A Rotor-Tilt-Free Tricopter UAV Design M
A Rotor-Tilt-Free Tricopter UAV Design M
A Rotor-Tilt-Free Tricopter UAV Design M
Belal H. Sababha*
Computer Engineering Department,
Princess Sumaya University for Technology,
Amman, 11941, Jordan
Email: [email protected]
*Corresponding author
Abstract: This paper presents an unconventional tri-rotor unmanned aerial vehicle (UAV) design
that only employs three brushless motors with three fixed pitch propellers for propulsion and
flight control. No additional mechanics for dynamically tilting motor(s), found on existing
tricopters, are used. The dynamic model of the proposed system is developed. Then a control
strategy to control the stability and manoeuvring of the UAV is presented. The control strategy is
achieved by only manipulating the rotational speeds of the propellers at each rotor. Two of the
rotors rotate in the same direction while the third rotates in the opposite direction. The control
methodology is novel compared to other systems that require either coaxial rotors or an extra
servo motor to control the yaw of the UAV. Results show that the proposed design achieved
stable flight with minimal position-attitude cross control effect. The fixed nature of the rotors in
the proposed design, reduced mechanical requirements and cost compared to existing vehicles of
its type.
Keywords: unmanned aerial vehicles; UAVs; tri-rotors; tricopters; vertical takeoff and landing;
VTOL.
Reference to this paper should be made as follows: Sababha, B.H., Al Zu’bi, H.M. and
Rawashdeh, O.A. (2015) ‘A rotor-tilt-free tricopter UAV: design, modelling, and stability
control’, Int. J. Mechatronics and Automation, Vol. 5, Nos. 2/3, pp.107–113.
Biographical notes: Belal H. Sababha is the Acting Dean of the King Abdullah II Faculty of
Engineering at Princess Sumaya University for Technology (PSUT) since 2015. He joined PSUT
as an Assistant Professor of Electrical and Computer Engineering in the Computer Engineering
Department on 2012. Before joining PSUT, he worked as a Senior Controls Engineer in the
Powertrain Controls Department at Chrysler Group LLC, Michigan. He received his PhD in
Electrical and Computer Engineering from Oakland University, Michigan in 2011. He has
extensive experience in embedded systems design, control algorithm design and software
development with applications related to gasoline engine controls and unmanned aerial vehicles
(UAVs). His research concentration areas are UAV development, sensor communication,
embedded RTOS, CAN networks, distributed embedded systems, graceful degradation in
embedded systems, rapid prototyping, and machine vision. He is an IEEE senior member and a
member of IEEE-RAS, ASME, AIAA and JEA.
Hamzeh Al Zu’bi is currently pursuing his PhD in Systems Engineering at Oakland University.
He received his BS in Mechanical Engineering with honours from Mutah University (Jordan),
and MS in Mechatronics Engineering from Oakland University in 2008, 2015 respectively.
Before joining Oakland, he has more than six years of experience in control design of unmanned
systems; his research interest is in autonomous control of unmanned aerial and ground vehicles.
He has four patents in technologies related to UAVs to his name.
Osamah A. Rawashdeh is an Associate Professor and the Academic Programs Coordinator in the
Department of Electrical and Computer Engineering at Oakland University, Michigan. He
received his BS with honours, MS, and PhD in Electrical Engineering from the University of
Kentucky in 2000, 2003, 2005 respectively. He is a member of ACM, AIAA, AUVSI, and a
senior member of IEEE. His research interests include unmanned systems, embedded systems
design, fault-tolerance, and reconfigurable computing. He is a Licensed Professional Engineer
(PE) in the State of Michigan.
Figure 1 Tri-rotor top view (Castillo et al., 2004; Padfield, 2007; Dorf and Bishop,
2010).
The dynamic model proposed for the tri-rotor presented
in this work is based on Newton-Euler 6-DOF mathematical
formulas for conventional rigid body multi-rotor UAVs
evolving in 3D space (Castillo et al., 2004; Padfield, 2007;
Dorf and Bishop, 2010). The following assumptions are
taken in consideration during the development of the
mathematical model of the tri-rotor UAV presented in this
work:
1 the body structure is rigid
2 the centre of gravity and the body fixed frame coincide
Figure 2 Tri-rotor front view
3 the propellers are rigid
4 thrust and drag are proportional to the square of
propellers’ speed
5 the drag factor of the rear rotor equals the summation of
the drag factor of the front two rotors.
