You are on page 1of 5

2020 3rd International Conference on Control and Robots

6WHHULQJ &RQWURO RI DQ $XWRQRPRXV 8QGHUZDWHU9HKLFOH XVLQJ 60&


7HFKQLTXHV

Muhammad Mudassir Attaullah Y. Memon


Department of Electronics and Power Engineering Department of Electronics and Power Engineering
PN Engineering College PN Engineering College
National University of Sciences and Technology National University of Sciences and Technology
Karachi, Pakistan Karachi, Pakistan
mmudassir.msee18pnec@student.nust.edu.pk attaullah@pnec.nust.edu.pk
2020 3rd International Conference on Control and Robots (ICCR) | 978-1-7281-7562-1/20/$31.00 ©2020 IEEE | DOI: 10.1109/ICCR51572.2020.9344394

Abstract—This paper presents the problem of steering control control design of AUVs due to complication of the AUV
for autonomous underwater vehicles (AUVs). A six degrees-of- nonlinear dynamics, effects of unmodelled fast dynamics,
freedom (DOF) system of coupled equations of motion of an system uncertainties and external disturbances. Moreover, the
AUV is first decoupled as a lateral model to control steering
angle of an AUV. Sliding Mode Control (SMC) technique is AUV models considered commonly are underactuated where
used to stabilize the system and track a constant reference yaw the size of input vector is smaller than that of the output
angle in presence of model perturbations and uncertainties. Using vector. Different types of methods have been proposed in
numerical simulations, it is shown that the closed-loop system order to deal with the complication and nonlinearity of AUV’s
under the proposed control scheme is stable and the tracking dynamics. In [3], linearization technique was proposed in
error converges to zero in presence of model perturbations and
uncertainties. order to reduce the complication of AUV nonlinear model
under consideration. Another method used in control design
Keywords—Autonomous Underwater Vehicle, Sliding Mode of AUVs is to reduce the number of DOF into three or
Control, Integral Sliding Mode Control less [4]. Motion control of unmanned underwater vehicles
is typically achieved using different control techniques such
I. I NTRODUCTION as state feedback linearization, proportional integral derivative
(PID), SMC, optimal control and adaptive control. Lately, Bo
Underwater robotics plays an important role in the ocean
hu [5] presented a fuzzy PID control to stabilize the depth of
exploring fields such as inspection and monitoring of under-
an AUV. By using this approach response of the system was
water man-made or natural structures, search, detection and
shown to be fast and achieve the required position accuracy
evaluation of underwater objects and a variety of military
with zero steady-state error.
missions. For some of these purposes, either remotely operated
unmanned underwater vehicles (ROVs) or autonomous under- Main contribution of this paper is use of a Sliding Mode
water vehicles (AUVs) are utilized. AUV offers advantages in Control (SMC) technique to achieve tracking of the yaw angle
term of its long endurance and ease of deployment, however, at of AUVs in steering plane. SMC is regarded as one of the most
the cost of challenges in its precise navigation, path planning, efficient control techniques to that yields in a robust controller
obstacle avoidance and recovery at the end of an underwater for a variety of input-output linearizable nonlinear systems.
deployment exercise. AUVs are required to follow a set of SMC is also referred to as a nonlinear control design technique
trajectories which are defined apriori, however the AUVs mo- which possesses two important capabilities. First, it can reject
tion depends heavily on continuous updation of these through the external disturbances when integral action is taken and
suitable obstacle avoidance/ path planning algorithms in order second, it is robust to model parameters perturbations. The
to cater for a variety of uncertainties and disturbances such as proposed control law tackles yaw angle problem in presence
ocean currents, tidal effects etc. of model perturbations and external disturbances.

Researchers such as Khac Duc Jie Pan [1], and Thor This paper is structured as follows: In Section II, the
I.Fossen [2], have made some important contributions in coordinate transformation, kinematics and dynamics modeling
marine vehicle system modelling and control techniques for are discussed. In Section III the control scheme is proposed
AUVs. One of the most challenging issues in AUVs is efficient that stabilizes the AUV. Section IV presents the simulation
and reliable motion control. There are a lot of challenges in results and validation of control law with and without model
perturbations. Section V presents the robustness properties,
and finally section VI draws the conclusions.

