance ratio) ber of revolutions per second, D is propeller diameter and V
(c) RPM = 13000 is free stream velocity. The non-dimensional thrust variation
with advance ratio for the four propeller at 5000, 8000 and
Fig. 10. Efficiency vs. Advance Ratio of the propeller at 13000 RPM is shown in Figs. 9(a), 9(b) and 9(c) respectively.
different operating RPM HK 6050 propeller is observed to have the consistently high
thrust output at all advance ratios and at all three RPMs tested.
8
The efficiency of the propeller (η ) is defined as: J = TV /P, carried out for the current geometry by using XFLR5 software
where P is the measured power output. The variation of the using 3D panel method at a Reynolds number of 200,000.
efficiency of the propeller with the advance ratio for 5000, XFLR5 is an open source analysis tool for airfoils, wings and
8000 and 13000 RPM is shown in Figs. 10(a), 10(b) and 10(c) planes operating at low Reynolds Numbers (Ref. 18). The
respectively. It should be noted that, the measured CT and η lift, drag and pitching moment coefficients are obtained for
observed for these small propellers operating in low Reynolds the angle of attack range of -2◦ to 10◦ . The comparison
number regime is consistent with those reported in literature of lift coefficient, drag coefficient, pitching moment coeffi-
for such small propellers (Ref. 17). Once again, HK 6050 of- cient and lift to drag ratio for the four airfoils is shown in
fers highest maximum efficiency which ranges between 0.52– Figs. 12(a), 12(b), 12(c) and 12(d) respectively. The key in-
0.57 for all three RPMs considered. This further reinforces ferences drawn from these graphs are summarized in Table 2.
the choice of the propeller made from the hover performance From Fig. 12(a), it is observed that S1223 airfoil offers high-
data. est Cl at any angle of attack among the four airfoils analyzed.
But, it also has very low stall angle, high drag (see Fig. 12(b))
and pitching moment (see Fig. 12(c)) when compared to other
airfoils. NACA0010 airfoil has very low pitching moment co-
efficient which may be useful in enhancing static stability but
it offers low value of lift coefficient and stall angle. Having
eliminated S1223 and NACA0010, AG36 and SD7062 airfoils
are studied closely. Both these airfoils have similar lift coef-
ficient variation, but AG36 offers higher value of lift to drag
ratio for up to 7 degree angle of attack as shown in Fig. 12(d).
Further, SD7062 has higher negative pitching moment com-
pared to AG36 (see Fig. 12(c)), which may result in greater
control actuation for stabilizing the aircraft. Based on these
observations, AG36 is chosen as the airfoil for both the wings.

Propeller Tilting Mechanism

The propeller, motor and a portion of the wing has to be tilted

by 90 degrees when the UAV changes from the hover mode to
the airplane mode. The tilting is achieved by a servo motor at-
Fig. 11. Airfoils considered for wing design tached to the non-tilting portion of the wing. A simple mech-
anism is designed for this purpose and is shown in Fig. 13.
The mechanism consists of the motor and propeller supported
by a circular aluminum shaft which is allowed to rotate in-
Wing Sizing and Airfoil Selection side a hollow carbon fiber tube of square cross section. The
guide is fixed to the non-tilting (main) part of the wing. The
For determining the wing area, a total gross weight of 1.6 kg use of guide eliminates the bending moment experienced by
is assumed for the vehicle. Based on the past experience of the servo head and enables desired tilting of the propeller and
making small fixed wing aeromodels a wing loading of ap- wing. The fixed end, as shown in the figure, is attached to the
proximately 5.5 kg/m2 is chosen. This results in required total tilting part of the wing as well as the shaft. It enables the ro-
wing area of 0.28 m2 . To keep the footprint of vehicle low and tation of both the propeller and the wing portion by the same
get sufficient rigidity from wing made of pink foam, an aspect servo.
ratio of 3.5 is chosen for each wing. Based on these choices,
the span of each wing is set to 0.7 m and the chord length of PID Controller for Hovering Flight
0.2 m.
The current tandem wing design has no “horizontal tail” to The quadrotor convertiplane UAV flies like a conventional
ensure the moment equilibrium on its own. Therefore, the air- quadrotor in hovering flight. The control of the vehicle is
foil selected should have low pitching moment co-efficient to achieved by using a Proportional Integral Derivative (PID)
ensure easy trimming by deflection of control surfaces. This controller which independently regulates the RPM of the pro-
requirement could be possibly satisfied by using a airfoil with pellers. The control input used for stabilizing and control-
low camber. The performance of four airfoils, with differ- ling the vehicle is angular speed of each propeller (Ω =
ent camber, commonly used in small fixed wing aeromodels [ω1 ω2 ω3 ω4 ]) which needs to be determined to match
are compared (see Fig. 11): 1) NACA 0010 (symmetric air- the desired four outputs: position (P = [x y z]) and head-
foil with zero camber), 2) AG36 (low camber), 3) SD7062 ing (ψ ). The controller is designed in two loops: outer loop,
(medium camber) and 4) S1223 (high camber). The estima- inner loop. The outer loop is responsible for trajectory track-
tion of the lift, drag and pitching moment for the airfoils is ing whereas the inner loop provides attitude stabilization. For
9
 1.8 0.2
1.6
1.4
0.15 Ag36
1.2
Sd7062
1 S1223
0.8 NACA0010