The avionics system manipulates the rotational speed of the Table 1 Main physical effects acting on a helicopter
propellers to control the attitude of the vehicle and to make
Effect Source Formulation
it capable of hovering and lateral manoeuvring. The design
eliminates the need to use coaxial rotors or an extra servo Aerodynamic Propeller rotation
effects
motor. All three propellers are initially set to rotate with Blades flapping CΩ2
similar rotational speed. However, the controller may Inertial counter Change in propeller
JΩ
increase or decrease the rotational speed of any of the torques rotation speed
propellers to maintain attitude stability. The angular speed Gravity effect Centre of mass position
and torque of the rotors are controlled by an onboard
Gyroscopic Change in orientation Iθψ
microcontroller. effects of the rigid body
The two front rotors rotate in the same rotational
Change in orientation IΩrθ, φ
direction. The rear rotor rotates in the opposite rotational of the propeller plane
direction, such that adverse torque forces and gyroscopic
Friction All helicopter motion Cφ,θ, ψ
moment forces are reduced or even cancelled. This leads to
yaw moment control. The rotational speed difference
between the two front rotors produces the roll moment, Figure 3 Tri-rotor UAV rigid body configuration with earth and
while the pitch moment is created due to the variation of body frames
speed between the rear rotor and the front two rotors
collectively.
The three used motors are brushless type DC motors.
The two front rotors produce the same amount of torque and
lift. The torque resulting from the rear rotor has to be twice
the torque resulting from any of the other two rotors. To
achieve the required torque of the rear rotor, the mechanical
specifications, propeller dimensions, motor dimensions and
power requirements of the rear rotor have to be higher than
those of the front two rotors.
3 Dynamic modelling
This section overviews the dynamic model of the tri-rotor
that is presented in this paper. In general, the main physical
effects that affect a multi-rotor aircraft are summarised in
Table 1 (Mullhaupt, 1999). These effects are used to derive
the six degree of freedom (6-DOF) nonlinear dynamic Figure 3 illustrates the subject tri-rotor UAV configuration
equations for rigid body conventional multi-rotor aircraft with reference frames. The 6-DOF rigid body motion
equations are expressed as follows:
110 B.H. Sababha et al.
• Kinematic equations: altitude of the UAV. Commands from the ground station or
remote pilot include the desired pitch, roll, and yaw angles
φ = p + tan θ (q sin φ + r cos φ ), in addition to the desired altitude. Each PID controller
θ = q cos φ − r sin φ , (1) calculates the difference between the feedback and the
ψ = (q sin φ + r cos φ ) sec θ , reference and tries to eliminate error. The output of each
controller is mixed in the three mixing blocks to produce the
• Tri-rotor moment inputs: desired pulse width modulation (PWM) signals and send
them to the motor controllers to achieve the desired motor
U x = b1 ( Ω12 − Ω 22 ) l , speed for each of the motors.
U T = b1 ( Ω12 + Ω 22 ) + b2 Ω32 , It has to be mentioned here that each of the motors
(2) should have a high frequency response. Higher motor
U y = b1 ( Ω12 + Ω 22 ) l cos β − b2 Ω32 l , frequency response will result in lower oscillation in the
U z = ld1 ( Ω12 + Ω 22 ) − ld 2 Ω32 , pitch angle due to the interdependencies of pitch and yaw
controllers during command signal mixing. Section 5 will
UT, Ux, Uy, Uz are the total thrust, rolling moment, show the effect of motor frequency response on the
pitching moment and yawing moment, respectively, behaviour of the aerial vehicle. The execution rate of the
and b, d are the thrust factor and the drag factor, pitch control loop must be higher than the execution rates of
respectively. both roll and yaw control loops. Also the execution rate of
the roll control loop has to be higher than the execution rate
• Equations of motion: of the yaw control loop.
I xxφ = θψ ( I yy − I zz ) + J rθΩ r + U x ,
I yyθ = θψ ( I zz − I xx ) − J rφΩ r + U y , 5 Simulation environment and results
I zz ψ ( I xx − I yy ) + J r Ω
= θφ r +Uz ,
(3) In this section, we present the simulation environment setup
mz = mg − (cos ψ cos φ )U T , showing all simulation parameters as well as the simulation
mx = (sin ψ sin φ + cos ψ sin θ cos φ )U T , results of the designed control system. The simulation
environment has been designed and implemented using
my = (− cos ψ sin φ + sin ψ sin θ cos φ )U T ,
MATLAB® Simulink®, FlightGear flight simulator, and
AC3DTM graphical 3D modelling software. MATLAB®
Simulink® was used to implement the dynamic model of the
4 Stability control tri-rotor UAV as well as all sensory units’ models,
Figure 4 shows the avionics control system of the tri-rotor actuators’ models and PID controllers. Moreover, a software
UAV. The system is composed of two main components: interface was implemented in MATLAB® Simulink® to send
the four PID controllers and the three motor command the tri-rotor UAV states to the FlightGear software.
mixers. The four PID controllers are roll, pitch, yaw, and FlightGear reflects the received states of the UAV
altitude controllers. The command mixing blocks are for the behaviour via a graphical display of a tri-rotor 3D shape that
front right, front left, and rear motors. The feedback states was originally designed using the AC3D™ software.
are received from the gyros, accelerometers, ultrasonic, and Figures 5 and 6 show screenshots of the simulator interface
pressure sensors. These are defined as the attitude and the display.
