Professional Documents
Culture Documents
José Agnelo Bezerra, João Francisco Silva Trentin, and Davi A. Santos
Abstract. This work is concerned with the attitude and position con-
trol of a fully actuated non-planar hexarotor aerial vehicle equipped with
reversible fixed rotors. The complete nonlinear dynamics of the vehicle is
modeled in a state error formulation with six degrees of freedom (DOF),
being three for position and three others for attitude, while the control
input is also a six-DOF quantity defined in terms of the resultant force
and torque acting on the system. A control law is designed using an
unit-vector global sliding mode control strategy, which ensures robust-
ness against bounded force and torque disturbances during the entire
flight since the sliding condition is guaranteed from the initial time. Fur-
thermore, the proposed controller also ensures global exponential stabil-
ity for both the closed-loop translational and rotational dynamics. Using
computational simulations, the designed control law is compared with
an inverse-dynamic stabilizing control, showing to be effective and to
perform much better.
1 Introduction
Urban applications of multirotor aerial vehicles (MAVs), such as in aerial ma-
nipulation, delivery, and air transport, are expected in the near future. These
tasks commonly require the vehicle to safely fly in a three-dimensional space
and, when demanded, to adopt an arbitrary body orientation during the flight,
while subject to unknown environmental disturbances (e.g., wind). Therefore,
such applications are quite suitable for fully actuated MAVs, which are those ve-
hicles capable of independently controlling their position and attitude dynamics.
This class of MAVs has attracted an increasing interest in the literature, thus
many different vehicle configurations and control designs have been proposed
in the last decade [1–11]. In this work, a fully actuated non-planar hexarotor
configuration is adopted.
Since the MAV applications commonly are subject to environmental dis-
turbances, they demand the use of a robust flight controller. Moreover, it is
well-known that the position and attitude dynamics of such vehicles are, in gen-
eral, coupled and nonlinear. Therefore, the sliding mode control (SMC) [12, 13]
2 Bezerra, Trentin and Santos
is quite suitable for this case, since it can ally robustness against bounded dis-
turbances with exponential stability for nonlinear systems using the linear first
order sliding manifold.
The conventional SMC design consists of two phases: reaching a specified slid-
ing manifold and sliding along this manifold [12]. However, during the reaching
phase, robustness is not guaranteed. Therefore, we use the global sliding mode
control (GSMC) scheme, which provides robustness from the initial condition
until the desired reference [12, 14]. Thus, the main contribution of this chapter
is to derive a GSMC control law in order to control both the MAV’s position
and attitude dynamics. The controller provides a six-dimensional command for
the resultant force and torque and it ensures robustness for all time.
This chapter is organized as follows. Section 2 presents the mathematical
modeling of the fully actuated hexarotor and formulates the control problem of
this work. Section 3 designs the global sliding mode controller and shows a stabil-
ity analysis. Section 4 reports the comparative simulation results of the proposed
control. Finally, Section 5 covers the concluding remarks of this chapter.
2 Problem Definition
This section is organized as follows: Subsection 2.1 introduces the notation em-
ployed in this work; Subsection 2.2 presents the rotor-set modeling of the adopted
MAV; and Subsection 2.3 is concerned with the vehicle’s dynamic modeling and
the control problem formulation.
2.1 Notation
Consider the hexarotor and the Cartesian Coordinate Systems (CCS) illustrated
in Figure 1. The ground CCS Sg ≜ {G; x̂g , ŷg , ẑg } is an inertial system fixed at a
known point G, while the body CCS Sb ≜ {B; x̂b , ŷb , ẑb } is fixed to the airframe,
with the origin at its center of mass B. The MAV has six fixed rotors with the
following positioning: rotors 1 and 2 are parallel to ẑb , while rotors 3 and 4 are
aligned with x̂b , and rotors 5 and 6 are parallel to ŷb . Therefore, the thrust force
and reaction torque that are produced by each rotor can be described by
where ⃗ei represents the i-th rotor direction, i.e., ⃗e1 = ⃗e2 = ẑb , ⃗e3 = ⃗e4 = x̂b , and
⃗e5 = ⃗e6 = ŷb . Moreover, fi , τi ∈ R represent, respectively, the force and torque
magnitudes and are described by the aerodynamic models [4, 16]
fi = kf ωi |ωi | , (3)
i+1
τi = (−1) kτ ωi |ωi | , (4)
6
X
f⃗ c = f⃗i , (5)
i=1
6
X
⃗τ c = ⃗ℓi × f⃗i + ⃗τi , (6)
i=1
where ⃗ℓi is the arm vector of the i-th rotor w.r.t. B and is such that ∥⃗ℓi ∥ = ℓ,
∀i ∈ {1, 2, . . . , 6}. Considering the modeling in (3)–(4), the Sb representations of
these efforts can be expressed as
4 Bezerra, Trentin and Santos
𝑓⃗5
𝐵 𝑓⃗1 𝜏⃗1
𝑓⃗3 +
𝜏⃗3 +
𝑓⃗6
𝒮𝑔 𝜏⃗6
Fig. 1. Schematic representation of the considered fully actuated hexarotor. The black
and blue arrows indicate the directions of the respective quantities.