978-1-7281-7562-1/20/$31.00 ©2020 IEEE 154

Authorized licensed use limited to: Central Mechanical Engineering Research Institute. Downloaded on October 21,2022 at 07:33:18 UTC from IEEE Xplore. Restrictions apply.
II. M ATHEMATICAL M ODELING OF AUV After summing up to the non-linear equation of motion of
A. Coordinate Systems AUV as,

While investigating the motion of AUV in 6 DOF it is ν̇ = M −1 [C(ν) ν + Dm(ν) ν + G(Q) + τr ]


essential to define two coordinate systems as shown in Fig. (4)
Q̇ = J(Q) ν
1. The moving coordinate system (O-xyz) is fixed to the
vehicle and is called the body-fixed reference system, [4]. The Equation (4) represents the dynamics and kinematics equations
motion of body-fixed reference system is depicted to earth of AUV.
fixed reference system (φ,θ,Ψ). C. Decoupled AUV Model
For controlling steering angle of AUV, we need to take
sway(v), velocity of yaw(r) and yaw(ψ). Steering motion can
be attained with the help of rudders and fins. To appropriately
express equations of motion in steering plane by neglecting z,
w, p, q, θ and φ equal to zero. Also neglecting gravitational
forces and damping, decoupled AUV system is given as,

[m − Yv̇ ]v̇ = (Xu̇ − m) u0 r + Yv v + Yr r


[Iz − Nṙ ]ṙ = (Yv̇ − Xu̇ ) v r + Yṙ u0 r + Nv v + Nr r + τr
ψ̇ = r
(5)
According to [4], the kinematics and dynamics equations are
shown in (5). It can be seen that there is no input in the
sway equation. Thus, the problem being addressed is that
of an underactuated system with only one actuator available
Fig. 1. Earth and Body fixed reference System.
that exerts force and a moment in the yaw direction. Table 1
represents the parameters of AUV.
B. AUV Modeling in 6-DOF
TABLE I
The six DOF AUV’s equation of motion follows rigid body PARAMETERS OF AUV
dynamics. Let us consider, Q=[N ,B], where N=[x,y,z] and
B=(φ,θ,Ψ) are the position and orientation of body and earth AUV Parameters Name
fixed reference system. Body fixed system B is transformed Y v , Y r , Nr Cross flow drag
into earth fixed reference system N. Linear and angular veloci- Xu̇ ,Yṙ ,Yv̇ ,Nṙ Added masses
ties are denoted as V1 =[u,v,w] and V2 =[p,q,r] respectively. The m mass of the AUV
τr Motion executor input force to AUV or yaw moment
AUV can be represented by the following dynamic equation:

M ν̇ + C(ν) ν + Dm(ν) ν + G(Q) = τ (1) III. CONTROL DESIGN


Where M is added mass, C(V) is coriolis forces, D(V) is This section focuses on design of controller that solves
damping matrix and G(Q) are gravitational matrix. The matrix stabilization problem for the AUV system under consideration.
J1 (ω) which describes orientation related to roll, pitch and We use SMC technique to control the yaw motion. The design
yaw is given as, process starts with selecting an appropriate sliding surface ‘s’,
described by,
J1 (ω) = rx ry rz S(x) = 0 (6)
J1−1 (ω) = J1 (ω) (2)
The objective of the control design is then to ensure that
J2−1 (μ) = J2 ( μ) the system trajectories not only reach the surface in finite
Where J2−1 (μ) is coordinate transformation matrix. Equa- time, but continue sliding along the surface afterwards. This
tion(2) results in overall Kinematics equation of motion for is practically achieved with the help of a high-frequency
an AUV in 6 DOF is given as, switching control signal. However, this approach oftentimes
causes the system to ”chatter” in the neighborhood of the
     sliding surface. Effect of chattering is usually reduced through
Q̇1 J (ω) 0 ν1 the use of a boundary layer around the sliding surface.
= 2
Q̇2 0 J2 (μ) ν2 (3) In essence, SMC technique uses infinite gain to force the
trajectories of the closed-loop system to slide along the desired

