Quad-cone-rotor: A Novel Tilt Quadrotor with Severe-fault-tolerant Ability

Zhe Shen, Yudong Ma, Takeshi Tsuchiya

The University of Tokyo [email protected]
Abstract
Conventional quadrotors received great attention in trajectory design and fault-tolerant control in these
years. The direction of each thrust is perpendicular to the body because of the geometrics in mechanical
design. Comparing with the conventional quadrotor, a novel quadrotor named quad-tilt-rotor brings
better freedom in manipulating the thrust vector. Quad-tilt-rotor augments the additional degrees of
freedom in the thrust, providing the possibility of violating the normal direction of the thrust in the
conventional quadrotor. This provides the ability of greater agility in control. This paper presents a novel
design of a quad-tilt-rotor (quad-cone-rotor) whose thrust can be assigned along the edge of a cone shape.
Besides the inheriting merits in agile from quad-tilt-rotor, the quad-cone-rotor is expected to take fault-
tolerant control in severe dynamic failure (total loss in all thrusts). We simulate the control result in a UAV
simulator in SIMULINK, MATLAB.

Keywords
Tilt-rotors, fault-tolerance, modeling, Design, Control

I. INTRODUCTION There are two typical designs of these quad-tilt-rotors. One is the
quad-tilt-rotor with eight inputs (four for the magnitude of the thrust;
four for the direction of the thrust) [9]. The four servo motors provide
In a conventional quadrotor, the thrust is perpendicular to the the possibility of changing the direction of each thrust by rotating
body. The inputs are the angular velocity of each motor which them along each arm. The other quad-tilt-rotor owns twelve inputs
generates the thrust to control attitude and the altitude. We should [10]. The function of the eight motors among its twelve motors is
note that the number of the input (four) is less than the degrees of identical to the function of the eight inputs in the previous quad-tilt-
freedom (6). Thus, following an arbitrary trajectory with the specific rotor. The extra four motors are used to further adjust the direction
requirement in attitude can be impossible for this under-actuated of thrusts, which provides more freedom in selection of the potential
system [1]. A common method to deal with this under-actuated direction of the thrust.
problem is to find the relationship between the actuated variables
and the variables not actuated directly. [2-5] approximates this The promising fact is that the number of the inputs (eight or
relationship using linearization and infinitesimal equivalence. Thus, twelve) is more than the quantity of the variables (six). Indeed, the
the degree of freedom is reduced to four, which is equal to the quad-tilt-rotor becomes an over-actuated system with input
number of the inputs. redundancy, which brings the possibility of tracking the desired
trajectory and the designed attitude at the same time [13]. However,
Although the conventional quadrotor can be stabilized to track the the complexity introduced by the extra inputs burdens the
desired trajectory [5-8], the inevitable attitude change in translational development of the proper controller to manipulate these novel
movement can influence the onboard camera. Thus, the camera- quad-tilt-rotors. A typical method in controller design is to decrease
based landing task can be a challenge. With these considerations, the number of degrees of freedom by restricting the allocation of the
novel quadrotors named quad-tilt-rotor or tilted rotor UAV are servo motor on purpose [2,3,14].
developed [9-12]. Compared with the conventional quadrotor, the
quad-tilt-rotor can adjust the direction of its thrust; several extra Another notable advantage of quad-tilt-rotors is its ability in fault-
servo motors make it possible. tolerant control [4,11,12,14-16]. The redundancy gifted by the tilt
mechanism entails the different fault-tolerant control results with the
conventional UAV. The common faults studied are partially or Apart from the blue sector, there are three red cones in Figure 2.
complete loss in one or several thrusts [15-18]. Note that the total These potential cones are the designed potential thrusts in our
loss in all thrusts is beyond the scope of these studies; these quad-tilt- quadrotor. The designed thrust can also change its direction, in a
rotors are determined to fall down without the possibility of getting unique pattern (along the surface of a cone), however. We decide one
controlled. cone in Figure 2 as the ‘orbit’ for our thrust by deciding on a proper
cone-angle. The relevant consideration in picking this cone-angle is
This research proposes a novel design of a quad-tilt-rotor whose detailed in Section V.
degree of freedom is eight. The direction of the thrust of each rotor
can be selected along the surface of a cone shape. Thus, we name it We name this kind of quad-tilt-rotor quad-cone-rotor because of the
quad-cone-rotor. The dynamics are deduced based on Newtown- cone shape of its potential direction of the thrust.
Euler equations. A simulator is subsequently modeled based on these
dynamics in SIMULINK, MATLAB. Symmetrical open-loop input test is
conducted for both cases where the thrusts are healthy and where
the driven-thrust motors are in stuck (total loss in all thrusts).