nr
X
fbc = fi ei,b , (7)
i=1
Xnr
τbc = fi [ℓi,b ×] + k(−1)i+1 I3 ei,b ,
(8)
i=1
u = Γ f̄, (9)
which is known as the control allocation equation. For the adopted hexarotor,
the matrix Γ ∈ R6×6 is given by
0 0 1 1 0 0
0 0 0 0 1 1
1 1 0 0 0 0
Γ ≜ . (10)
0 0
k −k −ℓ ℓ
−ℓ ℓ 0 0 k −k
k −k ℓ −ℓ 0 0
GSMC for a Fully Actuated Non-Planar Hexarotor Aerial Vehicle 5
One can note that rank(Γ ) = 6, thus the considered MAV is fully actuated
[16]. Therefore, Γ is non-singular and, for a given control command u, the thrust
commands in f̄ can be calculated as
f̄ = Γ −1 u, (11)
and, according to the model (3), the final rotor commands are given by
s
|f¯i |
sign f¯i ,
ω̄i = (12)
kf
∀i ∈ {1, 2, . . . , 6}. For a more complete explanation about the control allocation
problem for fully actuated multirotor aerial vehicles, the reader can read the
reference [17].
ṙb/g
g = vb/g
g , (13)
1 b/g T c
v̇gb/g = D fb + fbd + −ge3 , (14)
mh
b/g
i
b/g
Ḋ = − ωb × Db/g , (15)
h i
b/g b/g b/g
= J−1 τbc + τbd − J−1
ω̇b b b ωb × Jb ωb , (16)
∥fbd ∥ ≤ ρf , (17)
∥τbd ∥ ≤ ρτ , (18)
1 − ∥σ∥2 I3 + 2σ(σ)T − 2[σ×]
D(σ) ≜ . (19)
1 + ∥σ∥2
Therefore, from (15), one can derive the Gibbs vector kinematics
b/g
σ̇ b/g = M(σ b/g )ωb , (20)
with
1
σσ T + [σ×] + I3 .
M(σ) ≜ (21)
2
Now, assume Sb̄ ≜ {B; x̂b̄ , ŷb̄ , ẑb̄ } as a CCS that represents the desired orien-
tation for the Sb axes and is fixed to the vehicle’s center of mass B. Based on
b/g b/g
this system, consider that r̄g ∈ R3 and v̄g ∈ R3 are commands for position
b̄/g
and velocity in Sg , while σ b̄/g ∈ R3 and ωb̄ ∈ R3 represent the Gibbs vector
and angular velocity of Sb̄ w.r.t. Sg .
Using the previous commands, the error quantities are defined
r̃ ≜ rb/g
g − r̄b/g
g , (22)
D̃ ≜ Db/g Dg/b̄ , (23)
ṽ ≜ vb/g
g − v̄b/g
g , (24)
b/g b̄/g
ω̃ ≜ ωb − ωb . (25)
The attitude error in (23) can also be expressed in terms of Gibbs vector, as
follows
where
GSMC for a Fully Actuated Non-Planar Hexarotor Aerial Vehicle 7
ṽ
f1 (x1 , x2 ) ≜ ,
M(σ̃)ω̃
b/g
" #
−v̄˙ g − ge3h
f2 (x1 , x2 ) ≜ b̄/g b̄/g b/g
i
b/g
,
−D̃ω̇b̄ + [ω̃×] D̃ωb̄ − J−1
b ωb × Jb ωb
" T #
1 b/g
D 0 3×3
B(x1 ) ≜ m .