155

Authorized licensed use limited to: Central Mechanical Engineering Research Institute. Downloaded on October 21,2022 at 07:33:18 UTC from IEEE Xplore. Restrictions apply.
surface. As a result, the closed-loop system trajectories have Theorem 1: Consider the closed loop system of an AUV
desirable performance properties. The main strength of SMC
in the steering plane describing of (5) and (14). Let the
is its robustness. Since the control can be implemented as a
tracking error of the yaw angle shown in (7) with the desired
switching one, it does not need to be precise. As a result it is
yaw angle. When the stabilizing control law τr for the yaw
usually not sensitive to parameter variations that appear in the
moment, proposed in (14) applied to the AUV, then the
closed-loop system. Furthermore, since the control law is not
convergence of e1 to zero is guaranteed.
continuous, the trajectories can reach the surface in finite time
which is desirable when compared to an asymptotic behaviour
Proof: Take Lyapunov function candidate as,
obtained through conventional control methods. Let us define
the errors 1 2
V1 = S
e1 = ψ − ψd 2
V˙1 = S Ṡ
e˙1 = ψ̇ − ψ˙d (7)
V˙1 = S[a1 e˙1 + e¨1 ]
e¨1 = ψ̈ − ψ¨d (15)
V˙1 = S[a1 r + ṙ]
Where ψd represents desired yaw angle. After the change
V̇! = S[a1 r + (Iz − Nṙ )−1 ((Yv̇ − Xu̇ ) v r + Yṙ u0 r+
in variables, the system becomes,
Nv v + Nr r + τr )]
e¨1 = (Iz − Nṙ )−1 [(Yv̇ − Xu̇ ) v r + Yṙ u r + Nv v + Nr r + τr ] By substituting the value of τr in above equation we get,
σ̇ = (Iz − Nṙ )−1 [(Xu̇ − m) u r + Yv v + Yr r] S
V˙1 = S(Iz − Nṙ )[−βo Sat( )]
(8)
(16)
V˙1 ≤ − | S | γ
Define the sliding surface as,
This indicates that trajectory initiate from initial condition stay
S = a1 e1 + e˙1 (9)
inside the boundary set ω = | S |, reaches the boundary layer
Take derivative of (9) we get, in a finite time and stay inside in it. This also indicates that the
closed-loop system with control input τr reaches the sliding
surface in finite time and stay inside in it.
Ṡ = a1 e˙1 + e¨1 (10)
Let the Lyapunov function candidate as,
After substituting (8) in (10) we get, 1 2
V2 = σ
Ṡ = a1 e˙1 + (Iz − Nṙ ) −1
[(Yv̇ − Xu̇ ) v r + Yṙ u r + Nv v+ 2
N r r + τr ] V˙2 = σ σ̇
(11) V˙2 = σ(Iz − Nṙ )−1 [(Xu̇ − m) u r+
Now, design a state feedback SMC law that stabilizes the sys- Yv v + Yr r]
tem and track a constant yaw angle. There are two components V˙2 = σ[Z1 r + Z2 v] (17)
of SMC, first is stabilizing control, second is switching control. V˙2 = Z2 σ 2 + Z1 σr
Overall SMC is given as,
V˙2 = Z2 σ 2 + Z1 σr − Z2 θ1 σ 2 + Z2 θ1 σ 2
V˙2 = (1 − θ1 )Z2 σ 2 + Z1 σ + Z2 θ1 σ 2
τr = τ1 + τ2 (12)
Z1 r
V˙2 ≤ (1 − θ1 )Z2 σ 2 , ∀ | σ |≥ −
Here, Z 2 θ1
τ1 = (Iz − Nṙ )[−(Yv̇ − Xu̇ ) v r − Yṙ u r − Nv v − Nr r Here Z1 = (Iz − Nṙ )−1 [(Xu̇ − m) u + Yr ) and Z2 =
−a1 e˙1 ] (Iz − Nṙ )−1 Yv . With 0<θ1 <1. Now the system of (8) is semi
globally asymptotically stable.
τ2 = (Iz − Nṙ )βo Sat(S/
)
(13)
IV. SIMULATION RESULTS
Where
is the boundary layer for the switching control and This section presents simulation results of the closed-loop
it must be greater than zero, and β0 is the gain that must be system performance when the control law as proposed in
greater than zero. The advantage of saturation function is to the previous section is applied to the steering model (8)
remove chattering. Now the overall control law is given as, of an AUV. The control objective is to stabilize the AUV
steering angle and track a reference yaw angle (both with
τr = (Iz − Nṙ )[−(Yv̇ − Xu̇ ) v r − Yṙ u r − Nv v − Nr r and without model perturbations). For this purpose, we use
−a1 e˙1 + βo Sat(S/
)] Matlab Simulink to simulate the closed-loop system. It can be
(14) see that the control input goes to zero when the yaw angle goes

156

Authorized licensed use limited to: Central Mechanical Engineering Research Institute. Downloaded on October 21,2022 at 07:33:18 UTC from IEEE Xplore. Restrictions apply.
to the desired yaw angle ψd in a finite time. Consequently the
tracking error of yaw angle also goes to zero when desired
yaw motion is achieved. We consider u = 10 m/s, and set
initial condition of ψ(0) = 1.. All parameters of AUV used
during these simulations are listed in Table 2, [6]. Simulation
results thus obtained are shown in Fig. 2 through Fig. 4.

TABLE II
T HE AUV MODEL PARAMETERS USED FOR SIMULATIONS

Yv = −131 Xu̇ = −0.9 β = 20


Yr = −0.632 Yṙ = 1.93  = 0.5
Nv = −3.2 Nṙ = −4.9 a1 > 0
Yv̇ = −36 m = 40

Fig. 3. Tracking Error of Yaw.

Fig.2 shows the performance of proposed control law under


the nominal parameters of an AUV as indicated in the table(2),
which stabilizes the AUV yaw motion at desired yaw angle
ψd very efficiently in a finite set of time under the influence
of controller. Fig.3 shows the tracking error of yaw angle e1
between yaw motion ψ and desired yaw angle ψd goes to zero
when an AUV stabilizes at desired yaw motion.