d
Cl

0.1

C
0.6
0.4
Ag36
0.2 Sd7062 0.05
0 S1223
-0.2 NACA0010
-0.4 0
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
Angle of attack (deg) Angle of attack (deg)
(a) Lift coefficient (b) Drag coefficient

0.05 60

0
40
-0.05

20
-0.1
d
C /C
Cm

Ag36
Ag36
l

-0.15 Sd7062
0 Sd7062
S1223
S1223
NACA0010
-0.2 NACA0010
-20
-0.25

-0.3 -40
-2 0 2 4 6 8 10 -2 0 2 4 6 8 10
Angle of attack (deg) Angle of attack (deg)
(c) Pitching moment coefficient (d) Lift/Drag ratio

Fig. 12. Comparison of the performance of various airfoils using XFLR5

Table 2. Summary of airfoil properties (green: desired; red: not desired)

Airfoil Cl Stall Angle Cd Cm Cl /Cd
AG36 Satisfactory Satisfactory Low Very low High
SD7062 Satisfactory Satisfactory Low High Low
S1223 High Very low Very high Very high High
NACA0010 Very low Satisfactory High Very low Low

10
 50

40

30 Current state
Desired state

φ (deg)
20

10

Fig. 13. Schematic of the propeller tilting mechanism

0

-10
the outer loop, the inertial acceleration is computed as 0 0.5 1 1.5 2
Time (s)
Z t (a) Roll attitude
P̈ = −K po e p − Kio e p (t)dt − Kdo dote p (1)
o 50

where K po , Kio , Kdo are positive definite gain matrices and 40

e p = P − Pd , where Pd is the desired output of a quadrotor
in the inertial frame. The required thrust and desired roll and 30
pitch angles are computed from the desired states of position
θ (deg)

and heading as
20
p
Td = M ẍ2 + ÿ2 + (g − z̈)2 (2) 10
φd = sin−1 (ux sin ψd − uy cos ψd ) (3)
ux cos ψd +uy sin ψd 0
θd = sin−1 cos φd (4)
-10
0 0.5 1 1.5 2
where ux = mẍ/Td and uy = mÿ/Td . For attitude stabilization,
Time (s)
the vehicle moments, M = [l m n]T , are computed as
(b) Pitch attitude
 Z t 
= I −K pi eE − Kii eE (t)dt − Kdi doteE (5) 30
o
25

20
where K pi , Kii , Kdi are positive definite gain matrices and eE = Current state
15
[φ − φd θ − θd ψ − ψd ]T , where subscript ‘d’ is used to de- Desired state
ψ (deg)