03×3 J−1b
Given the definition of the disturbance vector d, one can obtain an upper
bound for this vector’s magnitude by using the limits in (17)–(18), as follows
∥d∥ ≤ ρ, (29)
where ρ ≜ ρf + ρτ .
Finally, the control problem of this work can be enunciated.
Problem 1. The MAV control problem is to design a control law u(x) that makes
x = 012 a globally exponentially stable equilibrium point, while keeping a sliding
motion for all time.
3 Controller Design
This section derives a global sliding mode control strategy for the complete dy-
namics of the fully actuated non-planar hexarotor. Thus, consider the following
sliding variable
S ≜ x ∈ R12 : s(t, x) = 0, ∀t ≥ 0 .
(33)
Therefore, to ensure a global sliding mode of system (27)–(28) in S, the following
unit-vector force-torque control law can be designed
8 Bezerra, Trentin and Santos
−1
∂f1 ∂f1 ∂f1 s
u=− B C+ f1 + f2 − Ṗ(t)σ(0) + κ , (34)
∂x2 ∂x1 ∂x2 ∥s∥
where κ ∈ R+ is the switching gain.
Theorem 1. The control law u given in (34) ensures that x = 012 is a globally
exponentially stable equilibrium point of the MAV’s complete error dynamics
(27)–(28), and guarantees the global sliding mode s(t, x) = 06 , ∀t ≥ 0.
V̇ = sT ṡ. (36)
Now, calculating the time derivative of the sliding variable in (30), it results
∂f1 ∂f1
ṡ = Cf1 + f1 + (f2 + Bu + Bd) − Ṗσ(0). (37)
∂x1 ∂x2
By substituting the proposed control law (34) into (37), one can obtain the
closed-loop s-dynamics
∂f1 s
ṡ = Bd − κ . (38)
∂x2 ∥s∥
The previous equation can then be substituted into (36), which yields
T ∂f1 s
V̇ = s Bd − κ , (39)
∂x2 ∥s∥
thus the following inequality holds
∂f1
V̇ ≤ − κ − ρ ∥B∥ ∥s∥ , (40)
∂x2
∂f1
κ>ρ ∥B∥ , (41)
∂x2
then (40) represents a sufficient condition for the existence of the sliding modes.
Finally, let us analyse the proposed sliding variable (30) at the initial time
Note that, at the initial time, the function P(t) is given by P(0) = I6 , which
demonstrates that the sliding motion is guaranteed from the initial time. Fur-
thermore, from (30), s = 06 implies that x1 and ẋ1 converge exponentially to
the equilibrium point. From (27), one can note that x2 also converges exponen-
tially to the origin. Therefore, the proposed control law ensures that x = 012 is a
globally exponentially stable equilibrium point of (27)–(28), which completes the
proof. □
4 Numerical Results
This section reports the numerical results obtained for the proposed GSMC law
and compares them to a feedback linearization PD control. The script was coded
in MATLAB using Euler’s method for integration with time step of 0.01 s. The
physical parameters adopted for the fully actuated hexarotor are: J = 0.01I3
kgm2 , m = 0.5 kg, g = 9.81 m/s2 , ℓ = 0.25 m, kf = 2.5 × 10−5 Ns2 , and kτ =
5 × 10−7 Nms2 .
The reference trajectory is a step of (1, 1.5, 2) m for position, combined with
a conic motion with frequency of 1/15 Hz and amplitude of 30◦ for attitude.
The disturbances are considered as sinusoidal signals with frequency of 0.1 Hz
and amplitudes of 0.04 N and 0.003 Nm for force and torque, respectively. The
parameters used for the GSMC are κ = 0.5, C = 0.6I6 , and P(t) = exp(−t)I6 .
On the other hand, for the PD controller, K1 = diag(2, 2, 2, 3, 3, 3) is the
proportional gain and K2 = 5I6 is the derivative gain.
The controlled position and velocity of the MAV are showed in Figure 2,
where one can observe that the proposed GSMC reaches the reference trajectory
first than the PD control and presenting no overshoot. One can also note that,
due to the disturbances applied in the simulations, the PD control presents some
oscillations around the desired reference.