This paper is structured as follows: Section II illustrates the design

and the dynamics of the quad-cone-rotor. The model of the thrusts
and drag moments is explained in Section III. The symmetrical open-
loop hovering test and the fault-tolerant open-loop hovering test are
detailed in Section IV and V, respectively. Section VI presents the
trade-off analysis in designing a cone-angle for a quad-cone-rotor. At
last, the conclusions and discussions are made in Section VII.

II. Quad-cone-rotor Dynamic Model

Fig. 2. The blue sector is the potential thrusts in [9]. The red cones
This section is to provide the kinematics design and the dynamical are the potential thrusts in our quad-cone-rotor.
model of the novel quad-cone-rotor. Before illustrating the
kinematics, we start with the introduction to the designation at the
end of each arm. The thorough mechanical fulfillment of this quad-cone-rotor is
beyond the scope of this study. However, we propose the draft of one
possible design (Figure 3) to explain the kinematics of our quad-cone-
Direction of The Thrust rotor. Figure 3 is the structure at the end of each arm. There are two
motors, the ‘cone motor’ and the ‘rotor motor’.
To better introduce the design of our quad-cone-rotor, we refer to
the quad-tilt-rotor designed in [9]. As plotted in Figure 1 in [9], they The cone motor is fixed at bottom. It can work as a servomotor
proposed the structure where the direction of the thrust can be adjusting the direction of the thrust. Specially, the link fixed on the
selected perpendicular to each arm. top of the cone motor is not parallel to the axis of the cone motor.
Instead, there is an angle biased from the axis of the cone motor.
This design determines the zone (cone shape) where the potential
directions of the thrust can be. Section V explain the selection of this
cone-angle, .

The other motor which directly drives the rotor is called ‘rotor
motor’. It mainly determines the magnitude of the thrust if the
coupling effect of the cone motor is not considered.

The length of the link between the cone motor and the rotor
motor is . The orientation of this link is carefully designed; it
precisely points upward when the cone motor is at the left-most
position.

Fig. 1. The design of a quad-tilt-rotor in [9] It is worth mentioning that the cone motor has two modes. One is
working as a servomotor; the other mode is the speed motor. While
the rotor motor has the speed motor mode only.
We plotted the direction of the potential thrust in blue in Figure 2.
These potential thrusts constitute a sector, assuming that the thrusts
are equal. Note that this sector is in a plane perpendicular to the arm.
 We assume the angle adjusted by the cone motor (fixed with )
is . Thus, the angular velocity of the cone motor is . Both
directions of and are CCW observed from the top of . In
addition, the angular velocity of the rotor motor (fixed with ) is
assumed to be . The direction of is CCW observed from the top
of .

Similarly, the relevant frames, , , and

, are defined at the end of the first, third, and fourth arm,
respectively. The corresponding angles and angular velocities
, , , are also defined. Note that we define the
Fig. 3. The structure at the end of each arm.
positive directions of angular velocities for both in
proposition 1.
At the end of each arm fixes this structure. So that the direction of Proposition 1: (positive directions of the angles and the angular
each thrust can be along a cone. We assume that the angle of the velocities) The positive direction of the angular velocities, at
cone for each thrust is identical ( ).
the end of the arm , are defined in (1):

(1)
Kinematics

Several coordinate frames are demonstrated in Figure 4. For

For instance, the positive directions of are along the
simplicity and without losing generality, we only plot typical frames
for one arm. The frames on the rest arms are similar. positive direction of and , respectively. While the positive
direction of are defined as the negative direction of and
, respectively.

