Development of a quad rotor tail-sitter VTOL UAV without control surfaces and experimental verification

Abstract

This paper presents development of a quad rotor tail-sitter VTOL UAV (Vertical Takeoff and Landing Unmanned Aerial Vehicle) which is composed of four rotors and a fixed wing. The conventional VTOL UAVs have a drawback in the accuracy of the attitude control in stationary hovering because they were developed based on a fixed-wing aircraft and they used the control surfaces, such as aileron, elevator, and rudder for the attitude control. To overcome such a drawback, we developed a quad rotor tail-sitter VTOL UAV. The quad rotor tail-sitter VTOL UAV realizes high accuracy in the attitude control with four rotors like a quad rotor helicopter and achieves level flight like a fixed-wing airplane. The remarkable characteristic of the developed quad rotor tail-sitter VTOL UAV is that it does not use any control surfaces even in the level flight. This paper shows the design concept of the developed UAV and experimental verification of all flight modes including hovering, transition flight and level flight.

Full text

Development of a Quad Rotor Tail-Sitter VTOL UAV
without Control Surfaces and Experimental Verification
Atsushi Oosedo1, Satoko Abiko1, Atsushi Konno2, Takuya Koizumi1, Tatuya Furui1and Masaru Uchiyama1
Abstract— This paper presents development of a quad rotor
tail-sitter VTOL UAV (Vertical Takeoff and Landing Unmanned
Aerial Vehicle) which is composed of four rotors and a fixed
wing. The conventional VTOL UAVs have a drawback in the
accuracy of the attitude control in stationary hovering because
they were developed based on a fixed-wing aircraft and they
used the control surfaces, such as aileron, elevator, and rudder
for the attitude control. To overcome such a drawback, we
developed a quad rotor tail-sitter VTOL UAV. The quad rotor
tail-sitter VTOL UAV realizes high accuracy in the attitude
control with four rotors like a quad rotor helicopter and
achieves level flight like a fixed-wing airplane. The remarkable
characteristic of the developed quad rotor tail-sitter VTOL UAV
is that it does not use any control surfaces even in the level
flight. This paper shows the design concept of the developed
UAV and experimental verification of all flight modes including
hovering, transition flight and level flight.
I. INTRODUCTION
Recently, UAVs are widely used to gather various types
of information from the sky above for a variety of purposes
in civil applications. In disasters, UAVs are expected to play
very significant roles to rescue victims, explore the disaster
areas or to deliver relief supplies to the isolated areas alone.
Especially, small VTOL UAVs are very promising in such
hazardous situations since they can fly long distance as fixed-
wing airplanes and can hover as conventional rotary-wing
aircraft by themselves.
Among the several types of VTOL aircraft (tilt-rotor,
vector-thrust etc.) [1], a tail-sitter aircraft is the simplest way
to achieve the VTOL maneuver since it does not require extra
actuators for the VTOL maneuver. The tail-sitter VTOL air-
craft can achieve both level flight and hovering by changing
its pitch angle of the fuselage by 90 ◦as shown in Fig. 1.
Up to now, several types of tail-sitter VTOL UAVs have
been developed. Stone et. al. developed the T-Wing tail-sitter
UAV with a canard wing and tandem rotors [2]. Kita et.
al. developed a simple tail-sitter VTOL UAV with a single
propeller R/C airplane [3]. The above mentioned tail-sitter
VTOL UAVs were developed based on fixed-wing aircraft.
Therefore, the attitude is controlled by control surfaces, such
as aileron. As a result, they had relatively poor performance
in stationary hovering compared to a rotary-wing aircraft,
although they achieved high stability in level flight.
1A. Oosedo, S. Abiko, T. Koizumi, T. Furui and M. Uchiyama
are with Department of Mechanical Systems and Design, Grad-
uate School of Engineering, Tohoku University, 6-6-01 Aramaki-
aza-Aoba, Aoba-ku, Sendai, 980-8579, Japan. {oosedo, abiko,
uchiyama}@space.mech.tohoku.ac.jp
2A. Konno is with Divi. of System Science and Informat-
ics, Hokkaido University, 14-9, Kita-ku, Sapporo, 060-0814, Japan.