The attitude obtained in the Gibbs vector is parameterized in terms of Euler
angles using a 1-2-3 (ϕ, θ, ψ) sequence for better visualization of the attitude mo-
tion. Thus, the controlled attitude and angular velocities of the MAV are showed
in Figure 3, where a significant difference between the control approaches can
be noted. Since the attitude dynamics is more sensible to disturbances than the
position one, the proposed robust approach has presented a better performance
when compared to the PD control. Although the plots of attitude and angular
velocity are charted side by side, it is important to remember that the time
derivative of the attitude does not result in the angular velocity.
The proposed GSMC approach provides a compound force-torque command.
Thereby, the control forces are illustrated in Figure 4 (a) and one can see that
the two controllers provide forces with very similar magnitude. However, the
magnitudes of the control torques presented in Figure 4 (b) are slightly larger
for the GSMC approach.
10 Bezerra, Trentin and Santos
1.2 0.4
1
0.3
0.8
0.2
v1 (m/s)
r1 (m)
0.6
0.1
0.4
0
0.2
0 -0.1
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
b/g b/g
(a) rg1 (b) vg1
2 0.6
0.5
1.5
0.4
v2 (m/s)
0.3
r2 (m)
1
0.2
0.1
0.5
0
0 -0.1
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
b/g b/g
(c) rg2 (d) vg2
2.5 0.8
2 0.6
1.5 0.4
v3 (m/s)
r3 (m)
1 0.2
0.5 0
0 -0.2
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
b/g b/g
(e) rg3 (f) vg3
60 0.4
0.3
40
0.2
ω1 (rad/s)
20
0.1
φ(t)(◦ )
0 0
-0.1
-20
-0.2
-40 -0.3
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
b/g
(a) ϕ(t) (b) ωb1
40 0.3
0.2
20
0.1
ω2 (rad/s)
θ(t)(◦ )
0 0
-0.1
-20
-0.2
-40 -0.3
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
b/g
(c) θ(t) (d) ωb2
40 0.2
0.1
20
0
ω3 (rad/s)
ψ(t)(◦ )
0 -0.1
-0.2
-20
-0.3
-40 -0.4
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
b/g
(e) ψ(t) (f) ωb3
Fig. 3. Controlled attitude in Euler angles and controlled angular velocity. Line – - –
indicates PD control, — indicates proposed GSMC and - - - indicates the command.
12 Bezerra, Trentin and Santos
5
0.02
(N)
(Nm)
0
0
fb,1
c
τb,1
-5
c
0 10 20 30 40 50 -0.02
0 10 20 30 40 50
5 0.02
(N)
(Nm)
0 0
fb,2
c
τb,2
c
-5 -0.02
0 10 20 30 40 50 0 10 20 30 40 50
10 0.02
(Nm)
(N)
5 0
fb,3
c
τb,3
c
0 -0.02
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
(a) Control force fcb (b) Control torque τbc
Fig. 4. Force and torque commands. Line – - – indicates PD control and — indicates
proposed GSMC.
0.2 0.2
0 0
s1
s4
-0.2 -0.2
0 10 20 30 40 50 0 10 20 30 40 50
0.2 0.2
0 0
s2
s5
-0.2 -0.2
0 10 20 30 40 50 0 10 20 30 40 50
0.2 0.2
0 0
s3
s6
-0.2 -0.2
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
(a) Position sliding variable (b) Attitude sliding variable
The main idea of designing a GSMC law for the fully actuated non-planar
hexarotor is to ensure robustness during all flight time and, for this purpose,
the sliding variable must be zero. In this condition, the GSMC can guarantee
robustness against bounded disturbances and unmodeled dynamics. Thus, Figure
5 exhibits the sliding variable which was separated for position and attitude only
for better visualization, where one can see that the sliding variable stays around
zero for all the simulation time.
5 Concluding Remarks
This work has presented the detailed dynamic modeling of a fully actuated non-
planar hexarotor aerial vehicle and designed a global sliding mode control law
suitable for this category of vehicles. The results for the proposed approach have
been compared to a proportional-derivative feedback linearization control law.
We can observe that the GSMC outperforms the PD control, while ensuring
robustness against bounded disturbances during all the flight.
Acknowledgements
The authors thank the São Paulo Research Foundation (FAPESP) for the finan-
cial support (grants 2019/05334-0 and 2020/12314-3). The third author is also
grateful for the support of CNPq/Brazil (grant 304300/2021-7).
