control of ducted fan flying object using thrust vectoring*

Journal of System Design and

Dynamics

Vol. 6, No. 3, 2012

322

Control of Ducted Fan Flying Object Using Thrust Vectoring*

Masafumi MIWA**, Yuki SHIGEMATSU*** and Takashi YAMASHITA*** **Institute of Technology and Science, The University of Tokushima,

2-1 Minamijyousanjima-cho, Tokushima 770-8506, JAPAN E-mail: [email protected]

***Graduate School of Advanced Technology and Science, The University of Tokushima, Abstract Recently, R/C helicopter is used in fields of aerial photography and aerial investigation. But helicopter rotor blades are not covered, and the thrust is generated by high rotational speed. Thus R/C helicopter has a high risk of damage. In this study, we developed a new flying object using ducted fans instead of rotor blades. At first, PD control was employed for pitch and roll attitude control, but it caused steady state error. Moreover, PI-D control was used instead of PD control, and it reduced the steady state error. We succeeded to achieve stable hovering by 3-axes (roll, pitch and yaw axis) attitude control.

Key words: UAV, Ducted Fan, Attitude Control

1. Introduction

In recent years, light aircraft and helicopter are used in fields of aerial photography and aerial investigation. However, UAV (unmanned aerial vehicle) such as radio controlled (R/C) helicopter takes the place of the real aircraft in proportion to the improvement of the radio control technology. The operation cost of the R/C helicopter is lower than the actual one. In addition, required heliport size is smaller than that of actual one. Thus, the R/C helicopter has reasons why those are more excellent than an actual one, there are many reports of automatic control system for R/C helicopter (1) (2) (3), and some additional auto pilot units are available in the market. Also another type autonomous UAV such as quad rotor helicopter (4) (more stable), tilt rotor plane (5) (long cruising distance), tail-sitter (6) (fixed wing VTOL) are reported.

However, UAV operation still involves risks, and it causes a fatal accident and serious personal injury. An inexperienced operator, radio wave interference, wind disturbance, etc. are the factors responsible for these accidents. These UAV obtain thrust by rotor blades with very high rotating speed, and it makes UAV as dangerous object. Additionally, Thrusters of these UAV are set on upper or front level of Center Of Gravity (COG). This structure has an advantage that gravity acts on the airframe as horizontal stabilizer. On the other hand, it is difficult to hang payload to the airframe directly because thruster backwash exists.

To clear up the risk of rotor blade and thruster backwash influence for payload, we present a new flying object that uses ducted fans instead of rotor blades. Fig. 1(a) shows our new proposed airframe model. We named this new airframe as "Ducted fan Flying Object (DFO)". Size of the ducted fan is small and rotor wing is placed in the duct. So ducted fan help to decrease accident risk. We present the DFO as a pusher type inverted VTOL UAV which likes an inverted pendulum, because the inverted airframe has an advantage to cut

*Received 30 Nov., 2011 (No. 11-0732) [DOI: 10.1299/jsdd.6.322]

Copyright © 2012 by JSME

Journal of System Design and Dynamics

Vol. 6, No. 3, 2012

323

Load

Back wash

down the takeoff / landing area. The dynamic model of DFO is same as a pusher type aircraft when it flies vertically as shown in Fig. 1(b). Also DFO has an advantage in aerial transportation. Ducted fans are set at the end of airframe, thruster backwash will not affect on payload set on the upper part of the airframe (Fig. 2(b)). It is clear that DFO has COG above thruster units which causes instability on attitude of DFO airframe by gravity and it is the disadvantage of inverted airframe. The main purpose of this study is how to stabilize the inverted airframe by control algorithm.

The aim of this study is to achieve stable flight of DFO by 3-axes (roll, pitch and yaw axis) attitude control. Attitude control method by thrust vectoring with servomotors is presented. Also, hovering maneuver is controlled by three axes thrust vectoring (yaw axis, pitch axis, and roll axis) using servomotors. In this paper, we present the result of DFO’s experiment.

(a) Ducted fan flying object (b) Pusher type aircraft

Fig.1 Model of Ducted fan Flying Object and Inverted Aircraft

(a) With Helicopter (b) With DFO

Fig. 2 Aerial transportation with UAV

2. Ducted Fan Flying Object

Fig. 3 shows the Axes of DFO. Each axis is rotated with the torque expressed by multiplication of thrust component and distance from COG to ducted fan. Thrust component is generated by titling the ducted fan with servo motor.