{konno}@ssi.ist.hokudai.ac.jp
Transition
from
Takeoff
Transition
to
Landing
Landing
Level Flight
Fig. 1. Take-off and landing of a tail sitter VTOL
On the other hand, quad rotor helicopters have achieved
high stability in positioning and attitude control with a simple
mechanism. Therefore, several researchers have commonly
used the quad rotor helicopters for their research applica-
tions, such as autonomous navigation in indoor environments
and multi-vehicle flight testbed [4][5]. However, it is difficult
for quad rotor helicopters to fly long distance and long
duration because the most of the thrusts are consumed to lift
the body up, and hence horizontal component of the thrust
is small.
One solution to overcome the above drawbacks is, so
called, a quad rotor tail-sitter VTOL UAV. The quad rotor
tail-sitter VTOL UAV equips with a fixed wing on the basis
of the quad rotor helicopter. The quad rotor tail-sitter VTOL
UAV provides the ability of long distance flight while holding
high stability in positioning and attitude control.
Young et. al. developed a quad rotor tail-sitter UAV with
an R/C airplane and succeeded in hovering and forward flight
in 2002 [6]. However, they did not discuss autonomous level
flight at all and have not reported further development since
then. Sinha et. al developed a quad rotor tail-sitter VTOL
UAV named Quadshot in 2012 [7]. Quadshot performed
high dynamic maneuverability by use of a combination of
differential thrust and two control surfaces, namely elevon.
Hence, this UAV requires six actuators for flying. However,
in principle, the quad rotor tail-sitter UAV does not require
additional control surfaces since only four rotors can realize
stable attitude control.
In our previous research published in [8], we performed
flight simulation of a quad rotor tail-sitter UAV without
using any control surfaces. The simulation result apparently
showed that the UAV is able to realize level flight without
control surfaces. Moreover, it is clearly shown from the
simulation results that the quad rotor tail-sitter UAV can
fly three times longer distance compared with that of the
conventional quad rotor helicopter.
This paper presents the development of a quad-rotor tail-
sitter UAV that consists of four rotors and a fixed wing based
2013 IEEE International Conference on Robotics and Automation (ICRA)
Karlsruhe, Germany, May 6-10, 2013
978-1-4673-5643-5/13/$31.00 ©2013 IEEE 317

45
Xb
Yb
Zb
(a) Cross type of quad rotor tail-sitter UAV
(b) Asterisk type of quad rotor tail-sitter UAV
Auxiliary wing
Main wing
Xb
Yb
Zb
Fig. 2. Two types of the developed quad rotor tail-sitter UAVs
on the conventional quad rotor helicopter. The remarkable
characteristic of the developed UAV is that it does not use
any control surfaces, but it can realize all flight modes,
namely hovering, transition and level flight with simple
mechanism.
Firstly, we describe design concept and system configura-
tion of the developed UAV. Secondly, we briefly describe
flight control system implemented to the system, which
has been developed in our previous research [9]. Finally,
experimental verification of all flight modes are carried out.
In the experimental verification, firstly, we compare the per-
formance of two types of UAVs in the attitude control since
the effect of slipstream was not modeled in the simulation in
[8]. Then, the UAV which has better performance is selected
to demonstrate the transition and level flights.
II. SYSTEM CONFIGURATION
A. Design concept of the system
Fig. 2 shows overview of the developed quad rotor tail-
sitter UAVs. The quad rotor tail-sitter UAV in Fig. 2(a) is
named here cross-type of UAV, in which the main wing is
aligned under the propellers. The quad rotor tail-sitter UAV
in Fig. 2(b) is named here asterisk-type of UAV, in which the
main wing is allocated on the place rotated by 45 [◦] around
the Zbaxis with respect to the allocation of the propellers.