We assume that generates the positive thrust (the

positive direction of ) by the definition. It means that CW rotation
along or generates the positive thrusts. While the CCW
rotation along or generates the positive thrusts. It is the
consequence of the mechanical structure of the propellers.

We also restrict that

(2)

We do not expect the negative angular velocity of the rotor motor.

Fig. 4. The frames in the quad-cone-rotor. However, the negative angular velocities of the cone motors ( ) are
allowed; the direction of each thrust can be adjusted by rotating the
cone motor CCW or CW.
Let be the body-fixed frame. It is fixed at In the following text, we use to represent the
the geometrical center of the quad-cone-rotor. The is along the rotation matrix along , respectively.
first arm; the is along the second arm. The number of the arm is In addition, we use and to represent and ,
assigned in CCW (counter clockwise). respectively.
is the frame fixed on the second cone The relevant matrix is demonstrated in Equation (3)-(5).
motor. It is transformed from the by rotating along by angle
. The negative sign represents the direction in rotation is CW
(3)
(clockwise).

The rotor frame is fixed on the second

rotor motor. Since is not fixed on the propeller, it does not rotate (4)
with the propeller. However, it will change the position with respect
to if the cone motor changes the angle.
 from the tilt rotor, ignoring the transient dynamics during the
(5) directional change in thrust. While [9] takes this effect into serious
consideration by analyzing the dynamics in each frame.

The transformations between each frame are illustrated in (6) We analyze the dynamics in each frame to describe the motion of
our quad-cone-rotor in a detail way in this study. The relationship
between each part is sketched in Figure 5. The drag (the cause for
drag moment) and the thrust affect the propeller directly. These
(6) effects are described in the rotor frame .
The notation, , represents the physical quantity, , with
The links the rotor frame and the cone
respect to the frame, , expressed in the frame, . When is
motor frame , conveying the dynamical effect of the drag and the
the inertial frame, is omitted (e.g., ). When is the frame
thrust. After that, the links the body-fixed frame
fixed on the body of , is omitted (e.g., ).
and the , further conveying the dynamical effect.
We use to represent the angular velocity of the propeller
expressed in the rotor frame, . We use to represent the
angular velocity of the cone motor expressed in its own frame (cone
frame, ). We use to represent the angular velocity of the body
expressed in its own frame (body-fixed frame, ). All of these
angular velocities are with respect to the inertial frame.
Fig. 5. The relationship between each part.
We have the following relationship for in Equation (7).

(7)
Several reasonable assumptions and approximations are made to
Considering the relationship in (6), Equation (7) can be rewritten simplify our modeling.
in Equation (8).
Proposition 2: (direction of the thrust) The direction of the thrust
is perpendicular to the propeller and is affecting at the mid of the
propeller. Thus, the thrust is in the positive direction of .

Remark 1: With proposition 2, the thrust contributes no moment

to the

Proposition 3: (direction of the drag moment) The direction of the

and are illustrated in Equation (9)-(10).
drag moment is perpendicular to the propeller and is affecting at the
mid of the propeller. Furthermore, the thrust is in the negative
(9) direction of (along ). Thus, it tends to decrease the angular
velocity of the relevant rotor motor.

Proposition 4: (mass of the body) The mass of the body is much

larger than the mass of the cone motor or the rotor.
(10)
Remark 2: With proposition 4, the thrust can be regarded as the
force exerting directly on the body.
Similarly, we have the relationship for in Equation (11).
Remark 3: Based on Proposition 2-4 and Remarks 1-2, the thrust
can is affecting on the body directly. Thus, we have the equivalent
(11)
relationship for each part in Figure (6).

Dynamics

The method in deducing the dynamic equation for our quad-cone-

rotor in this study is Newtown-Euler method. The total movement of
the vehicle comprises rotation and translation.

1) Rotation
Fig. 6. The equivalent relationship between each part.
Several studies (e.g., [2,10]) regard their targeted quad-tilt-rotor
as an entirety without considering the effects brought by the rotation
 Based on this, we write the dynamic equation for each of these
three separated parts based on Figure 5.