(a) Pitch axis (b) Roll axis (c) Yaw axis

Fig. 3 Axes of Ducted fan Flying Object (DFO)

Load

Back wash


Vol. 6, No. 3, 2012

324

Fig. 4 Rotation model of single axis (pitch or roll) Fig.5 Rotation model of yaw axis In this study, we assume that each rotational motion on 3-axes (roll, pitch, and yaw) is

controllable independently when the attitude angles are small. Fig. 4 shows the roll/pitch axis rotational motion model of DFO, where θd is the target attitude angle of DFO, θf is angle of the ducted fan tilted by the servo motor. θo is attitude angle of DFO, G is the center of gravity, M is the mass of airframe (includes battery, sensors and CPU, etc.), m is then mass of the ducted fan units, R is the distance between to G and M, r is the distance between to G and m, and J is the entire moment of inertia of the airframe. Then equation of motion becomes Eq. (1).

oofo mgrMgRrFJ θθθθ sinsinsin +−= (1)

From the centrobaric momentum equilibrium, oMgR θsin and omgr θsin are countering with each other. When θ is so small, θθ ≅sin . And Eq. (1) becomes to Eq. (2) by Laplace transform.

f2

o ΘΘ rFJs = (2)

From Eq. (2), transfer function of DFO's attitude of roll/pitch axis is expressed in Eq. (3).

2f

o

ΘΘ

JsrF

= (3)

Fig. 5 shows the yaw axis rotational motion model of DFO, where θyd is the target attitude angle of yaw axis, θyo is rotate angle of yaw axis, r is the distance between yaw axis and the center of ducted fan, F is thrust of ducted fan, and θyf is the tilted angle of ducted fan. The equation of yaw axis rotational motion is expressed in Eq. (4),

yfy rFJ θsin20 =θ (4)

When θyf is so small, yfyf θθ ≅sin . And Eq. (4) becomes to Eq. (5) by Laplace transform.

yfy rFJs Θ=Θ 220 (5)

From Eq. (5), transfer function of DFO's yaw axis rotation is expressed in Eq. (6).

20 2

JsrF

yf

y =Θ

Θ (6)

3. Experimental Methods and Results

3.1 Experimental setup Fig. 6 shows the schematic diagram of experimental setup. Control gain and set point

are chosen by PC. These values are sent to CPU (AKI -H8/3069F : Akiduki E-Commerce) mounted on DFO through Bluetooth (BTX022 : Best technology). CPU receives triaxial angle and angular velocity from attitude sensor (3DM-GX3-25: MicroStrain). CPU transmits PWM signal for two ducted fan drive amplifiers (Pro.C S25: Tamazo) and 3 servomotors (W-150Power: WAYPOINT) to control thrust and its direction.


Vol. 6, No. 3, 2012

325

Yaw axis

Roll axis Pitch axis

Ducted fan

battery microcomputer sensor

Servo motor

Fig. 6 Schematic Diagram of Experimental Setup

Fig. 7 Experimental Airframe (DFO 1)

Weight 778g Maximum thrust 900gf Payload 122g Diameter of Duct 55mm Flight time 1min

Bluetoot

DFO

Sensor

ServomotorMicrocomputer

PWM signalTarget value

AttitudeState data

PC

Bluetooth

Target value


Vol. 6, No. 3, 2012

326

Fig. 7 shows the experimental setup of DFO. Airframe of DFO is made of MDF plate (2.5mm thickness) cut by laser cutter. DFO equipped with two ducted fan units (DF-55mk-2: Tamazo), and each ducted fan is mounted on gimbals respectively. First and second servomotors are capable of tilting these ducted fans along pitch axis, independently, which act as a yaw (rudder) and pitch (elevator) axis controller of airframe. Third servomotor tilts these ducted fans along roll axis in parallel, which acts as a roll (aileron) axis controller of airframe. Lithium polymer battery (11.1V-3S-1300: Hobby Net) is used as a power source for ducted fans and Lithium ion battery (ICR17335: YASHIMA DENGYO) is used to provide power for control system (CPU, sensor, Bluetooth). Maximum thrust of the experimental ducted fan is 450gf. Experimental airframe has two ducted fans, therefore, total thrust is 900gf while experimental airframe weight is 746g.