If the control surfaces are used to control the attitude
as the T-wing and Quadshot, the control surfaces must be
placed under the propellers. On the other hand, if this is
not the case, the main wing is placed at the area where the
propeller slipstream does not affect the control performance.
As one can see in Fig. 2, the cross-type of UAV has a
main wing under the propellers. Therefore, one can imagine
that the slipstream influences the performance of attitude
control. In the simulation analysis of the quad-rotor tail-
sitter VTOL UAV in [8], we did not include the detail effect
of the slipstream. Therefore, we quantitatively analyze the
influence of the slipstream through experiment verification
by comparing the performance of the cross-type and asterisk-
type of UAVs. The detail analysis of this is shown in Section
IV.
B. Coordinate Systems
In each UAV, the coordinates of the aircraft body (Xb,
Yb,Zb) are defined as shown in Fig. 2. The rotations about
the Xbaxis, Ybaxis and Zbaxis are defined as roll, pitch,
and yaw, respectively. The earth fixed coordinates (inertial
coordinates) define Xiaxis as true north, Yiaxis as east, and
Ziaxis as perpendicular downward. Xbaxis of the aircraft
coordinates coincides with Xiaxis of the inertial coordinates
when pitch and yaw angles are equal to 0. The coordinate
system of the aircraft is consistent in every flight modes.
C. Fuselage
The following subsections show the common specification
of both developed quad rotor tail-sitter UAVs. The airframe is
made of aluminum alloy and EPP (Expanded PolyProylene).
The main wing is a part of a commercially available R/C
airplane and the airfoil is NACA0010 whose span is 0.99 [m],
the chord is 0.28 [m], and taper ratio is 0.6. Since we do
not use the aileron, we fixed the aileron to prevent it from
moving in this research.
A fixed pitch propeller and a brushless DC motor are used
as propulsion units. The propeller diameter is 0.205 [m] and
the pitch is 0.152 [m]. As a result, a thrust to weight ratio
is more than or equal to 1.39.
D. Electronics
The UAV is equipped with the following processing units
and sensors.
The main computer is commercially available microcom-
puter board (Alpha project Co., STK-7125) which has an
SH2 microcomputer (Rnesas Technology Co.). This com-
puter executes control computation, sensor information pro-
cessing and transmission of a command signal (PWM:Pulse
Width Modulated) to each motor. The cycle of control
computation is 50 [Hz]. Flight logs are recorded on a micro-
SD card.
The attitude, three-axis angular velocity and translational
acceleration are measured by 3DM-GX1 producted by Mi-
crostrain Co. The sampling time is 100 [Hz]. The maximum
angle error is ±2 [◦]. GPS 18-5Hz from Garmin Co. is
used to obtain the position of the UAV. The accuracy of
horizontal direction is 4–5 meters and the accuracy of vertical
position is 10–20 meters. Therefore, altitude is mainly ob-
tained from one ultrasonic distance sensor (USS) for precise
measurements. The resolution of the USS is 0.025 meter and
the measurable range of the USS is 0 to 6.45 meter. One
ultrasonic distance sensor and one R/C servo are mounted at
the tail of the UAV. The attitude of tail-sitter UAV greatly
varies in each flight mode. Therefore, in order to measure
altitude with one sensor, the servo rotates the USS around
the Ybaxis. The movement of this servo does not affect the
attitude of the aircraft.
318

Altitude
δ
Distributor Aircraft
dynamics
Roll angle
Yaw angle
Pitch angle
Reference
Roll angle
Reference
Pitch angle
PWM1
Reference
Yaw angle
-
Reference
Altitude +
PID
Controller
Attitude
transition
strategy
∆hTtotal
Transform
into
Rotation Matrix
Transform
into
Rotation Matrix
Rref
1
2
3
+-
Rcur
1
δ2
δ3
Attitude and altitude controller
ω
Command
from
Operator Flight
Planner
ω
ω
PWM2
PWM3
PWM4
Fig. 3. Flight controller
III. CONTROL SYSTEM
A. Flight control system
Fig. 3 shows the block diagram of the flight controller. The
flight control system is designed based on the PID controller.