note the desired attitude. The inertia matrix is represented by 10

I. The complete description of the PID controller is available
5
in Bouabdallah et al. (Ref. 6).
0
The PID controller described above for stabilization of the
convertiplane in hover mode has been implemented and tested -5
for a simulated quadrotor model using Matlab. The simu- -10
lated vehicle under hover condition is arbitrarily disturbed 0 0.5 1 1.5 2
in roll, pitch and yaw attitude by 50◦ , 30◦ and 20◦ respec- Time (s)
tively. The vehicle regains its undisturbed hover attitude (c) Yaw attitude
quickly within half a second for roll, pitch and yaw as shown
in Figs. 14(a), 14(b) and 14(c) respectively. The stock PID Fig. 14. Attitude stabilization using PID controller for the
controller available on a PixHawk autopilot board is then used simulated hover mode of the convertiplane UAV
to stabilize the vehicle in hovering flight. The gains required
for stable flight are tuned by trial and error through experi-
mentation.
11
Avionics and Telemetry 9800 RPM. The corresponding power required from each mo-
tor is 45 W as shown in Fig. 15. This is 50% less than the
The quadrotor convertiplane is inherently unstable and there- power required for hovering flight. Therefore, the prototype
fore an on-board autopilot is required to stabilize the vehi- is estimated to have an endurance of approximately 10 min-
cle continuously in various modes of flight. The open source utes with the current battery.
PixHawk autopilot board with on-board ARM Cortex M4
processor (STM32f407) based micro-controller with a clock 250
speed of 168 MHz is used for stabilization and control. It
has nine degrees-of-freedom IMU (3 axis digital MEMS gyro
200
with range of ±2000 deg/s, 3 axis digital MEMS accelerom-
eter with measuring range of ±16 g and 3 axis digital MEMS

Power (W)
Magnetometer with range of ±12 Gauss), pressure sensor and 150
GPS. The autopilot loop on the PixHawk runs at 200 Hz.
The autopilot software uses open source Mavlink communi- 100
cation protocol for all RTOS inter-process communications as
well as ground link telemetry communications related to ve- 50
hicle health, flight modes, alerts, current position, autopilot
arm / disarm status, way-point transfer / switching and sensor
0
calibration. The Mavlink communication protocol also offers -100 0 100 200 300 400 500
cyclic redundancy check (CRC) checksum to reduce data cor- Thrust (gram)
ruption during communication between the ground station and
the vehicle. The Linux version of open-source QGroundCon- Fig. 15. Curve-fit to measured propeller power vs thrust at
trol ground station software is used for real-time monitoring 15 m/s forward speed
and logging of telemetry data. The telemetry module pair op-
erating at 433MHz is used for all vehicle to ground station
communication.
Table 3. Weight breakdown for the second prototype
Part Weight (in gram) % of Total
Estimation of Endurance
Weight
The endurance of the vehicle is estimated in hovering and for- Motors 129.2 8.23
ward flight using experimental measurements and estimates Propellers 18.8 1.20
based on XFLR5 software. The total thrust to be generated by Structure 825.1 52.55
the four propellers for hover is 1.57 kg. Therefore, each pro- Avionics 231 14.71
peller needs to generate approximately 0.4 kg of thrust. Based ESCs 60.4 3.85
on the hover tests performed on the motor-propeller set cho- Battery 208.7 13.29
sen for the prototype, each motor consumes 89.4 W (based Servos 96.8 6.17
on power variation for tilt wing span = R case of Fig. 8(b)) Total Weight 1570
for generating the desired thrust. Therefore, the total power
consumed by all four motors for hovering flight is 357.6 W.
A 3-cell Lithium-Polymer battery pack of 2650 mAh is used Vehicle Integration and Flight Testing
for the flight test, this results in a hover endurance of approx-
Based on the experiments and analysis explained earlier in the
imately 5 minutes. This is clearly low, but can be improved
text, various components of the vehicle are fabricated and fi-
by selecting a higher capacity battery. Higher capacity battery
nally assembled to complete the prototype. The layout is kept
is also going to be heavier, and therefore selection should be
modular to enable easy assembly and repair of components.
done carefully to ensure that the extra energy available from
For this, the fuselage is made like a frame with rectangular
the higher capacity battery doesn’t get canceled by increased
cross section as shown in Fig. 16. The wings of the UAV are
weight and power requirement. It should be noted that during
fabricated using pink foam and reinforced using aluminum T-
hovering flight each motor-propeller set can produce a maxi-
spar. In the first prototype, aluminum square channels were
mum thrust of 0.76 kg, allowing for additional lifting capacity
used to construct the frame. This resulted in an all up weight
of approximately 1.4 kg beyond the current AUW.
(AUW) of 2.2 kg for the prototype (shown in Fig. 16(a)) which
The total thrust required for the vehicle to cruise at 15 m/s was too high. A second prototype (shown in Fig. 16(b)) was
in forward flight mode is estimated using XFLR5, which pre- made by using carbon fiber rods and the AUW was brought
dicts the drag-coefficient for the vehicle to be 0.13. There- down to 1.6 kg. The weight breakdown for the various com-
fore, the total thrust required is 0.51 kg which puts the thrust ponents of the quadrotor is shown in Table 3. In the current
requirement from each propeller at 127 grams approximately. prototype, 52% of the weight is that of the structure, which
The thrust of 127 grams at 15 m/s is obtained at approximately would be optimized in the future iterations of the vehicle.
12
 back to hover would be undertaken in future work. The aero-
dynamic performance of low and high tandem wing design
would be also studied experimentally.