3.2 Experiment 1: PD control on yaw axis PD control was adopted for yaw axis control. Fig. 8 represents the block diagram of PD

control system, where, Kp is P gain, KD is D gain, GM is transfer function of servomotor, and F is thrust (constant). Fig. 9 shows the experimental method of yaw axis rotation control. PD control parameters were decided by repeated experiments. DFO was hanged with wire, and yaw axis rotation was free. Experiments were conducted as follows.

Step1: Test airframe was hold with error offset by hands. Step2: When control was started, test airframe was released. Step3: PD control measurements were continued until yaw axis gets stabilized for 3 to

5 seconds. Fig. 10 (a) shows the experimental result with -80 degrees error offset and Fig. 10 (b)

shows the result with 80 degrees error offset.

Fig. 8 Block diagram of PD control system

Fig. 9 Experimental method of yaw axis rotation control

In Fig. 10, steady state error existed on yaw axis. Therefore, we tried to find the cause of steady state error by changing ducted fan thrust in same yaw axis control experiments. Ducted fan thrust is varied by rotating speed of ducted fan motor. Thrust amplitude was set as 3.0N, 3.5N, and 4.2N. Fig. 11 shows the results.


Vol. 6, No. 3, 2012

327

(a) With +80 degree error offset (b) With -80 degree error offset

Fig. 10 Response of PD control on Yaw axis

(a) 3.0N

(b) 3.5N

(c) 4.0N

Fig. 11 Response of PD control on Yaw axis with 3 thrust condition In Fig. 11, steady state error was varied with the ducted fan thrust. Ducted fan thrust is

proportional to the rotational speed of motor and ducted fan motor torque increases with rotating speed, so the torque reaction of ducted fan motor is grown with rotating speed. Hence ducted fan thrust is varied by rotational speed of ducted fan motor. As a result, torque reaction of ducted fan motors affect steady state errors. In this study, ducted fan thrust was almost constant during experimental period and it means that the torque reaction was almost

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)time(sec)

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

time(sec)

angl

e(de

g)

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)


Vol. 6, No. 3, 2012

328

constant, too. Generally, I-Control is effective method to cancel steady state error caused by disturbance, but it needs integral time. In this case, the torque reaction is the source of the steady state error and we can estimate it from thrust amplitude and tilting angle of the ducted fan. Consequently, we select to put control offset instead of I control to omit the delay due to integral time. Then, we estimated the torque reaction and set control offset value for yaw axis control and adjusted it by repeated experiments.

(a) With +80 degree error offset (b) With -90 degree error offset Fig. 12 Response of PD control on Yaw axis with control offset

In Fig. 12, steady state error was decreased by added offsets. As can be seen from Fig.

12(a) and Fig. 12(b), small amplitude steady state error still existed on yaw axis. However, we assume that the steady state error does not affect other axis tuning.

3.3 Experiment 2: PI-D control on roll and pitch axis Next, we tried to realize hovering maneuver with triaxis attitude control by PD control.

PD control parameters for roll axis and pitch axis were decided by repeated hovering experiments. Yaw axis PD parameter and control offset were set as same as those in § 3.2. Fig. 13 shows the scene of hovering experiments. Experiments were performed as follows. Step1: Test airframe was hold with error offset by hands. : Fig. 13(a) Step2: When control was started, test airframe was released. : Fig. 13(b) Step3: About 5 second later, test airframe was caught, and experiment was completed.

Test airframe drifts during hovering maneuver because position control is not installed yet. Therefore, we selected the target angles of roll and pitch attitude control that reduce airframe drift by repeated experiments. These target angles are not zero but vertical component of thrust passes the COG of test airframe when attitude converge to the target angle. According to repeated hovering experiment results and test field size, hovering test time was within 5 seconds.

(a) Before hovering test (b) During hovering test

Fig. 13 scene of hovering experiments

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)


Vol. 6, No. 3, 2012

329

As a result, stable attitude holding were obtained in all three axes, and stable hovering was achieved. However, steady state error of each attitude control existed. For example, when we added 15 degree error offset, steady state error was 5 degree. Also we added -15 degree error offset, steady state error was -4 degree. To eliminate these steady state errors, PI-D control (7) was adapted to roll and pitch axis attitude control. Fig. 13 shows the block diagram of PI-D control system. KI is I gain.

