Control of unmanned vehicles

20 de septiembre de 2024

Control of unmanned vehicles 20 de septiembre de 2024 1 / 46

Content

1 Modeling the UAV: basic concepts

2 Orientation and position of the aircraft

3 Newton-Euler model

4 Euler-Lagrange model

Control of unmanned vehicles 20 de septiembre de 2024 2 / 46

Modeling the UAV: basic concepts

Quadrotor
a cross configuration

Control of unmanned vehicles 20 de septiembre de 2024 3 / 46

Modeling the UAV: basic concepts

Quadrotor
a cross configuration
the cross structure is quite thin and light

Control of unmanned vehicles 20 de septiembre de 2024 3 / 46

Modeling the UAV: basic concepts

Quadrotor
a cross configuration
the cross structure is quite thin and light
four rotos

Control of unmanned vehicles 20 de septiembre de 2024 3 / 46

Modeling the UAV: basic concepts

Quadrotor
a cross configuration
the cross structure is quite thin and light
four rotos
all the propellers axes of rotation are fixedand paralle

Control of unmanned vehicles 20 de septiembre de 2024 3 / 46

Modeling the UAV: basic concepts

Quadrotor
a cross configuration
the cross structure is quite thin and light
four rotos
all the propellers axes of rotation are fixedand paralle
all rotors with fixed-pitch blades

Control of unmanned vehicles 20 de septiembre de 2024 3 / 46

Modeling the UAV: basic concepts

Quadrotor
a cross configuration
the cross structure is quite thin and light
four rotos
all the propellers axes of rotation are fixedand paralle
all rotors with fixed-pitch blades

Control of unmanned vehicles 20 de septiembre de 2024 3 / 46

Modeling the UAV: basic concepts
Quadrotor

Figura: Quadrotor from Mikrokopter

Control of unmanned vehicles 20 de septiembre de 2024 4 / 46

Modeling the UAV: basic concepts

Figura: Two cross configurations

Control of unmanned vehicles 20 de septiembre de 2024 5 / 46

Modeling the UAV: basic concepts

Control of unmanned vehicles 20 de septiembre de 2024 6 / 46

Modeling the UAV: basic concepts

Figura: Lift forces and torques

P4 P4
Flift = 1 fi 1 τi =0 (1)
Control of unmanned vehicles 20 de septiembre de 2024 7 / 46
Modeling the UAV: basic concepts
Yaw moment
Simultaneously, increasing the speed of two rotors that rotate clockwise and
decreasing the speed of other rotors that rotate counter-clockwise.

Control of unmanned vehicles 20 de septiembre de 2024 8 / 46

Modeling the UAV: basic concepts
Pitch moment
Simultaneously, increasing the speed of rotor 1 and decreasing the speed rotor 2
while the speed of the other two rotors does not change.

Control of unmanned vehicles 20 de septiembre de 2024 9 / 46

Modeling the UAV: basic concepts
Pitch moment
Simultaneously, increasing the speed of rotor 2 and decreasing the speed rotor 1
while the speed of the other two rotors does not change.

Control of unmanned vehicles 20 de septiembre de 2024 10 / 46

Modeling the UAV: basic concepts
Roll moment
Simultaneously, increasing the speed of rotor 2 and decreasing the speed rotor 4
while the speed of the other two rotors does not change.

Control of unmanned vehicles 20 de septiembre de 2024 11 / 46

Modeling the UAV: basic concepts
Roll moment
Simultaneously, increasing the speed of rotor 4 and decreasing the speed rotor 2
while the speed of the other two rotors does not change.

Control of unmanned vehicles 20 de septiembre de 2024 12 / 46

Orientation and position of the aircraft

Inertial frame I-frame

Body frame B-frame
The B-frame is a right-hand oriented coordinate system:
The positive x-axis direction is forward movement.
The positive y -axis direction is to right side.
The positive z-axis point down into the earth.
Three angles that are individually called roll φ, pitch θ and yaw
ψ.