Firstly, the flight plan and reference parameters are in-
stalled in the “Flight planner”. After the UAV receives
start command from an operator, the planner generates the
reference attitude and altitude based on the prearranged flight
plan and sensor information including attitude and altitude.
Then, these references and current attitude are transformed
into rotation matrices, Rre f and Rcur , which are sent to the
“Attitude transition strategy”.
In the tail-sitter aircraft, quaternion feedback control is
often used due to no singular point. However, when the
attitude error is large, the quaternion feedback control may
fail to stabilize the UAV. Therefore, in this paper, “Resolved
Tilt-Twist Angle Feedback Control”, proposed in our previ-
ous research [9], is used to calculate the attitude error for a
tail-sitter VTOL UAV. The resolved tilt-twist angle feedback
control increases stability against large attitude disturbance.
Section III.Bbriefly reviews this control method.
The block of “Attitude transition strategy” generates errors
around the Xb,Yband Zbaxis of the coordinates of the aircraft
body. These errors are defined as
ω
j,(j=1∼3). These
attitude errors and altitude error are sent to the PID controller.
The PID controller generates the desired differential thrusts
for Xb,Ybaxis control, the desired torque for Zbaxis control
and desired total thrust for altitude control. These desired
difference thrusts, torque and total thrust are given as follows:
δ
j=KP
ω
j+KI
ω
jdt +KD˙
ω
j,(j=1∼3)(1)
Ttotal =KP∆h+KI∆hdt +KD∆˙
h+mg,(2)
where
δ
1,
δ
2, and
δ
3are the desired differential thrust for Xb
axis control, Ybaxis control, and the desired torque for Zb
axis control, respectively. Tt otal is the desired total thrust and
∆hdenotes the error between reference and current altitude.
The PID gains are provided by the ultimate sensitivity
method, and empirically tuned up.
These values are sent to the distributor, which calculates
revolution speed of each motor and transforms into control
command (PWM signal). The calculated PWM signals are
finally sent to each motor.
B. Attitude transition strategy
This subsection briefly reviews “Resolved Tilt-Twist An-
gle Feedback Control” [9]. In the above control sequence,
the attitude error is resolved into the tilt and twist angles.
The tilt angle is composed of two angles of orthogonal axes.
The method is composed of the following steps.
In the first step, the pitch and yaw errors are derived in
the analogy of inverted pendulum. The pitch and yaw errors
provide the tilt angle of the aircraft. Firstly, the error rotation
matrix REbetween the reference orientation Rre f and the
current orientation Rcur is determined as follows:
RE=RT
cur Rre f =

RE11 RE12 RE13
RE21 RE22 RE23
RE31 RE32 RE33
,(3)
Rre f =exr eyr ezr ,(4)
Rcur =exc eyc ezc.(5)
The elements of Xbaxis in REgives pitch and yaw errors
as follows:
θ
Y=atan2(RE31,RE11 ),(6)
θ
Z=atan2(RE21,RE11 ).(7)
Then,
θ
Yand
θ
Zdefine the tilt angle
θ
tilt as follows:
θ
tilt =
θ
2
Y+
θ
2
Z,(8)
In the second step, the roll error is derived. The attitude
of the aircraft after rotation of
θ
tilt becomes as follows:
Rv=exp(
v
θ
tilt ),for RE=E
E,for RE=E(9)
where Eis a 3×3 identity matrix, vis the rotation axis vector
given by normalized cross product of exr and exc as follows:
v=exc ×exr
|exc ×exr|≡vxvyvzT
,(10)
The hat operator 
{·} indicates a skew-symmetric matrix.