For , we have Equation (12).

(12)

is the drag moment from the air. We will discuss it in detail

later in Section III.

is the torque contributed by .

Fig. 7. The simulator of the quad-cone-rotor.
For , we have Equation (13).

(13)
As can be seen from Figure 7, there are eight inputs in total. Four
of them are the angular velocities of the cone motors. And the rest
is the torque contributed by .
are the angular velocities of the rotor motors.
For , we have Equation (14).

III. Thrusts and Drag Moments

(14)

is the torque provided by the thrusts. It is detailed together

This section explains the method we model the thrust and the drag
with later in Section III. moments.

2) Translation Several studies (e.g., [20,21]) take the effect of blade flapping into
consideration. To simplify our modeling process, this effect is beyond
The translational motion of the quad-cone-rotor is
the scope of this research and can be a further topic.
straightforward. It is described in Equation (15).
Typical formulas modeling the thrust and drag moment in are in
Equation (19)-(20), respectively.
(16)
(19)

in Equation (16) is a rotation matrix transforming the (20)

thrust expressed in the body-fixed frame into the thrust expressed in
the inertial frame. It is detailed in Equation (17). These thrust and drag moment models are generally used in the
conventional quadrotors [22,23] and the tilt UAV [24]. It is worth
mentioning that in Equation (19) and (20) in a conventional
(17)
quadrotor is the angular velocity of the propeller with respect to the
body-fixed frame. Although a more reasonable way is to deduce the
Where are the roll, pitch, yaw, respectively. angular velocity with respect to the inertial frame, we still deduce it
with respect to the body-fixed frame.
The relationship [19] between roll, pitch, yaw and the angular
The angular velocity of the propeller in the quad-tilt-rotor is a little
velocities is given in Equation (18).
different. In [9], the drag-moment-related angular velocity is with
respect to the body-fixed frame. While the thrust-related angular
velocity is with respect to its previous frame; it is the angular velocity
(18)
of the speed motor.

In this research, both the drag-moment-related and the thrust-

related angular velocities of the quad-cone-rotor are with respect to
the body-fixed frame.
Simulator
Proposition 5: (thrust and drag moment) The thrust and the drag
Based on the aforementioned relationships in the kinematics and
moment in quad-cone-rotor are modeled in Equation (21)-(22).
the dynamics, we build the simulator for our quad-cone-rotor in
SIMULINK, MATLAB, environment (Figure 7). (21)

(22)
 in Equation (21)-(22) is the part of the angular velocity of the The angular velocity, , and the translational velocity, ,
propeller along with respect to the body-fixed frame, expressed in the rotor-motor frame, , are exemplified for the
, expressed in the rotor frame, . second motor rotor frame, , in Figure 8.

For example, the angular velocity of the propeller with respect to

the body-fixed frame, , in (23) means that there are rotations
along . The required in
Equation (21)-(22) is the part of the angular velocity along
of (Equation (24)).

(23)

(24)

can be calculated similarly with Equation (8). However, the

first term in Equation (8) vanishes in since is with respect
to the body-fixed frame. The result is in Equation (25).
Fig. 8. The angular velocity and the translational velocity in the rotor
frame

(25)

IV. Symmetric Open-loop Hovering

The closed-loop controller design is beyond the scope of this

Notice that there are rotations along and . research. We only analyze the symmetric open-loop hovering test in
Proposition 5 believes that only the angular velocity along this study. We restrict our study in the case with the fixed direction of
contributes to the thrust and drag moment. The term is in the cone-motors .
Equation (26).
Our goal is to find the relevant angular velocities of the rotor-
(26) motors ( ) for different designed cone-angle, .

The parameters used in this simulation are given in Figure 9.