References
1. Crowther, B., Lanzon, A., Maya-Gonzalez, M., Langkamp, D.: Kinematic analysis
and control design for a nonplanar multirotor vehicle. Journal of Guidance, Control,
and Dynamics 34(4), 1157–1171 (2011)
2. Rajappa, S., Ryll, M., Bülthoff, H.H., Franchi, A.: Modeling, control and design
optimization for a fully-actuated hexarotor aerial vehicle with tilted propellers. In:
2015 IEEE International Conference on Robotics and Automation (ICRA). pp.
4006–4013. Seattle (2015)
3. Park, S., Her, J., Kim, J., Lee, D.: Design, modeling and control of omni-directional
aerial robot. In: 2016 IEEE/RSJ International Conference on Intelligent Robots
and Systems (IROS). pp. 1570–1575. Daejeon (2016)
4. Brescianini, D., D’Andrea, R.: Design, modeling and control of an omni-directional
aerial vehicle. In: 2016 IEEE International Conference on Robotics and Automation
(ICRA). pp. 3261–3266 (2016)
5. Kamel, M., Verling, S., Elkhatib, O., Sprecher, C., Wulkop, P., Taylor, Z., Sieg-
wart, R., Gilitschenski, I.: The Voliro omniorientational hexacopter: An agile and
maneuverable tiltable-rotor aerial vehicle. IEEE Robotics Automation Magazine
25(4), 34–44 (2018)
6. Tognon, M., Franchi, A.: Omnidirectional aerial vehicles with unidirectional
thrusters: Theory, optimal design, and control. IEEE Robotics and Automation
Letters 3(3), 2277–2282 (2018)
14 Bezerra, Trentin and Santos
7. Dyer, E., Sirouspour, S., Jafarinasab, M.: Energy optimal control allocation in a
redundantly actuated omnidirectional UAV. In: 2019 International Conference on
Robotics and Automation (ICRA). pp. 5316–5322. Montreal (2019)
8. Rashad, R., Goerres, J., Aarts, R., Engelen, J.B.C., Stramigioli, S.: Fully actuated
multirotor UAVs: A literature review. IEEE Robotics Automation Magazine 27(3),
97–107 (2020)
9. Hamandi, M., Usai, F., Sablé, Q., Staub, N., Tognon, M., Franchi, A.: Design of
multirotor aerial vehicles: A taxonomy based on input allocation. The International
Journal of Robotics Research 40(8-9), 1015–1044 (2021)
10. Ryll, M., Bicego, D., Giurato, M., Lovera, M., Franchi, A.: FAST-Hex – A mor-
phing hexarotor: Design, mechanical implementation, control and experimental
validation. IEEE/ASME Transactions on Mechatronics (2021)
11. Bezerra, J.A., Santos, D.A.: On the guidance of fully-actuated multirotor aerial
vehicles under control allocation constraints using the receding-horizon strategy.
ISA Transactions (2021)
12. Shtessel, Y., Edwards, C., Fridman, L., Levant, A.: Sliding mode control and ob-
servation, vol. 10. Springer (2014)
13. Utkin, V., Poznyak, A., Orlov, Y.V., Polyakov, A.: Road map for sliding mode
control design. Springer (2020)
14. Bartoszewicz, A.: Time-varying sliding modes for second-order systems. IEE Pro-
ceedings - Control Theory and Applications 143, 455–462(7) (September 1996)
15. Shuster, M.D.: A survey on attitude representations. Journal of the Astronautical
Sciences 41(4), 439–517 (1993)
16. Michieletto, G., Ryll, M., Franchi, A.: Fundamental actuation properties of multi-
rotors: Force–moment decoupling and fail–safe robustness. IEEE Transactions on
Robotics 34(3), 702–715 (2018)
17. Santos, D.A., Bezerra, J.A.: On the control allocation of fully actuated multirotor
aerial vehicles. Aerospace Science and Technology 122, 107424 (2022), https://
www.sciencedirect.com/science/article/pii/S1270963822000980
18. Silva, A.L., Santos, D.A.: Fast nonsingular terminal sliding mode flight control for
multirotor aerial vehicles. IEEE Transactions on Aerospace and Electronic Systems
56(6), 4288–4299 (2020)