The UAV attitude RPafter
θ
tilt attitude change is given
by using Rvas follows:
RP=RvRc≡exp ey p ezp ,(11)
319

where ejp (j=x,y,z)are the unit vectors along jaxis of
the body coordinate frame after compensating the tilt with
respect to the inertial coordinate frame. The absolute roll
error is defined as follows:
θ
twist =cos−1ez p ·ezr
|ezp ||ezr|(12)
Since the range of roll angle of the aircraft is from -180 [◦]
to 180 [◦], the sign of the roll error must be identified. In
order to identify the sign of the roll error
θ
X,
θ
sign is defined
as follows:
θ
sign =cos−1eyp ·ezr
|eyp ||ezr|(13)
By using
θ
sign, the roll error
θ
Xof the UAV is identified as
follows:
θ
X=
θ
twist ,for
θ
sign ≤
π
2
−
θ
twist ,for
θ
sign ≥
π
2.
(14)
In the third step, the errors around the Xb,Yband Zbaxis
on the aircraft body coordinates are calculated to compensate
each axis error simultaneously. Therefore, the pitch and yaw
errors in the inertial coordinates must be projected onto the
aircraft body coordinates, which is expressed by rolling
θ
X
around Xbaxis in the inertial coordinates as follows:


ω
1
ω
2
ω
3
=

1 0 0
0 cos
θ
X−sin
θ
X
0 sin
θ
Xcos
θ
X


θ
X
θ
Y
θ
Z
.(15)
The above errors are finally sent to the PID controller.
IV. EXP ER IM EN TAL VERIFICATION
This section presents experimental verification of the
developed quad rotor tail-sitter VTOL UAVs. Firstly, we
compare the performance of the attitude control between the
cross-type of UAV and the asterisk-type of UAV to qualita-
tively analyze the effect of slipstream in the flight. Then,
the UAV with better performance is used to demonstrate
transition and level flight.
A. Effect of slipstream
In our previous research in [8], we confirmed that it is
possible to develop the quad rotor tail-sitter VTOL UAV
without using any control surfaces. However, the simulation
did not include the influence of the slipstream in detail.
Therefore, to experimentally analyze the effect of the slip-
stream, we developed two types of UAVs, namely cross-type
and asterisk-type of UAVs as shown in Section II. As one
can imagine, when the main wing is allocated right under
the propellers, the slipstream generates forces to rotate the
airframe to opposite direction from the rotational direction
produced by the anti-torque of the propellers around the Zb
axis since the slipstream hits on the surface of the main
wing. In the cross-type of UAV equipped with only main
wing, the effect of the anti-torque of the propellers right
above the main wing is counteracted by the effect of the
slipstream. Therefore, the UAV always easily rotate in one
direction. Accordingly, the effect of the slipstream leads to
the difficulty of the attitude control. To solve this problem,
0 2000 4000 6000 8000
0
0.05
0.1
0.15
0.2
0.25
Torque [Nm]
Propeller revolution speed [rpm]
Asterisk
Cross (only main wing)
Cross (with auxiliary wing)
CQ Asterisk = 0.027
Torque
Coefficients
Asterisk type
CQ cross= 0.014
Cross type
CQ cross auxiliary= 0.027
Cross type
(auxiliary wing)
Fig. 4. Torque of slipstream and torque coefficients
there are two solutions. One is to attach auxiliary wings
under the other propellers in the cross-type of UAV. Another
solution is to place the main wing in the area where the
slipstream does not affect to the attitude control, which is
here the asterisk-type of UAV.
In the following, we investigate the effect of the slipstream
by observing the change of the torque. The change of the
torque around the Zbaxis is observed by a torque sensor
on which the airframe is completely fixed. Fig. 4 shows the
experimental result in which the torque generated around the
Zbaxis is plotted with respect to the revolution speed of the
propeller. Fig. 4 clearly shows that the torque of the cross-
type of UAV is drastically reduced compared to that of the
asterisk-type of UAV. In general, the torque generated by the
propellers is expressed by the following equation.
Q=
ρ
n2D5CQ(16)
where nstands for the revolution speed of the propeller,
ρ
is the atmospheric density and Dis the propeller diameter.