Figure 5 Screenshot of the simulation environment display during a hover case of the tri-rotor UAV (see online version for colours)
Figure 6 Screenshot of the simulation environment display during a manoeuvre case of the tri-rotor UAV (see online version for colours)
Table 2 shows the tri-rotor computer aided design (CAD) Table 3 PID controllers’ gain values
model parameters that were used in the simulation.
kp kd ki
Similarly, Table 3 represents all attitude and altitude PID
Roll 6.6667 3.2456 0.02
controllers’ gain values.
Pitch 13.3334 11.18075 0
Table 2 Tri-rotor model parameters Yaw 0.5 0.147225 0
Altitude 10.5409 9.1423 0.01
Parameters Values
Mass (m) 2.375 kg First order transfer function used for the brushless motors
Moment of inertia (Ix) 51.2 ∗ 10−3 kg.m2 ( τ s1+1 ), is used to adjust the speed response of the motor in
Moment of inertia (Iy) 44.1 ∗ 10−3 kg.m2 the simulation environment. Figure 7 illustrates the tri-rotor
Moment of inertia (Iz) 3.8 ∗ 10−3 kg.m2 UAV behaviour while commanded to hold position and
Arm length (l) 0.288 m altitude. These results were collected when running the
Gravitational acceleration (g) 9.81 m/s2 simulation with the motor response set to 50 Hz. The red
Rotor inertia (Jr) 6 ∗ 10−5 kg.m2 lines represent the command, while the blue lines represent
Thrust coefficient (b1) 7.8 ∗ 10−5 N.s2 the response. Figure 8, shows the same results at 12 Hz
Thrust coefficient (b2) 13.3 ∗ 10−5 N.s2 motor response. It can be observed that the UAV’s
behaviour and stability was better at higher motor response.
Drag coefficient (d1) 7.5 ∗ 10−7 N.s2
The amount of drift in the X, Y position coordinates as well
Drag coefficient (d2) 15 ∗ 10−7 N.s2
as the oscillation in the roll, pitch, and yaw angles are much
less when the motor response was higher.
112 B.H. Sababha et al.
Figure 7 Hover case results at 50 Hz motor response (see online Figure 9 UAV behaviour while manoeuvring in a square path
version for colours) (see online version for colours)
Figure 8 Hover case results at 12 Hz motor response (see online Figure 10 3D plot of the UAV’s square path (see online version
version for colours) for colours)
6 Conclusions
Both figures show that the yaw angle drifts with time. The This paper presented a novel tri-rotor UAV design and
amount of drift is about 2.5 ∗ 10–3 rad ≈ 0.14 degrees in control system. The design only employs three brushless
about 40 seconds. It has to be mentioned that this drift could DC motors with three single fixed pitch propellers. The yaw
be automatically corrected by the avionics when the drift moment control is achieved by only manipulating the
reaches a certain threshold. The pitch-yaw cross control rotational speeds of the propellers at each rotor, without the
resulted in less than ±0.3 degrees of pitch angle oscillation. use of any conventional mechanical components for
Figure 9 represents the tri-rotor UAV behaviour while dynamically tilting motor(s) found on existing tricopters.
manoeuvring in a square path. In this manoeuvre the The paper presented the dynamic model of the proposed
altitude was set to 100 metres. The motors response was set system, as well as the control strategy to control the stability
to 50 Hz. The UAV was initially commanded to head ten and manoeuvring of the tri-rotor UAV. Simulation results
metres north, then ten metres west, then ten metres south show that the proposed design achieved stable flight with
and finally ten metres east. minimal position-attitude cross control effect. The reduced
During the manoeuvre, the yaw angle was set to always mechanical requirements resulted in a lower weight, and
point to the north (zero degrees). A positive or negative thus, cost less than existing vehicles of its type. On the other
pitch angle would result in heading north or south hand, reducing the mechanics resulted in a less than
respectively. Moreover, a positive or negative roll angle ±0.3 degrees of pitch angle oscillation due to the pitch-yaw
would result in moving the tri-rotor east or west cross control effect. Practically, this oscillation is
respectively. considered to be acceptable in many UAV applications.
A rotor-tilt-free tricopter UAV 113