It can be seen from Equation (26), not only the angular velocity of
the rotor-motor, , which directly drives the propeller but also
angular velocity of the cone-motor, , which changes the direction
of the thrust contribute to . When all four rotor-motors suffer length of the arm 0.1785 meter
from stuck (complete loss in thrusts. ), the d 0.01 meter
components of is still able to provide thrusts to some extent,
g 9.8 N/kg
making possibility of the fault-tolerant control in this severe case.
total mass 0.429 kg
Besides the angular velocity, , another quantity worth
IB diag(2.238E-3, 2.985E-3, 4.804E-3) kgm2
investigation is the translational velocity. The translational velocity of
the geometric center of the propeller with respect to the body-fixed IC i diag(1E-10, 1E-10, 2.030E-5) kgm2
frame, , expressed in the rotor frame, , is Equation (27).
IR i diag(1E-10, 1E-10, 2.030E-5) kgm2

km i 2.423E-7 Nm
(27)
kf i 8.048E-6 N

Fig. 9. The parameters in the Quad-cone-rotor.

in Equation (27) is the length between the origin of and the
origin of . We ignore this term in our analysis since it can be
ignored if is designed small.
 Equation (28)-(29) elucidate relationship between the angular in a quad-cone-rotor while all the rotors suffer from a total loss in
velocities of the rotor-motors, , in a symmetric thrust.
open-loop hovering.
In this study, we analyze the fault-tolerant ability in fulfilling an
open-loop hovering when all the speed-motors are unhealthy
(28) (complete loss in thrust, ). The quad-cone-
rotor generates the thrusts totally relying on the angular velocities of
(29) the cone-motors ( ).

Likewise, the detail of closed-loop fault-tolerant control is one of

The result is plotted in Figure 10. The cone-angle varies from 0 the further studies and is not considered in this research.
radius to radius (asymptote). The blue curve and the red curve are
Proposition 6: (cone motor and speed motor) Cone motor has two
the angular velocity in Equation (28) and (29), respectively. The red modes. One is working as a servomotor; the other mode is the speed
stars and the blue stars are the situations verified in the quad-cone- motor. In fault-tolerant control, the cone motors are working as
rotor simulator; the zero vertical acceleration is received for these speed motors. The failed ( ) rotor motors are fixed at the
angular velocities with the corresponding cone-angles. positions generating the largest thrust.

Equation (30)-(31) illustrate the relationships of the angular

velocities of the cone motors and the rotor motors, respectively, a
complete-loss-in-all-thrust fault.

(30)
angular velocity [rad/s]

(31)

Proposition 7: (cone motor and speed motor) The angular velocity

of the cone motor, , results from Equation (32)-(33).

(32)

Fig. 10. Angular velocities in symmetric open-loop hovering.

(33)

Proof: see Appendix

V. Fault-tolerant Open-loop Hovering Instead of solving Equation (32)-(33), we find in the quad-cone-
rotor simulator to make it near hovering for different cone-angle, .
The result is plotted in Figure 11.
Fault diagnosis and fault-tolerant control received significant
interests in quadrotor and tilt quadrotor control. [25] analyzes the
case where two rotors suffer from complete thrusts loss and provides
fault-tolerant control methods with a free yaw in a conventional UAV.
To solve the uncontrolled yaw in [25], [11] proposed a tilt structure
providing the possibility of yaw control.
angular velocity [rad/s]

However, there lacks the study for the case where all the thrusts
suffer from complete loss. The reason is that controlling a
conventional quadrotor or a quad-cone-rotor suffering from such a
severe fault is impossible. We need the admissible thrusts [26,27] to
realize the attitude and the altitude control; zero thrust in all rotors is
not within the admissible region.

However, the situation is different in our quad-cone-rotor. It can

be seen from Equation (25)-(26), even if all the rotor-motors suffer
from complete thrust loss ( ), the actual thrust
can be non-zero if we assign the properly with a non-zero cone- Fig. 11. Angular velocities in fault-tolerant open-loop hovering
angle ( ). This indicates that a fault-tolerant control is possible
 It is worth mentioning that an intensive oscillation is observed in
the vertical acceleration for any cone-angle, . Figure 12a is the

vertical acceleration for the case . It is reasonable since that

square of magnitude [rad2/s 2]

the projection of the thrust varies, inducing the changing
vertical acceleration, (Equation (32)-(33)). Consequently, the
oscillation in the vertical acceleration occurs.