The torque coefficient, CQdepends on the propeller form and
indicates the performance of the propeller. Fig. 4 shows the
torque coefficient of two different types of UAV. As shown
in Fig. 4, the asterisk-type of UAV performs almost twice
better than the cross-type of UAV.
B. Attitude control capability
We also carried out experiment of attitude control in
autonomous hovering mode with the cross-type of UAV, the
asterisk-type of UAV and the conventional quad rotor UAV
which has no wing. The total weight of the cross-type of
UAV is 1.18 [kg], the asterisk-type of UAV is 1.39 [kg] and
the conventional quad rotor UAV is 1.19 [kg].
In the experiment, the reference roll, pitch and yaw angles
are 0 [◦]. The flight experiment was performed outdoors, and
wind was blowing from the east to the west about 0.5 ∼1.0
[m/s]. After flying for twenty-three seconds, the experiment
was terminated and these UAVs landed on the ground.
Figs. 5 to 7 show the attitude profiles in the experiment.
In the figures, the dashed line depicts the reference attitude,
the red dotted line depicts the attitude of the cross-type of
UAV, the green solid line depicts the attitude of the asterisk-
type of UAV and the blue bold line depicts the attitude of the
conventional quad rotor UAV, respectively. As can be seen
from Figs. 5 and 6, these UAVs follow up the reference
320

0 5 10 15 20
−45
−30
−15
0
15
30
45
Roll angle [°]
Time [s]
Reference
Cross type
Asterisk type
Conventional type
Fig. 5. Comparison of roll angle
0 5 10 15 20
−45
−30
−15
0
15
30
45
Pitch angle [°]
Time [s]
Reference
Cross type
Asterisk type
Conventional type
Fig. 6. Comparison of pitch angle
0 5 10 15 20
−45
−30
−15
0
15
30
45
Yaw angle [°]
Time [s]
Reference
Cross type
Asterisk type
Conventional type
Fig. 7. Comparison of yaw angle
on the roll and the pitch. In the yaw axis, the attitude of
the asterisk-type of UAV and the conventional quad rotor
UAV converge to the reference while the cross-type of UAV
did not converge to the reference in the experiment. The
yaw control capability of both the asterisk-type of UAV and
the conventional quad rotor UAV was almost same in the
experiment. However, in the case when large disturbance
occurs such as strong wind, it can be predicted that the
yaw control capability of the asterisk-type of UAV should
be degraded.
In any case, the attitude of the asterisk-type of UAV is
more stable than that of the cross-type of UAV. Especially,
the attitude of the asterisk-type of UAV in the yaw axis shows
fast response compared to that of the cross-type of UAV.
Therefore, the experiment of transition and level flight is
carried out with the asterisk-type UAV.
C. Experiment of transition and level flight
The transition and level flight control strategy are experi-
mentally verified using the asterisk-type of UAV. The block
diagram of the flight controller is illustrated in Fig. 3. This
flight experiment was performed outdoors, and wind was
blowing from the east to the west at 1.0 [m/s].
The UAV precedes flight according to the following pre-
defined flight plan.
Step.1–Lifting up to start experiment: The UAV is lifted
up to about 6 [m] in hovering with manual control mode.
Step.2–Switching control mode: The control mode is
automatically switched to autonomous control.
Step.3-Stationary hovering: The UAV performs au-
tonomous stationary hovering on site for 2 seconds. At this
time, the reference roll and pitch are 0 [◦] and the reference
yaw is -35 [◦].
Step.4-Transition to level flight: After stationary hover-
ing, the UAV initiates transition flight to the level flight. At
this time, the reference roll is 0 [◦], yaw is -35 [◦] and pitch
angle is -75 [◦]. Step.4 continues until the pitch angle reaches
-75 [◦].
Step.5-Level flight: After the pitch angle reaches -75 [◦],
the UAV flies with the above mentioned reference attitude
and altitude for 8 seconds. The reference pitch angle is -
75 [◦] and reference altitude is 6 [m].
Step.6-Transition to hovering to terminate the exper-
iment: After the 8 second level flight, the UAV transits to
hovering flight and flies down to ground.