Control of unmanned vehicles 20 de septiembre de 2024 13 / 46

Orientation and position of the aircraft
Rotation matrix
RI = Rz,ψ Ry ,θ Rx,φ
   
cos ψ − sin ψ 0 cos θ 0 sin θ 1 0 0
=  sin ψ cos ψ 0   0 1 0  0 cos ψ − sin ψ 
0 0 1 − sin θ 0 cos θ 0 sin ψ cos ψ
 
Cφ Sθ Cψ Sθ Sφ − Sψ Sφ Cψ Sθ Cφ + Sψ Sφ
=  Sφ Cθ Sψ Sθ Sφ + Cψ Cφ Sψ Sθ Cφ − Sφ Cψ  (2)
−Sθ Cθ Sφ Cθ Cφ

where Cδ = cos δ and Sδ = sin δ.

Control of unmanned vehicles 20 de septiembre de 2024 14 / 46

Orientation and position of the aircraft
Rotation matrix
RI = Rz,ψ Ry ,θ Rx,φ
   
cos ψ − sin ψ 0 cos θ 0 sin θ 1 0 0
=  sin ψ cos ψ 0   0 1 0  0 cos ψ − sin ψ 
0 0 1 − sin θ 0 cos θ 0 sin ψ cos ψ
 
Cφ Sθ Cψ Sθ Sφ − Sψ Sφ Cψ Sθ Cφ + Sψ Sφ
=  Sφ Cθ Sψ Sθ Sφ + Cψ Cφ Sψ Sθ Cφ − Sφ Cψ  (2)
−Sθ Cθ Sφ Cθ Cφ

where Cδ = cos δ and Sδ = sin δ.

RI ∈ SO(3)

Control of unmanned vehicles 20 de septiembre de 2024 14 / 46

Orientation and position of the aircraft
Rotation matrix
RI = Rz,ψ Ry ,θ Rx,φ
   
cos ψ − sin ψ 0 cos θ 0 sin θ 1 0 0
=  sin ψ cos ψ 0   0 1 0  0 cos ψ − sin ψ 
0 0 1 − sin θ 0 cos θ 0 sin ψ cos ψ
 
Cφ Sθ Cψ Sθ Sφ − Sψ Sφ Cψ Sθ Cφ + Sψ Sφ
=  Sφ Cθ Sψ Sθ Sφ + Cψ Cφ Sψ Sθ Cφ − Sφ Cψ  (2)
−Sθ Cθ Sφ Cθ Cφ

where Cδ = cos δ and Sδ = sin δ.

RI ∈ SO(3)
−1
RT
I = RI

Control of unmanned vehicles 20 de septiembre de 2024 14 / 46

Orientation and position of the aircraft
Rotation matrix
RI = Rz,ψ Ry ,θ Rx,φ
   
cos ψ − sin ψ 0 cos θ 0 sin θ 1 0 0
=  sin ψ cos ψ 0   0 1 0  0 cos ψ − sin ψ 
0 0 1 − sin θ 0 cos θ 0 sin ψ cos ψ
 
Cφ Sθ Cψ Sθ Sφ − Sψ Sφ Cψ Sθ Cφ + Sψ Sφ
=  Sφ Cθ Sψ Sθ Sφ + Cψ Cφ Sψ Sθ Cφ − Sφ Cψ  (2)
−Sθ Cθ Sφ Cθ Cφ

where Cδ = cos δ and Sδ = sin δ.

RI ∈ SO(3)
−1
RT
I = RI

detRI = 1

Control of unmanned vehicles 20 de septiembre de 2024 14 / 46

Orientation and position of the aircraft
Rotation matrix
RI = Rz,ψ Ry ,θ Rx,φ
   
cos ψ − sin ψ 0 cos θ 0 sin θ 1 0 0
=  sin ψ cos ψ 0   0 1 0  0 cos ψ − sin ψ 
0 0 1 − sin θ 0 cos θ 0 sin ψ cos ψ
 
Cφ Sθ Cψ Sθ Sφ − Sψ Sφ Cψ Sθ Cφ + Sψ Sφ
=  Sφ Cθ Sψ Sθ Sφ + Cψ Cφ Sψ Sθ Cφ − Sφ Cψ  (2)
−Sθ Cθ Sφ Cθ Cφ

where Cδ = cos δ and Sδ = sin δ.