Fig. 14 Block diagram of PI-D control system

(a) With +13 degrees error offset (b) With -30 degrees error offset

Fig. 15 Response of PI-D control on Roll axis

(a) With +22 degrees error offset (b) With -16 degrees error offset

Fig. 16 Step Response of PI-D control on Pitch Axis

Fig.15 shows the experimental results of PI-D control on roll axis. Fig.15 (a) shows experiment started with + 13 degrees error offset and Fig.15 (b) is a graph started with -30 degrees error offset. In Fig. 15(a), roll angle attitude oscillated. Its oscillation center was almost same as target angle, and frequency was about 1Hz. In Fig. 15(b), steady state error became as small as about 2 degree.

Fig. 16 shows the experimental results of PI-D control on pitch axis. In Fig. 16 (a), experiment started with + 22 degrees, Fig. 16 (b) is a graph started with -16 degrees. In both case, pitch angle reached target angle at once. But pitch angle went out from target point. The reason is not clear, but it might be due to battery voltage drop during hovering experiments.

-30

-20

-10

0

10

20

30

0 1 2 3 4

θd

θo

angl

e(de

g)

time(sec)

-30

-20

-10

0

10

20

30

0 1 2 3 4

θd

θo

angl

e(de

g)time(sec)

-30

-20

-10

0

10

20

30

0 1 2 3 4

θd

θo

angl

e(de

g)

time(sec)

-30

-20

-10

0

10

20

30

0 1 2 3 4

θd

θo

angl

e(de

g)

time(sec)


Vol. 6, No. 3, 2012

330

Weight 1720g → 942g UPMaximum thrust 1960gf →1060gf UPPayload 240g → 118g UP Diameter of Duct 69mm → 14mm UPFlight time 5min → 4min UP

From Fig. 15 and Fig. 16, PI-D control does not eliminate steady state error completely but it reduces steady state error and makes attitude close to target angle. It clearly shows the effectiveness of PI-D control.

4. DFO2

4.1 New Experimental Airframe DFO in § 3 has too small payload to equip larger battery. Therefore flight time is so

short and altitude decreased by voltage drop during the entire experiment. Thus, we developed a new airframe with larger ducted fan. We named this new airframe as "DFO2". Fig. 16 shows DFO2. Airframe of DFO2 is made of plywood plate (4mm thickness) cut by laser cutter. Basic structure is same as DFO. DFO2 consists of large ducted fan (DF-69 typeⅡ：Tamazo) , ducted fan drive amplifier (Phoenix Ice Lite 50: Castle creations), servomotor (MG90 : Tower Pro), attitude sensor and MPU. Two lithium polymer batteries (14.8V-4S- 2200mAh: TURBO) are used for each ducted fan, and Lithium ion battery (ICR17335: YASHIMA DENGYO) is used for control system (CPU, sensor, Bluetooth and servomotors). Bluetooth（Parani SD1000 : SENA） is used for communication between PC and CPU. Thrust of the ducted fan has max 1000gf. DFO2 has two ducted fans, therefore total thrust is 2000gf. Weight of DFO2 is 1590g.

Fig. 16 Experimental Airframe (DFO2)

4.2 Experiment 4: PI-D control on yaw axis At first, PD control with step offset same as described in § 3 was tested for yaw axis.

Fig. 17 shows the experimental results.


Fig. 17 Response of PD control on Yaw axis with control offset As the results, we succeed to decrease the steady state error, but it still existed. Then,

PI-D control was used in yaw axis to improve the precision of control. Fig. 18 shows the

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(se

c)

time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)


Vol. 6, No. 3, 2012

331

experimental results. As shown in Fig. 18, PI-D succeeded to eliminate steady state error on yaw axis.


Fig. 18 Response of PI-D control on Yaw axis

4.3 Experimental 5: PI-D control on 3 axes. Next, we tried PI-D control on pitch, roll and yaw axis at the same time. Fig. 19 shows

the experimental results. As shown in Fig. 19, we succeed to converge and eliminate steady state error at the target value on all axes. Also PI-D control succeeded to achieve hovering maneuver.