Fig. 8 shows snapshots of the transition and level flight
experiment. Please refer to the attached movie for more
information. The UAV begins to lift up in manual control
mode at t=0 [s]. At 20.2 seconds after the takeoff, the
control mode switches to autonomous control. As can be seen
in Fig. 8, the UAV maintained almost reference attitude, but
the altitude significantly changed after transition from the
hovering to the level flight. This cause is discussed later.
After flying for eight seconds in level flight mode, the UAV
switched to hovering to terminate the experiment and the
UAV landed on the ground. The total flight time was thirty
three seconds.
Figs. 9 to 11 show the attitude profiles during the experi-
ment. The attitude on every axes was successfully controlled
and it did not have large error until the end of the level flight.
During the transition flight from the hovering to the level
flight, which was started at t=22.2 [s], the pitch reached
the desired angle, -75 [◦], at t=23.0 [s]. As shown in the
figure, the level flight was successfully accomplished with
constant pitch angle. The maximum pitch error was 7 [◦] in
the level flight.
After the level flight, attitude was disturbed from t=
30.2 [s] due to the aerodynamics drag generated by transition
flight against the fixed wing. By rapidly lifting the pitch up,
unequal huge drag was generated in right and left side of the
main wing.
In the experiment, the altitude of the UAV was also
controlled by following the controller shown in Fig. 3.
However, the altitude significantly fell in transition flight
and it changed in level flight. The cause of altitude loss in
transition flight is a lack of wing lift. The transition flight
was performed quickly, resulting in producing very little
flight speed. At t=23.0 [s], the calculated flight speed is
about 5.3 [m/s], and the vertical component of the lifting
force (wing lift and rotor thrust) is 8.2[N]. As a result, the
wing lift is insufficient to compensate for the loss in rotor
lifting thrust. Hence, the UAV significantly descended. In
addition, the reason of the altitude change in level flight is
that the USS could not measure the proper distance from the
UAV to the ground due to grass field condition. Nevertheless,
we can see from Fig. 8 that the altitude in level flight was
lifted up. When we analyze the thrust distribution between
the vertical and horizontal component from the total thrust,
the maximum vertical component of thrust becomes 4.8 [N].
This indicates that the controller generated the lift force
321

20.99 [s] 22.19 [s] 22.28 [s] 23.01 [s] 23.13 [s] 23.22 [s]
Hover Transition Transition Transition
Level flight Level flight
Fig. 8. Sequential photographs of the transition and level flight
15 20 25 30
−90
−45
0
45
90
Rotation around Z axis of
ZXY Euler angle [°]
Time [s]
Reference
Result
Manual
Hovering
Autonomous
Hovering
Transition
to
Level flight
Level Flight
Transition
to
Hovering
Fig. 9. Rotation around Z axis of ZXY Euler
angle
15 20 25 30
−90
−45
0
45
90
Rotation around X axis of
ZXY Euler angle [°]
Time [s]
Reference
Result
Manual
Hovering
Autonomous
Hovering
Transition
to
Level flight
Level Flight
Transition
to
Hovering
Fig. 10. Rotation around X axis of ZXY Euler
angle
15 20 25 30
−90
−45
0
45
90
Rotation around Y axis of
ZXY Euler angle [°]
Time [s]
Reference
Result
Manual
Hovering
Autonomous
Hovering
Transition
to
Level flight
Level Flight
Transition
to
Hovering
Fig. 11. Rotation around Y axis of ZXY Euler
angle
greater than the aircraft weight despite of the measured
altitude was not accurate enough.
Consequently, we could verify the transition and level
flight with the developed quad rotor tail-sitter VTOL UAV,
which does not have any control surfaces. The control of the
altitude, however, was not perfectly achieved for the reason
of inaccurate altitude measurement by the USS due to the
grass field. This problem will be solved soon by carrying out
again in the different field condition and establishing more
sophisticated control strategy with a combination of several
sensors so that the UAV can fly in the altitude of our scope,
which is about several meters to several dozen of meters or
indoors of buildings.