(a) Airframe of Prototype - I

Fig. 17. Hover flight of the quadrotor convertiplane proto-

type

SUMMARY AND CONCLUSIONS

The systematic design and construction of a novel quadrotor
convertiplane UAV has been carried out in this paper. The de-
sign studied, offers novel features, in which the segment of
the wing under the propeller downwash is tilted with the pro-
peller to improve the hovering efficiency of the UAV. In addi-
tion, tandem low and high wing design was used to mitigate
the negative influence of fore wing and propeller downwash
(b) Completed Prototype - II on the aft wing and the propellers. The motor and propeller
suitable for the design were chosen based on the performance
Fig. 16. Quadrotor convertiplane prototypes measurements carried out using a hover test stand. The effect
of tilting the wing segment in the rotor downwash was stud-
A PID based controller for attitude stabilization, described ied by creating a modified hover stand with wing included in
earlier, is implemented on PixHawk autopilot board and gains the setup. The motor propeller selected in the process were
for the controller were tuned through trial and error. Finally, used to fabricate the quadrotor convertiplane vehicle. A PID
remote controlled hovering flight of the prototype is achieved controller based autopilot was developed and implemented
as shown in Fig. 17. on PixHawk autopilot board to stabilize the vehicle in hover
mode and enable flight testing. The autopilot would be ex-
tended in future to enable transition and forward flight of the
Future Work UAV. The key conclusions from this study can be summarized
as follows:
A proof-of-concept prototype has been built and hovering
flight tests have been performed using a PID controller im- 1. For the current configuration and size, the proposed
plemented on the PixHawk autopilot board. The improved novel quadrotor convertiplane design, with the wing seg-
efficiency of the current design compared to a conventional ment in the propeller downwash allowed to tilt with it,
quad-tiltrotor design has been established. In the next phase, offers significantly higher hovering efficiency in compar-
the prototype would be fitted with wheels and flown like a ison to a conventional quad tiltrotor with no wing tilting.
fixed wing airplane. The vehicle would be stabilized using It is observed that the propeller looses 40% of its thrust
in-house fixed wing autopilot software implemented for Pix- due to the download penalty on the wing and the power
Hawk, the gains would be first tuned using a simple hardware- loading decreases by 50% of that of an isolated rotor.
in-the-loop-simulator (HILS) to ensure stable flight. After,
tuning of the gains in simulation, it would be tested on the ac- 2. If the wing segment with a span of at least 75% of the
tual vehicle for forward flight. Finally, the controller design radius of the propeller, is allowed to tilt and align with
for autonomous transition from hover to forward flight and the propeller downwash, the loss in thrust is less than
13
 3%. If the span of the tilting portion of the wing is in- 9 Hrishikeshavan, V., Bogdanowicz, C. and Chopra, I., “De-
creased to 100% of the radius of the propeller, then the sign, Performance and Testing of a Quad Rotor Biplane Micro
propeller operates with nearly same power loading as the Air Vehicle for Multirole Missions,” International Journal of
isolated rotor. This is due to the elimination of the down- Micro Air Vehicles, Vol. 6, No. 3, September 2014, pp. 155-
load penalty of the rotor wake on the wing. 174.
3. The convertiplane is estimated to consume 50% less 10
Bogdanowicz, C., Hrishikeshavan, V. and Chopra, I., “De-
power during airplane mode in comparison to the velopment of a Quad-Rotor Biplane MAV with Enhanced Roll
quadrotor mode thereby establishing the advantage of Control Authority in Fixed Wing Mode,” 71st Annual Forum
the proposed convertiplane vehicle over regular quadro- of the American Helicopter Society, Virginia Beach, VA, May
tor UAVs. Due to this the UAV is estimated to have two 2015.
fold increase in endurance during forward flight mode
11 Flores, G.R., Escareno, J., Lozano, R., and Salazaret S.,
compared to the hover mode.
“Four Tilting Rotor Convertible MAV: Modeling, Control and
Author contact: Abhishek, [email protected]; M. Rama Kr- Real-time Hover Flight Control,” Journal of Intelligent and
ishna, [email protected]; Sourav Sinha, [email protected]; Robotic System, Vol. 65, No. 1, January 2012, pp. 457–471,
Joydeep Bhowmik, [email protected]; Debopam Das, doi:10.1007/s10846-011-9589-x
[email protected] 12 Cetinsoy, E., Dikyar, S., Hancer, C., Oner, K.T., Sirimoglu,