(a) Pitch axis (b) Roll axis

(c) Yaw axis

Fig. 19 Response of PI-D control on triaxis

4.4 Experimental 5: Take-off test. Take-off test was conducted using following step by step procedure. Throttle was

opened by manual operation. Step1: Test airframe was set on floor, and target angles of each 3 axis were set. Target angle

of pitch was -6 degree, roll was 0 degree, and yaw was 120 degree. Step2: Each PI-D control was started. Step3: Throttle was opened slowly. Step4: After DFO took off and it reached appropriate height, throttle was fixed to hold the

height.

0

20

40

60

80

100

120

140

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)time(sec)

-60

-40

-20

0

20

40

60

80

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

90

100

110

120

130

0 1 2 3 4 5

θyd

θyo

angl

e(de

g)

time(sec)

-20

-10

0

10

20

0 1 2 3 4 5

θd

θo

angl

e(de

g)

time(sec)

-20

-10

0

10

20

0 1 2 3 4 5

θd

θo

angl

e(de

g)

time(sec)


Vol. 6, No. 3, 2012

332

-20

-10

0

10

20

0 1 2 3 4

θd

θo

angl

e(de

g)

time(sec)

-20

-10

0

10

20

0 1 2 3 4

θd

θo

angl

e(de

g)

time(sec)

(a) Pitch axis (b) Roll axis

100

110

120

130

140

0 1 2 3 4

θyo

θyo

angl

e(de

g)

time(sec)

(c) Yaw axis Fig. 20 Response of PI-D control of take-off test

Fig.20 shows the attitude change during take-off test. At 0.6 second from experiment

start, DFO2 took off. At 2.0 second after experiment start, DFO2 altitude reached 30 cm and it stared hovering with constant altitude. Before take-off, roll and pitch angles were different from each target angle because DFO2 was put on floor and it was vertical to the floor. After DFO2 took off, it started hovering while PI-D control held the attitude and both angles were converged to target angles. In yaw axis, yaw angle started to vary at 0.3 second before take-off timing. This change was caused the contact/non-contact condition change between DFO2 landing gear. Contact / non-contact condition causes floor reaction force to change, and it affected yaw control. Finally, yaw angle was converged to target angles, too. This result shows that PI-D control for three axes is effective for take-off operation.

4.5 Experimental 5: Manual flight test with PI-D control. Next, we tried manual flight test with PI-D control. In this experiment, we set target

angles for roll pitch angle and throttle using PC with Bluetooth device during test flight. Target angle for yaw angle was kept constant. When new target angle was set, PI-D control tilts the DFO airframe to the target angle. Then, ducted fan is tilted to hold the DFO attitude, and it generate thrust horizontal component. This component will move DFO. So we can able to move DFO by tilting it. Manual flight test was done as followings. Step1: Test airframe was hold. Step2: When control was started, test airframe was released. Step3: Target values of roll and pitch axis were changed during test flight. Step4: About 8 second later, test airframe was caught, and experiment was finished.

Fig. 21 shows the results of manual flight test. In Fig. 21, attitude angles were followed to the target angles, but DFO2 was oscillated. In pitch and roll axis, oscillation center were followed to each target angle and amplitude was about 2 degree as shown in Fig. 21(a) and (b). In yaw axis, relatively large error (up to 8 degree) appeared at 4 second and 7 second. However, PI-D control reduced the relative error. But oscillation was existed as shown in Fig. 21(c). Additionally PI-D control could not suppress oscillation and steady state error

Take-off timing Take-off timing

Take-off timing


Vol. 6, No. 3, 2012

333

completely, moreover, DFO2 moved following to desired direction as the result of PI-D control for attitude. These oscillations have an insignificant effect on manual flight.