The frequency in that oscillation is identical to the frequency of

the angular velocity of the cone motor, . This is verified by the
power spectral density analysis (Figure 12b and Figure 13).

Fig. 13. The power of the coupled oscillation in different angular

velocities.

VI. Cone-angle Trade-off

The cone-angle in the quad-cone-rotor, , is one of the most

important parameters. It determines the range that the direction of
the thrust can be selected. A naïve thought is to design it as large as
possible. However, a larger than generates no positive

vertical thrust when for some angles in the cone-motor. In this

consideration, we restrict to be no more than .
(a)
Another fact that should pay attention to is the centripetal force.
Before calculating the centripetal force of a motor while working in a
fault-tolerant control mode mentioned in the Section VI (Equation
(30)-(31)), we make some reasonable approximation to the direction
of the thrust to simplify our trade-off analysis.
Assume that the thrust is in the direction pointing from the apex
of the cone to the center of the circle of the cone. Notice that this is
the approximation only in this trade-off analysis.
Then, the resulting centripetal force, , for fault-tolerate
control (hovering) based on this approximation is Equation (34).
(34)

is a monotonically increasing function within the interval

, this indicates that the centripetal force increases while

picking a larger cone-angle.

Another parameter affects the centripetal force is . It is the

(b) length between the origin of and the origin of . A smaller

not only produces less in Equation (34) but also gives less
Fig. 12. The oscillation in acceleration ( ) translational velocity in (27). Thus, a small is encouraged.
To balance the conflict in larger range in direction and the less
centripetal force, we formulate the following multiple-objective
optimization problem in Equation (35).
We calculate the square of the magnitude (power) of the coupled
oscillation for each corresponding cone angle, , in each fault-
tolerant open-loop hovering test. The in Figure 13
(35)
represents the power. The represents the angular velocity
for the corresponding cone angle.

The way we solve this multiple-objective optimization problem is

the weighted sum method [28], transferring Equation (35) into a changing direction of the propeller generates different thrust in each
single-objective optimization problem in Equation (36). posture. That is also part of the reason why we make Proposition 6.

Although some pairs generate thrust, modeling it using a precise

mathematical way can be challenged. To better design our controller,
(36) we make the following restriction to our inputs in Equation (37).

(37)

The result is plotted in Figure 14. It is a Pareto frontier of the trade-

off between the range and the centripetal force. Since the When the quad-cone-rotor is working in the normal mode, the
represents the negative of the range, the less (more negative) the is much larger than . This restriction assigns to be the angular
better. The represents the centripetal force, the less the
velocity mainly for generating the magnitude of the thrust. While
better. However, decreasing the centripetal force increases the
serves as a direction adjuster for the thrust. The other admissible
negative range, causing the range loss for the direction of the
working mode is usually applicable for a fault-tolerant control. The
designed thrust.
only angular velocity allowed is . While is prohibited for
introducing the difficulty in analysis.

The relevant admissible region is painted in blue in Figure 15. It

consists of the triangular part in the first and second quadrants and
the positive direction of the .
centripetal force (F)

Our further research will base on this admissible region to design

the relevant controllers.

Fig. 14. The Pareto frontier of the trade-off between the range and
the centripetal force.

Fig. 15. The admissible input region of the quad-cone-rotor.

VII. Conclusion and Discussion

The quad-cone-rotor realizes the open-loop hovering in a healthy

state. Moreover, it shows unique potential in fault-tolerant control
when a severe fault happens (complete loss in all thrusts). The
oscillation is observed while taking this special fault-tolerant control.
The frequency of this oscillation is identical to the frequency of the
angular velocity.

Our ongoing further works are closed-loop control and fault-

Appendix: Proof of Proposition 7
tolerant control. It might be tempting to control the translational
movement by setting . In this setting, the To prove Proposition 7, we first prove the following lemma.
quad-cone-rotor can balance the altitude while receiving a level force,
Lemma: The included angle ( ) between the direction of the
making it move like an autonomous car. However, the drag moment
cannot be balanced, making it rotate. thrust and the in the rotor frame is Equation (38).