V. CONCLUSIONS AND FU TU RE WORKS
This paper described development of two different types of
quad rotor tail-sitter VTOL UAVs named here cross-type and
asterisk-type of UAV. We discussed the autonomous flight
control including hovering, transition and level flight of the
developed quad rotor tail-sitter VTOL UAV.
Firstly, we verified attitude control capability of the two
developed UAVs to analyze the effect of the slipstream which
was not completely modeled in the simulation analysis in
our previous research in [8]. The verification showed that
the asterisk-type of UAV is superior to the cross-type of
UAV in terms of the attitude control capability. Therefore,
the experiment of transition and level flight was performed
with the asterisk-type of UAV. As a result, the asterisk-type
of UAV has succeeded in hovering, transition and level flight.
In the experiment, the attitude was controlled very well and it
could converge to the reference attitude, but the altitude was
not perfectly controlled. Since the lift force was generated
greater than the aircraft weight, the controller itself was
functional. However, the measurement of the altitude seemed
to have an error due to the grass field and the USS capability.
In the future, we will verify the improvement of energy
efficiency of the UAV in the level flight compared with
the conventional quad rotor helicopter. Moreover, we will
establish more sophisticated altitude transition strategy of the
quad rotor tail-sitter UAV for low altitude flight and flight
indoors building. Furthermore, we will analyze aerodynamic
characteristic of the developed UAV by using of wind tunnel
and try to find out optimal aerodynamic configuration for
hovering and level flight.
REFERENCES
[1] B. W. McCormick, “Aerodynamics of V/STOL Flight”, Academic
Press, 1969.
[2] H. Stone, G. Clarke, “Optimization of Transition Maneuvers for a
Tail-Sitter Unmanned Air Vehicle (UAV)”, in Proc. 5th Australian
International Aerospace Congress, Canberra, pp. 105, 2001.
[3] K. Kita, A. Konno, M. Uchiyama, “Transition between Level Flight
and Hovering of a Tail-Sitter Vertical Takeoff and Landing Aerial
Robot”, Advanced Robotics, Vol. 24, pp. 763-781, 2010.
[4] S. Shen, N. Michael and V. Kumar, “Autonomous Multi-Floor Indoor
Navigation with a Computationally Constrained MAV”, in Proc. IEEE
Int. Conf. on Robotics and Automation, pp. 20-25, Shanghai, 2011.
[5] M. Valenti, B. Bethke, D. Dale, A. Frank, “The MIT Indoor Multi-
Vehicle Flight Testbed”, in Proc. IEEE Int. Conf. on Robotics and
Automation, Rome, 2758-2759, 2007.
[6] L. A. Young, E. W. Aiken, J. L. Johnson, R. Demblewski, “New Con-
cepts and Perspectives on Micro-Rotorcraft and Small Autonomous
Rotary-Wing Vehicles”, Ames Research Center, Moffett Field, 2002.
[7] P. Sinha, P. Esden-Tempski, C. A. Forrette, J. K. Gibboney, G. M.
Horn “Versatile, Modular, Extensible VTOL Aerial Platform with
Autonomous Flight Mode Transitions”, in Proc. 2012 IEEE Aerospace
Conference, Montana, pp. 1-17, 2012.
[8] A. Oosedo A. Konno, T. Matsumoto, K. Go, K. Masuko, M.
Uchiyama“Design and Attitude Control of a Quad Rotor Tail-Sitter
Vertical Takeoff and Landing Unmanned Aerial Vehicle”, Advanced
Robotics, Vol. 26, pp. 307-326, 2012.
[9] T. Matsumoto, K. Kita, R. Suzuki, A. Oosedo, K. Go, Y. Hoshino,
A.Konno and M. Uchiyama, “A Hovering Control Strategy for a Tail-
Sitter VTOL UAV that Increases Stability Against Large Disturbance”,
in Proc. IEEE Int. Conf. on Robotics and Automation, Alaska, pp.54-
59, 2010.
322