E., Unel, M. and Aksit, M.F., “Design and Construction of a

REFERENCES Novel Quad Tilt-wing UAV,” Mechatronics, Vol. 22, No. 6,
1 Seshadri,P., Benedict, M., and Chopra, I., “Experimental September 2012, pp. 723–745.
Investigation of an Insect-based Flapping Wing Hovering Mi- 13
Anonymous, Bell Boeing Quad TiltRotor, Wikipedia,
cro Air Vehicle”, Presented at the Aeromechanics Conference
URL: https://goo.gl/zdQngA retreived on 29 September 2016.
of the American Helicopter Society, Jan 20–22, San Fran-
cisco, 2010. 14 Radhakrishnan, A.,. “An Experimental Investigation of
2 Bohorquez, F., Samuel, P., Sirohi, J., Pines, D., Rudd, L. Ground Effect on a Quad Tilt Rotor in Hover and Low Speed
Forward Flight,” PhD Thesis, University of Maryland, 2006.
and Perel, R., “Design, Analysis and Hover Performance of
a Rotary Wing Micro Air Vehicle”, Journal of the American 15
Gupta, V., “Quad Tilt Rotor Simulations in Helicopter
Helicopter Society, Vol. 48, No. 2, April 2003, pp. 80–90. Mode using Computational Fluid Dynamics”, PhD Thesis,
3 Schafroth, University of Maryland, 2005.
D., Bouabdallah, S., Bermes, C., and Siegwart,
R., “From the Test Benches to the First Prototype of the muFly 16 Sheng, C. and Narramore, J. C., “Computational Simula-
Micro Helicopter”, Journal of Intelligent & Robotic Systems, tion and Analysis of Bell Boeing Quad Tiltrotor Aero Inter-
Vol. 54, No. 1, 2009, pp. 245–260. action,” Journal of the American Helicopter Society, Vol. 54,
4
Ulrich, E., Pines, D. and Gerardi, S., “Autonomous Flight No.4 , 2009, pp. 42002-1–42002-15.
of a Samara MAV”, Proceedings of the American Helicopter 17 Brandt,J. and Selig, M., “Propeller Performance Data
Society 65th Annual Forum, Grapevine Texas, 2009. at Low Reynolds Numbers,” Proceedings of 49th AIAA
5 Moble,
B., “Fundamental Understanding of the Cycloidal- Aerospace Sciences Meeting including the New Horizons Fo-
Rotor Concept for Micro Air Vehicle Applications”, Ph.D rum and Aerospace Exposition, Orlando, FL, January 2011,
Thesis, Department of Aerospace Engineering, University of pp. 1255–1272.
Maryland, College Park, December 2010. 18 XFLR5 Software, URL: http://www.xflr5.com, as retrieved

6 Bouabdallah, S., Noth, A., and Siegwart, R., “PID vs LQ on 10 March, 2017.
Control Techniques Applied to an Indoor Micro Quadrotor,”
Proceedings of 2004 IEEE/RSJ International Conference on
Intelligent Robots and Systems, Vol. 3, pp. 2451–2456.
7
Stone, R.H., Anderson, P., Hutchison, C., Tsai, A., Gibbens,
P., and Wong, K.C., “Flight Testing of the T-Wing Tail-Sitter
Unmanned Air Vehicle,” Journal of Aircraft, Vol. 45, No. 2,
2008, pp. 673-685.
8 Frank,
A., McGrew, J.S., Valenti, M., Levine, D., and How,
J.P., “Hover, Transition, and Level Flight Control Design for a
Single-Propeller Indoor Airplane,” AIAA Guidance, Naviga-
tion and Control Conference and Exhibit, Hilton Head, South
Carolina, AIAA Paper 2007-6318, 2007.
14

View publication stats