RI ∈ SO(3)
−1
RT
I = RI

detRI = 1

Control of unmanned vehicles 20 de septiembre de 2024 14 / 46

Orientation and position of the aircraft
Angular rates
For the orthonormal matrix RI we have

RI RIT = I3 (3)

Differentiating (3) yields

R˙I RIT + RI ṘIT = 03

 T
R˙I RIT + R˙I RIT = 03 (4)

Defining S = R˙I RIT , equation (4) may also be written

S + S T = 03 (5)

where S is a skew-symmetric matrix.

Control of unmanned vehicles 20 de septiembre de 2024 15 / 46
Orientation and position of the aircraft
Angular rates
Hence, a property relating the derivative of orthonormal matrices
with a skew-symmetric matrices exist and can be stated as [7]:

S = R˙I RI−1 (6)

From (6), the kinematic equations to determine the quadrotor’s

posture are given by
R˙I = RI S(ω) (7)
 T
where ω = p q r is the angular velocities vector in the
B-frame and S(ω)
 
0 −r q
S(ω) =  r 0 −p  (8)
−q p 0

Control of unmanned vehicles 20 de septiembre de 2024 16 / 46

Orientation and position of the aircraft

Angular rates
From (6), we can obtain the relationship of the angular velocities
between both reference frames
    
φ̇ 1 sin φ tan θ cos φ tan θ p
 θ̇  =  0 cos φ − sin φ   q  (9)
ψ̇ 0 sin φ sec θ cos φ sec θ r
   
φ̇ p
 θ̇  = WI  q 
ψ̇ r

Control of unmanned vehicles 20 de septiembre de 2024 17 / 46

Orientation and position of the aircraft

Rotational movement
The rotational movement of the UAV is given by the components of
the angular velocities in all three axes:
roll angular velocity, p on xB axis
pitch angular velocity, q on yB axis
yaw angular velocity, r on zB axis
which are due to the forces exerted on the UAV causing the
moment of rolling (τφ )
moment of pitch (τθ )
moment of yaw (τψ )

Control of unmanned vehicles 20 de septiembre de 2024 18 / 46

Orientation and position of the aircraft

Translational movement
The translation movement is given by the components of the speed
 T
v = u0 v0 w0 on the three inertial axes in relation to the
 T
absolute speed of the UAV expressed in B-frame V = u v w
   
u0 u
 v0  = RI  v  (10)
w0 w

Control of unmanned vehicles 20 de septiembre de 2024 19 / 46

Newton-Euler model

Consider the following assumptions:

The structure is supposed rigid and symmetrical.
The B-frame origin 0B is coincident with the center of gravity
(COG) of the body.
The propellers are supposed rigid.
 T
Defining the linear and angular position in I ξ = x y z and
 T
η = φ θ ψ , and the linear and angular velocities in B
 T  T
VB = u v w and ωB = p q r
 T
ζ= ξ η
 T
ν = VB ω B

Control of unmanned vehicles 20 de septiembre de 2024 20 / 46

Newton-Euler model

The relation between VB and ξ˙ involves the rotation matrix RI

according to equation
ξ˙ = RI VB (11)
The relation between ωB and η involves the rotation matrix TI
according to equation
η̇ = WI ωB (12)
So,
ξ˙
      
RI 03×3 VB VB
= = JI (13)
η̇ 03×3 WI ωB ωB

Control of unmanned vehicles 20 de septiembre de 2024 21 / 46

Newton-Euler model

From the Euler’s first axiom of the Newton’s second law follows the
derivation of the linear components of the body motion, according to

mξ¨ = FI
d
m (RI VB ) = RI FB
 dt 
m RI V̇B + ṘI VB = RI FB
 
mRI V̇B + ωB × VB = RI FB
 
m V̇B + ωB × VB = FB (14)

Control of unmanned vehicles 20 de septiembre de 2024 22 / 46

Newton-Euler model

From the Euler’s first axiom of the Newton’s second law follows the
derivation of the linear components of the body motion, according to

Jη̇I = τI
d
(WI JωB ) = WI τB
dt
WI Jω̇B + ẆI ωB = WI τB
WI (Jω̇B + ωB × JωB ) = WI τB
Jω̇B + ωB × JωB = τB (15)

Control of unmanned vehicles 20 de septiembre de 2024 23 / 46

Newton-Euler model

By putting together (14) and (15), the dynamic equations of UAV

(rigid body) subject to external forces applied to the center of mass
and expressed in B-frame

mV̇B + ωB × mVB = FB (16)

Jω̇B + ωB × JωB = τB (17)

where J ∈ R3×3 is the inertia matrix.