Fig.4 shows the performance of proposed control law under


parametric perturbations, where the parameters of an AUV as
indicated in table 2 are perturbed to 20% of the actual value. It
shows that the control law works very well under parametric
perturbations and perform desired yaw motion in a finite set
of time with zero tracking error e1 .

Fig. 4. Yaw Angle and Tracking Error of r.

V. DISTURBANCE REJECTION USING ISMC


In this section, we deal with the external disturbance
rejection. When external disturbances caused due to envi-
ronmental change such as ocean current etc. For external
disturbance rejection, we include an integrator in the sliding
surface equation. ISMC performs a robust operation to external
disturbances, [7]. The uncertainties are added in the AUV
model as shown by (5). Now the new AUV model becomes.

[m − Yv̇ ]v̇ = (Xu̇ − m) u0 r + Yv v + Yr r


[Iz − Nṙ ]ṙ = (Yv̇ − Xu̇ ) v r + Yṙ u0 r + Nv v + Nr r+
τr + d1 (t)
ψ̇ = r
Fig. 2. Yaw Angle and Control Input. (18)
Where d1 (t) is the external disturbance. The following two
uncertainties cases are taken in the simulations,

157

Authorized licensed use limited to: Central Mechanical Engineering Research Institute. Downloaded on October 21,2022 at 07:33:18 UTC from IEEE Xplore. Restrictions apply.
1. Constant Disturbance
In this case, a constant disturbance applied to an AUV.

d1 (t) = 2 (19)

2. Sinusoidal Disturbance
In this case, sinusoidal disturbance applied to an AUV.

d1 (t) = 2 cos t (20)

For compensating the effect of unknown uncertainties the


integral action is taken in the sliding surface. The robust
sliding surface is shown in (21).

Fig. 7. Yaw angle of an AUV under a variety of conditions

performance of proposed ISMC applied to an AUV which


rejects the external disturbance and stabilizes the AUV at
desired yaw angle ψd . It also shows that the proposed ISMC
compensate for the effect of external disturbance. Fig. 7 shows
the performance of an AUV under a variety of conditions. It
shows that in all cases like unperturbed, perturbed, constant
and sinusoidal disturbance the proposed controller works very
well.
VI. C ONCLUSION
Fig. 5. Yaw angle with Constant Disturbance In this paper, a robust feedback dynamic controller tech-
nique has been addressed. A nonlinear steering model of
an AUV has been considered. With the help of numerical
simulations through Matlab Simulink, it is seen that the
proposed control law works very well with and without model
perturbations. Furthermore, external disturbances are added
to the system and ISMC technique adopted for rejection of
external disturbances. Moreover, the numerical simulations
imply that the proposed control design techniques are robust
to model perturbations and disturbances.
R EFERENCES
[1] Khac duc Do, Jie PAn, Control of Ships and Underwater Vehicles. Mill
Valley, Crawley: The University of Western Australia, 2009.
[2] Thor I.Fossen, Marine Control Systems Guidance, Navigation and
Control of Ships, Rigs and Underwater Vehicles,Trondheim: Norwegian
University of Science and Technology,2002.
[3] U Adeely, AA Zaidiz, AY Memon, “Path tracking of a heavy weight
Fig. 6. Yaw angle with Sinusoidal Disturbances torpedo in diving plane using an output feedback sliding mode con-
troller,” 12th International Bhurban Conference on Applied Sciences and
Technology (IBCAST),2015.
 [4] Thor I.Fossen, Handbook of marine craft and motion control,John Wiley
SR = a1 e1 + e˙1 + a2 e1 dt (21) and Sons,2011.
[5] Bo Hu , Hai Tian , Jiani Qian , Guochao Xie , Linlang Mo , Shuo
Zhang “A fuzzy-PID method to improve the depth control of AUV,”
Now, the overall robust sliding mode control is shown in (22). IEEE International Conference on Mechatronics and Automation,2013.
[6] Prestero, Timothy “Verification of a six-degree of freedom sim-
τrobust = (Iz − Nṙ )[−(Yv̇ − Xu̇ ) v r − Yṙ u r − Nv v − Nr r ulation model for the REMUS autonomous underwater vehicle”
−a1 e˙1 + βo Sat(SR /
)] http://hdl.handle.net/1721.1/65068,2001.
[7] Yong Ping Pan, Changuang Yang “Integral sliding mode control: Perfor-
(22) mance, Modification and Improvement”, IEEE Transactions on Industrial
The simulation results of the two scenarios given by Informatics,2018.
(19) and (20) are present. In Fig. 5 and Fig. 6 shows the

158

Authorized licensed use limited to: Central Mechanical Engineering Research Institute. Downloaded on October 21,2022 at 07:33:18 UTC from IEEE Xplore. Restrictions apply.

You might also like