Another special restriction in quad-cone-rotor is the admissible (38)

analyzing control input. Judged from Equation (25)-(26), it seems that
and have the same effect affecting the angular
Proof of Lemma:
velocity generating the thrust. However, pairing and in some
relationships are prohibited. When is not very large, the rotating The relationship of each quantity is plotted in Figure 16.

propeller can be a burden to generate a model of the thrust; the

 [2] Nemati, A., 2016. Designing, modeling and control of a tilting rotor
quadcopter (Doctoral dissertation, University of Cincinnati).

[3] Kumar, R., 2017. Position, Attitude, and Fault-Tolerant Control of

Tilting-Rotor Quadcopter (Doctoral dissertation, University of
Cincinnati).

[4] Kumar, R., Sridhar, S., Cazaurang, F., Cohen, K. and Kumar, M.,
2018, September. Reconfigurable fault-tolerant tilt-rotor quadcopter
system. In Dynamic Systems and Control Conference (Vol. 51913, p.
V003T37A008). American Society of Mechanical Engineers.

[5] Mellinger, D., Michael, N. and Kumar, V., 2012. Trajectory

generation and control for precise aggressive maneuvers with
Fig. 16. The relationship between , , .
quadrotors. The International Journal of Robotics Research, 31(5),
pp.664-674.

We calculate and in Equation (39) based on the [6] Lee, T., Leok, M. and McClamroch, N.H., 2010, December.
Geometric tracking control of a quadrotor UAV on SE (3). In 49th IEEE
geometric relationship in a cone.
conference on decision and control (CDC) (pp. 5420-5425). IEEE.
(39)
[7] Lee, H., Kim, S., Ryan, T. and Kim, H.J., 2013, October. Backstepping
In , we have Equation (40). control on se (3) of a micro quadrotor for stable trajectory tracking.
In 2013 IEEE international conference on systems, man, and
cybernetics (pp. 4522-4527). IEEE.
(40)
[8] Fernando, T., Chandiramani, J., Lee, T. and Gutierrez, H., 2011,
In , we have Equation (41), law of cosines. December. Robust adaptive geometric tracking controls on SO (3)
with an application to the attitude dynamics of a quadrotor UAV.
(41) In 2011 50th IEEE conference on decision and control and European
control conference (pp. 7380-7385). IEEE.

Substitute Equation (39)-(40) into Equation (41), we receive [9] Ryll, M., Bülthoff, H.H. and Giordano, P.R., 2012, May. Modeling
Equation (38). The proof to lemma is complete. and control of a quadrotor UAV with tilting propellers. In 2012 IEEE
International Conference on Robotics and Automation (pp. 4606-
Each rotor in the symmetric hovering analysis satisfies Equation
4613). IEEE.
(42), Newtown’s second law.
[10] Şenkul, F. and Altuğ, E., 2013, May. Modeling and control of a
(42) novel tilt—Roll rotor quadrotor UAV. In 2013 International
Conference on Unmanned Aircraft Systems (ICUAS) (pp. 1071-1076).
Substituting Equation (21), (26), (38) into Equation (42) yields IEEE.
Equation (43). [11] Yang, D., Li, Z., Zhou, P. and Lu, J., 2020, November. Control
System Design for Tiltable Quad-rotor with Propeller Failure. In 2020
Chinese Automation Congress (CAC) (pp. 7473-7478). IEEE.