FB and τB are the external forces and torques applied to the UAV:
weight, aerodynamic forces, thrust torques developed by the four
rotors.

Control of unmanned vehicles 20 de septiembre de 2024 24 / 46

Newton-Euler model

Therefore it is possible to rewrite equation (16)-(17) in a matrix form

MB ν̇ + CB (ν)ν = Λ (18)

where MB is the system inertia matrix, CB (ν) is the

Coriolis-centripetal matrix and Λ is the generalized force vector.
 T  T
Λ= FB τB = Fx Fy Fz τx τy τz (19)

Control of unmanned vehicles 20 de septiembre de 2024 25 / 46

Newton-Euler model
Gravitational vector
The first contribution is the gravitational vector given from the
acceleration due to gravity g . This force affects only the linear
equations since it is not a torque, and it is defined en I-frame.
 
0
FGI =  0  (20)
−mg

Therefore, in B-frame
   
0 mg sin θ
FGB = R−1 I T 
I FG = RI 0  =  −mg cos θ sin φ  (21)
−mg −mg cos θ cos φ

Control of unmanned vehicles 20 de septiembre de 2024 26 / 46

Newton-Euler model

Lift force vector

The force directly produced by the main action of four rotors. From
aerodynamics consideration, the force of each rotor is proportional to
the squared propellers’ speed, f = bΩ2 . Therefore the throttle force
is obtained by the sum of each individual force, and is in opposite
direction of gravity force
   
0 0
FLB =  0  =  0  (22)
2 2 2 2
u1 b (Ω1 + Ω2 + Ω3 + Ω4 )

The force u1 affects all the three linear equations through the
rotation matrix RI .

Control of unmanned vehicles 20 de septiembre de 2024 27 / 46

Newton-Euler model

Gyroscopic vector
The gyroscopic effects are produced by the propeller rotation:
Two rotors are rotating in clockwise and other two in
counterclockwise
Imbalance when rotor speeds are not equal
When roll or pitch rates are different than zero

Control of unmanned vehicles 20 de septiembre de 2024 28 / 46

Newton-Euler model
Gyroscopic vector
 
4
X 0
B
τgyro = − JR ωB ×  0  (−1)i Ωi
i=1 1
 
X 4 q
= − JR  −p  (−1)i Ωi
i=1 0
 
q
= −JR  −p  (−Ω1 + Ω2 − Ω3 + Ω4 )
0
 
q
= JR Ω  −p  (23)
0

whit Ω = −Ω1 + Ω2 − Ω3 + Ω4
Control of unmanned vehicles 20 de septiembre de 2024 29 / 46
Newton-Euler model

Roll, pitch and yaw moments vector τaB

This vector express the torques developed by the four rotors
roll moment is produced by the difference of thrust between the
left rotor and the right rotor
pitch moment is produced by the difference of thrust between
the left front and the back rotor
yaw moment is produced by the difference of torque between the
two clockwise turning rotors and the two counter-clockwise
turning rotors


Lb Ω22 − Ω24

     
L(f2 − f4 ) lu2
τaB =  L(f3 − f1 )  =  Lb Ω33 − Ω21
= lu3
   (24)
P4 2 3 2 2
τ
i=1 Mi d Ω 1 + Ω 3 − Ω 2 − Ω 4 u 4

with l the distnace between the rotors and the center of gravity

Control of unmanned vehicles 20 de septiembre de 2024 30 / 46

Newton-Euler model

Aerodynamic force and torque vectors

 
Ax
ABL =  Ay  (25)
Az
 
Ap
ABR =  Aq  (26)
Ar

Control of unmanned vehicles 20 de septiembre de 2024 31 / 46

Newton-Euler model

The generalized force vector Λ can be rewritten as follows

FB = FGB + FLB + FTB (27)

     
mg sin θ 0 Ax
=  −mg cos θ sin φ  +  0  +  Ay 
−mg cos θ cos φ u1 Az

B
τB = τgyro + τaB + τRB (28)
     