- 1 0

- 5

0

5

1 0

0 1 2 3 4 5 6 7 8 9

θ d

θ o

angl

e(de

g)

t i m e ( s e c )

(a) Pitch axis

- 1 0

- 5

0

5

1 0

0 1 2 3 4 5 6 7 8 9

θ d

θ o

angl

e(de

g)

t i m e ( s e c )

(b) Roll axis

- 8 0

- 7 0

- 6 0

- 5 0

- 4 0

0 1 2 3 4 5 6 7 8 9

θ y d

θ y o

angl

e(de

g)

t i m e ( s e c ) (c) Yaw axis

Fig. 21 Response of PI-D control of manual flight test In § 4.4, steady state error was converged in take-off procedure but it was not

suppressed in this experiment. The reason is not clear, however, the difference between take-off and manual flight is change of target attitude angle. It might be explained as followings. Ducted fan is a high rotational speed machine, and it will generate Gyro moment when it tilted in a large angle. When take-off test, target angle was not changed, and ducted fan was tilted in a little angle to hold the attitude. Thus, generated Gyro moment was too small to affect on attitude. On the other hand, target angle was changed in manual


Vol. 6, No. 3, 2012

334

flight, ducted fan was tilted in a large angle, generated Gyro moment was large enough to affect on attitude.

5. Conclusions

In this study, we developed new type airframe DFO. PI-D control was introduced to hold the attitude angle of DFO and it succeeded to keep hovering maneuver. Since DFO has a small payload, the flight time is very short because DFO is able to load light battery only. Thus, we developed DFO2 to increase the flight time. DFO2 was developed using larger ducted fan and loaded bigger battery. As a result of an experiment, PI-D control succeeded to eliminate steady state error at the target value on all axis. Further, take-off test and manual flight test were successfully carried out. However, in manual control, small oscillation and small steady state error appears. These oscillation and steady state error can be attributed to Gyro moment of ducted fan tilting motion. When we started this study, only normal rotation ducted fan was available and its torque reaction causes steady state error on yaw axis control. But now, we can get normal rotation and reverse rotation ducted fan from the market that it will balance out Gyro moment. Thus, we have developed a new DFO3 equipped with normal and reverse rotation ducted fan. We will report automatic take-off / landing, position control and cruise control with GPS and altitude sensor with DFO3 in the future study.

Flying time of DFO is shorter than normal aircraft because the maneuver of DFO is achieved by the reaction of ducted fan thrust only and it causes early battery drain. Moreover, the dynamic model of DFO is almost same as a pusher type fixed wing aircraft when it flies vertically. So, we will develop a new DFO4 by applying this study results to a pusher type aircraft which equips thrust-vectoring. Take-off / landing and hovering motion of DFO4 will be done with vertical mode (same as DFO1, 2, 3) and DFO4 cruises in horizontal mode same as normal airplane. We will study DFO4 as a new type VTOL aircrafts in near future.

Acknowledgement

This work was supported by NSK Foundation for the Advancement of Mechatronics.

References

(1) J. Shin, D. Fujiwara, K. Hazawa and K. Nonami, Model Based Optimal Attitude and Positioning Control of Small-Scale Unmanned Helicopter, Tran. Of the Japan Society of Mechanical Engineers, Vol.70, No.697 (2004), C, pp.2631-2637

(2) S. Suzuki, D. Nakazwa, K. Nonami and M. Tawara, Attitude Control of Small Electric Helicopter by Using Quaternion Feedback, Journal of System Design and Dynamics, Vol.5, No.2 (2011), pp.231-247.

(3) M. Miwa, I. Shiraishi, M. Matsushima and K. Minami, Remote Control Support System for R/C Helicopter, ServiceRobotics and Mechatronics, (2010), pp.125-130, Springer-Verlag

(4) D. Pebrianti, F. Kendoul, S. Azrad, W. Wang and K. Nonami, Autonomous Hovering and Landing of a Quad-rotor Micro Aerial Vehicle by Means of on Ground Stereo Vision System, Journal of System Design and Dynamics, Vol.4, No.2 (2010), pp.269-284

(5) S. Suzuki, R. Zhijia, Y. Horita, K. Nonami, G. Kimura, T. Bando, D. Hirabayashi, M. Furuya and K. Yasuda, Attitude Control of Quad Rotors QTW-UAV with Tilt Wing Mechanism, Journal of System Design and Dynamics, Vol.4, No.3 (2010), pp.416-428

(6) D. Kubo and S. Suzuki, Transitional Flights of a Tail-Sitter Vertical Takeoff and Landing Small Aerial Robot, Proceedings of 45th AIRCRAFT SYMPOSIUM, (2007 ) , Page2H2

(7) M. Kawata and K. Nishioka, Easy Learning Text for Control Engineering with Matlab & Simulink (in Japanese), (2001), pp.92, Morikita.

control of ducted fan flying object using thrust vectoring*

Documents