(43) [12] Giribet, J.I., Pose, C.D., Ghersin, A.S. and Mas, I., 2018.
Experimental validation of a fault-tolerant hexacopter with tilted
Integrating the right side of Equation (43) yields the velocity- rotors.
related term, Equation (32).
[13] Ryll, M., Bülthoff, H.H. and Giordano, P.R., 2015. A Novel
Integrating Equation (32) again yields the altitude-related term, Overactuated Quadrotor UAV: Modelling, Control and Experimental
Equation (33). The expected altitude change in a period, , is zero. Validation. IEEE Transactions on Control Systems Technology,
Institute of Electrical and Electronics Engineers, 23(2), pp.510-556.
So far, the Proposition 7 is proved.
[14] Nemati, A., Kumar, R. and Kumar, M., 2016, October. Stabilizing
and control of tilting-rotor quadcopter in case of a propeller failure.
In Dynamic Systems and Control Conference (Vol. 50695, p.
REFERENCES V001T05A005). American Society of Mechanical Engineers.
[1] Raffo, G.V., Ortega, M.G. and Rubio, F.R., 2010. An integral [15] Caliskan, F. and Hajiyev, C., 2016. Active fault-tolerant control of
predictive/nonlinear H∞ control structure for a quadrotor helicopter. UAV dynamics against sensor-actuator failures. Journal of Aerospace
Automatica, 46(1), pp.29-39. Engineering, 29(4), p.04016012.
[16] Sadeghzadeh, I., Mehta, A., Chamseddine, A. and Zhang, Y., 2012,
April. Active fault tolerant control of a quadrotor uav based on
gainscheduled pid control. In 2012 25th IEEE Canadian conference on
electrical and computer engineering (CCECE) (pp. 1-4). IEEE.

[17] Caliskan, F. and Hacizade, C., 2014. Sensor and actuator FDI
applied to an UAV dynamic model. IFAC Proceedings Volumes, 47(3),
pp.12220-12225.

[18] Nguyen, N.P. and Hong, S.K., 2018. Sliding mode thau observer
for actuator fault diagnosis of quadcopter UAVs. Applied
Sciences, 8(10), p.1893.

[19] Al-Ali, I., Zweiri, Y., AMoosa, N., Taha, T., Dias, J. and Senevirtane,
L., 2020. State of the art in tilt-quadrotors, modelling, control and
fault recovery. Proceedings of the Institution of Mechanical Engineers,
Part C: Journal of Mechanical Engineering Science, 234(2), pp.474-486.

[20] Huang, H., Hoffmann, G.M., Waslander, S.L. and Tomlin, C.J.,
2009, May. Aerodynamics and control of autonomous quadrotor
helicopters in aggressive maneuvering. In 2009 IEEE international
conference on robotics and automation (pp. 3277-3282). IEEE.

[21] Craig, W., Yeo, D. and Paley, D.A., 2020. Geometric attitude and
position control of a quadrotor in wind. Journal of Guidance, Control,
and Dynamics, 43(5), pp.870-883.

[22] Bouabdallah, S., Noth, A. and Siegwart, R., 2004, September. PID
vs LQ control techniques applied to an indoor micro quadrotor.
In 2004 IEEE/RSJ International Conference on Intelligent Robots and
Systems (IROS)(IEEE Cat. No. 04CH37566) (Vol. 3, pp. 2451-2456).
IEEE.

[23] Hasan, A. and Johansen, T.A., 2018, June. Model-based actuator

fault diagnosis in multirotor UAVs. In 2018 International Conference
on Unmanned Aircraft Systems (ICUAS) (pp. 1017-1024). IEEE.

[24] Prach, A. and Kayacan, E., 2018. An MPC-based position

controller for a tilt-rotor tricopter VTOL UAV. Optimal Control
Applications and Methods, 39(1), pp.343-356.

[25] Michieletto, G., Ryll, M. and Franchi, A., 2018. Fundamental

actuation properties of multirotors: Force–moment decoupling and
fail–safe robustness. IEEE Transactions on Robotics, 34(3), pp.702-
715.

[26] Giribet, J.I., Sanchez-Pena, R.S. and Ghersin, A.S., 2016. Analysis
and design of a tilted rotor hexacopter for fault tolerance. IEEE
Transactions on Aerospace and Electronic Systems, 52(4), pp.1555-
1567.

[27] Ducard, G. and Hua, M.D., 2011. Discussion and practical aspects
on control allocation for a multi-rotor helicopter. International
Archives of the Photogrammetry, Remote Sensing and Spatial
Information Sciences, 38(1/C22), pp.95-100.

[28] Marler, R.T. and Arora, J.S., 2004. Survey of multi-objective

optimization methods for engineering. Structural and
multidisciplinary optimization, 26(6), pp.369-395.