−q Lu2 Ap
= JR Ω  p  +  Lu3  +  Aq 
0 u4 Ar

Control of unmanned vehicles 20 de septiembre de 2024 32 / 46

Newton-Euler model

In I-frame only the vectors FGB and FLB are different from that ones
in the B-frame because affects all the three linear equations through
the rotation matrix RI . The gyroscopic effects produced by the
propeller rotation is unvaried because it affects only the angular
equations referred to the B-frame.
 
0
FGI = RI FGB =  0  (29)
−mg
 
(cos ψ sin θ cos φ + sin ψ sin φ)u1
FLI = RI FLB =  (sin ψ sin θ cos φ − cos ψ sin φ)u1  (30)
(cos φ cos θ)u1

Control of unmanned vehicles 20 de septiembre de 2024 33 / 46

Newton-Euler model

For linear movement equations

   
ẋ u0
ξ˙ = v −→  ẏ  =  v0  (31)
ż w0
mv̇ = R F
  I B     
u̇0 0 (Cψ Sθ Cφ + Sψ Sφ )u1 Ax
m  v̇0  =  0  +  (Sψ Sθ Cφ − Cψ Sφ )u1  +  Ay  (32)
ẇ0 −mg Cφ Cθ u1 Az

Control of unmanned vehicles 20 de septiembre de 2024 34 / 46

Newton-Euler model

For angular movement equations

R˙I = RI S(ω) −→ η̇ = WI−1 ω (33)

Jω̇B = −ωB × JωB + τB

       
Ixx ṗ (Iyy − Izz )qr −q Lu2
 Iyy q̇  =  (Izz − Ixx )pr  + JR Ω  p  +  Lu3 
Izz r˙ (Ixx − Iyy )pq 0 u4
 
Ap
+ Aq 
 (34)
Ar

Control of unmanned vehicles 20 de septiembre de 2024 35 / 46

Newton-Euler model

mẍ = (cos ψ sin θ cos φ + sin ψ sin φ)u1 + Ax (35)

mÿ = (sin ψ sin θ cos φ − cos ψ sin φ)u1 + Ay (36)
mz̈ = −mg + cos φ cos θu1 + Az (37)
φ̇ = p + q sin φ tan θ + r cos φ tan θ (38)
θ̇ = q cos φ − r sin φ (39)
ψ̇ = q sin φ sec θ + r cos φ sec θ (40)
Ixx ṗ = (Iyy − Izz )qr − JR Ωq + Lu2 (41)
Iyy q̇ = (Izz − Ixx )pr + JR Ωp + Lu3 (42)
Izz r˙ = (Ixx − Iyy )pq + u4 (43)

Control of unmanned vehicles 20 de septiembre de 2024 36 / 46

Newton-Euler model

The mathematical model (35)-(43) is assumed to be sufficiently

accurate in representing all UAV functional motions. However,
dependence on aerodynamic forces AL and moments AR does not
make it suitable for control design. Thus, during control design these
vectors are considered as external disturbances.

Control of unmanned vehicles 20 de septiembre de 2024 37 / 46

Euler-Lagrange model

The motion equations can also be obtained by directly applying

the Lagrange-Euler formulation.
This method considers the balance of the potential and kinetic
energies of the system.
In order to develop the equations, remember that we deal with
two orthogonal reference frames I and B, that is 0I xI yI zI and
0B xB yB zB , respectively.
The translationalT and rotational
 T in I are
coordinates
ξ= x y z and η = φ θ ψ

Control of unmanned vehicles 20 de septiembre de 2024 38 / 46

Euler-Lagrange model

The general form of the equations of motion in Lagrange method is

 
d ∂L ∂L
Γi = − (44)
dt ∂ q̇i ∂qi

where
L is the Lagrangian, the different between the kinematic and
potential energies, L = T − U
 T
qi is the generalized coordinates q = x y z φ θ ψ
Γi is the external generalized forces

Control of unmanned vehicles 20 de septiembre de 2024 39 / 46

Euler-Lagrange model
The translational kinetic energy of the UAV is
m
Ttrans = ξ˙T ξ˙ (45)
2
where m denote the mass of the rotorcraft
The rotational kinetic energy is
1
Trot = η̇ T J (η)η̇ (46)
2
The matrix J is the inertia matrix for the full rotational kinetic
energy of the rotorcraft expressed directly in terms of the
generalized coordinates η.
The only potential energy which needs to be considered is the
standard gravitational potential given by
U = mgz (47)
Control of unmanned vehicles 20 de septiembre de 2024 40 / 46
Euler-Lagrange model

The external generalized force Γ include the τ generalized moments

and FGI applied to the UAV due to the control inputs.
 
(cos ψ sin θ cos φ + sin ψ sin φ)u1
 I   (sin ψ sin θ cos φ − cos ψ sin φ)u1 
 
FG  (cos φ cos θ)u1 
Γ= I =  (48)
τa 
 Lu 2


 Lu3 
u4

with L the distance between the rotors and the center of gravity

Control of unmanned vehicles 20 de septiembre de 2024 41 / 46

Euler-Lagrange model
The Euler-Lagrange equation for translational dynamics are
L(q, q̇) = Ttrans + Trot − U
m ˙T ˙ 1 T
= ξ ξ + η̇ J (η)η̇ − mgz (49)
2 2
Since the Lagrangian (49) contains no cross-terms in the kinetic
energy T combining ξ˙ and η̇, the Euler-Lagrange equation can be
partitioned into the dynamics for the ξ coordinates and the η
dynamics.
!
d ˙
∂L(ξ, ξ) ˙
∂L(ξ, ξ)
− = FGI (50)
dt ∂ ξ˙ ∂ξ
 
d ∂L(η, η̇) ∂L(η, η̇)
− = τaI (51)
dt ∂ η̇ ∂η
Control of unmanned vehicles 20 de septiembre de 2024 42 / 46
Euler-Lagrange model

Developing (50) and (51) yields

 
0
mξ¨ +  0  = FGI
mg
J (η)η̈ + C(η, η̈)η̇ = τaI

which are the same equations in (35)-(43).

Control of unmanned vehicles 20 de septiembre de 2024 43 / 46

References I

J-H. Park, Ji-H. Bae, M-H. Baeg

Estimation of Fluctuation Characterizations by USV-Operation
Simulations in Sea State 3.
Modelling and Simulation in Engineering, Hindawi. pp.
1687–1691, 2017. https://doi.org/10.1155/2017/1305752
D. O. B. Jones, A. R. Gates, V. A. I. Huvenne, A. B. Phillips,
B. J. Bett
Autonomous marine environmental monitoring: Application in
decommissioned oil fields.
Science of the Total Environment, Vol. 668, pp. 835–853, 2019.
https://doi.org/10.1016/j.scitotenv.2019.02.310

Control of unmanned vehicles 20 de septiembre de 2024 44 / 46

References II

R. Siegwart, I. Nourbakhsht
Introduction to autonomous mobile robots
The MIT Press, 2004. ISBN 0-262-19502-X
P. Castillo, R. Lozano, A. E. Dzul
Modelling and control of mini-flying machines. (Advances in
industrial control)
Springer, 2005. ISBN 1-85233-957-8
S. A. Wadoo, P. Kachroo
Autonomous Underwater Vehicles: modeling, control design, and
simulation
CRC Press, 2011. ISBN 978-1-4398-1831-2

Control of unmanned vehicles 20 de septiembre de 2024 45 / 46

References III

M. W. Spong, S. Hutchinson, M. Vidyasagar

Robot Modeling and Control
Jhon Wiley and Sons, Inc., 2005.
J. J. Craig
Introduction to robotics. Mechanics and control
Pearson Education, Inc., 2005.
M. J. Velueta Guzmán
Navegación autónoma de un vehı́culo acuático notripulado
dedicado a la exploración de zonas costeras
Master Thesis, 2018. Universidad Autánoma del Carmen

Control of unmanned vehicles 20 de septiembre de 2024 46 / 46