IET Intelligent Transport Systems

Special Issue: AI Applications to Intelligent Vehicles for Advancing

Intelligent Transport Systems

Three-dimensional trajectory tracking of an ISSN 1751-956X

Received on 3rd June 2019
Revised 13th August 2019
underactuated AUV based on fuzzy dynamic Accepted on 25th September 2019
E-First on 20th November 2019
surface control doi: 10.1049/iet-its.2019.0347
www.ietdl.org

Xiao Liang1,2, Xingru Qu1, Ning Wang3 , Rubo Zhang4, Ye Li1,2

1School of Naval Architecture and Ocean Engineering, Dalian Maritime University, Dalian, Liaoning Province, People's Republic of China
2Science and Technology on Underwater Vehicle Technology, Harbin, Heilongjiang Province, People's Republic of China
3School of Marine Electrical Engineering, Dalian Maritime University, Dalian, Liaoning Province, People's Republic of China
4School of Mechanical and Electrical Engineering, Dalian Minzu University, Dalian, Liaoning Province, People's Republic of China

E-mail: [email protected]

Abstract: The three-dimensional trajectory tracking of an underactuated autonomous underwater vehicle (AUV) under the
complex unknowns including model uncertainties and time-varying disturbances is studied. The reference pitch angle and yaw
angle are designed in kinematics, based on the time-varying reference trajectory. Dynamics controllers are developed by
incorporating the first-order filter into the dynamic surface control (DSC), which simplifies the design process and overcome
differential explosion in the traditional backstepping. To reduce the influence of the complex unknowns, an adaptive fuzzy-based
DSC scheme is employed to identify the lumped disturbances with arbitrary accuracy and further enhance system robustness.
Lyapunov stability analysis demonstrates that tracking errors are bounded and converge to an arbitrarily small neighbourhood of
zero. Finally, simulation studies and comparisons with DSC scheme are carried out to illustrate the effectiveness and superiority
of the proposed scheme.

1 Introduction novel sideslip observer is designed for estimation, which

strengthens system stability and enhances tracking performance
In recent years, autonomous underwater vehicles (AUVs) have [24]. Mu et al. improved adaptive line-of-sight guidance [25] using
been the focus of marine applications due to their autonomous a fuzzy algorithm and further constructed the neural shunting
navigation and decision-making ability. These applications include model to solve the differential explosion [26]. In addition to finite-
ocean sampling, marine survey and military operation, and more. time uncertainty observers, Wang et al. researched the global
Unfortunately, AUVs suffer from model uncertainties and current asymptotic trajectory of an asymmetric USV and designed yaw-
disturbances [1], and control systems must consider the time-delay guided dynamics without persistent excitation [27]. Moreover,
[2, 3] and finite time convergence [4, 5] in engineering similar concern on finite-time observer was investigated in [28–
applications, which cause great difficulties to collect data, such as 30]. Recently, to estimate model uncertainties and time-varying
research for the MH370 disappeared in the Indian Ocean on March disturbances, a universal approximator using neural network and
8, 2014. In this aspect, trajectory tracking of an underactuated fuzzy logic is employed in tracking control. Based on the
AUV has played a central role and aroused wide public concern [6, combination of a fuzzy logic system and dynamic surface control
7]. (DSC), the unknown non-linear function can be estimated with
However, the underactuated AUV control has many difficulties arbitrary accuracy [31]. In [32], an online fuzzy approximator with
in comparison with the fully actuated vehicle, since the control self-constructing T-S fuzzy rules is proposed. Considering the
inputs with only three actuators are less than the system freedom finite-time and time-delay control in engineering applications,
[8, 9]. By virtue of the control theory development, different robust controllers can be researched in [33] using fuzzy logic and
control techniques have been used such as backstepping control sliding mode [34, 35]. Therefore, it is of practical importance for
[10–12], sliding mode control [13, 14], neural network control vehicles under complex unknowns to derive robust controllers.
[15], fuzzy control [16–18], and so on. Backstepping control of an Motivated by the above observations, this study focuses on the
underactuated AUV is used in [7] and tracking error dynamics is three-dimensional trajectory tracking of an underactuated AUV
solved via virtual guidance. The controller synthesis using under the complex unknowns. To simplify the traditional
parameter projection and backstepping is developed in [10], and backstepping and overcome differential explosion, the DSC is
tracking errors can be guaranteed with arbitrary small. To avoid the proposed. Besides, an adaptive fuzzy approximator with self-
sway dynamics influence and overcome underactuated difficulties, constructing fuzzy rules is created to estimate the lumped
two coordinate transformations with ship position and tracking disturbances including model uncertainties and time-varying
errors are designed, which provide convenience for the global disturbances. By virtue of the adaptive fuzzy-based DSC (AF-
smooth controllers [11]. The sliding mode control is not sensitive DSC) scheme, robust controllers and adaptive laws are proposed.
to model uncertainties and external disturbances, thereby high- Finally, simulation studies and comparisons with the DSC scheme
order sliding mode and terminal sliding mode are further designed are carried out to illustrate the superiority of the proposed scheme,
for surge control and heading control [19, 20]. In [21], a high-order which can provide technical guidance for vehicle engineering
sliding mode control for an unmanned surface vehicle (USV) is applications.
employed, which avoids chattering in the traditional sliding mode. The rest of this paper is organised as follows. In Section 2, the
In addition, external disturbances can be estimated using non-linear five-degree-of-freedom (5-DOF) dynamics and tracking error
observers. In [22], a position-tracking controller and a current dynamics are presented. In Section 3, a fuzzy approximator is
observer [23] with globally K-exponentially stability are designed proposed. An AF-DSC scheme is presented in Section 4. In
for AUV in the presence of unknown current disturbances. Note Sections 5 and 6, stability analysis of the closed-loop system and
that the sideslip angle would enlarge tracking errors and affect simulation studies are provided, respectively. Finally, conclusions
tracking performance. To identify an unknown sideslip angle, a are given in Section 7.

IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370 364
© The Institution of Engineering and Technology 2019
 17519578, 2020, 5, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-its.2019.0347 by Readcube (Labtiva Inc.), Wiley Online Library on [09/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.2 Tracking error dynamics
The reference trajectory in the earth-fixed frame can be defined as
[xd(t), yd(t), zd(t)]T, which is related to time. Based on the time-
varying reference trajectory, the reference pitch angle and yaw
angle are given by

arctan(ẏd /ẋd), ẋd ≥ 0

ψd = (5)
π + arctan(ẏd / ẋd), ẋd < 0

and
Fig. 1 Earth-fixed frame and body-fixed frame for AUV
θd = − arctan żd / ẋ2d + ẏ2d (6)
2 Problem formulation
Using the AUV states and the reference variables, the tracking
2.1 AUV kinematic and dynamic models errors are expressed as
Neglect the influence of high-order non-linear hydrodynamic
damping and AUV rolling motion. As shown in Fig. 1, Let xe x − xd
η = [x, y, z, θ, ψ]T be the 5-DOF position (x, y, z) and pitch angle as ye y − yd
well as yaw angle (θ, ψ) of the vehicle in the earth-fixed frame and ze = JT(η) z − zd (7)
let v = [u, v, w, q, r]T be the corresponding linear velocities and θe θ − θd
angular velocities in the body-fixed frame.
The mathematical models of an underactuated AUV can be ψe ψ − ψd
described as [7]
where JT(η) is the transposed matrix of J(η); xe, ye, ze and θe, ψ e are
η̇ = J(η)v position errors and attitude angle errors in the body-fixed frame,
(1)
M(v)v̇ = τ(v(t)) − C(v)v − D(v)v − g(v) + d(t) respectively.
Taking the time derivative for tracking errors, and using (1)
where τ(v(t)) = [τu, 0, 0, τq, τr]T denotes the vector of control inputs yield
and d(t) denotes the vector of time-varying disturbances. M(v) is
the inertial matrix including the added mass, C(v) denotes the ẋe = rye − vpcos θdcos θcos ψ e − qze + u
Coriolis and centripetal matrix caused by the rotation, D(v) is −vpsin θdsin θ
hydrodynamic damping matrix and g(v) is the vector of ẏe = − rxe − zertan θ + vpcosθRsin ψ e + v (8)
gravitational/buoyancy forces and moments. J(η) is the rotation
matrix given by że = yertan θ + vpsin θdcos θ + w + qxe
−vpcos θdsin θcos ψ e
cos ψcos θ −sin ψ cos ψsin θ 0 0
sin ψcos θ cos ψ sin ψsin θ 0 0 where vP = ẋ2R + ẏ2R + ż2R is the reference velocity and satisfies
J(η) = −sin θ 0 cos θ 0 0 (2) vP > 0.
0 0 0 1 0 The control objective is to design robust controllers for an
0 0 0 0 1/cos θ underactuated AUV with the dynamics (3), and further to ensure
that the AUV tracks a given three-dimensional trajectory under the
To simplify the design process and improve the control complex unknowns. Moreover, tracking errors satisfy
performance, the lumped disturbance terms including model
uncertainties and time-varying disturbances are established. lim xe ≤ ε1
t→∞
Therefore, the AUV dynamics with the lumped disturbances in 5-
DOF can be described as lim ye ≤ ε2 (9)
t→∞

m11u̇ = m11 f u + τu lim ze ≤ ε3

t→∞
m22v̇ = m22 f v
for some small constants ε1, ε2 and ε3.
m33ẇ = m33 f w (3)
m55q̇ = m55 f q + τq 3 Fuzzy approximator
m66ṙ = m66 f r + τr In this section, a universal approximator is employed to
approximate the complex unknowns including model uncertainties
where the positive constant terms mii(1 ≤ i ≤ 6) present the AUV and time-varying disturbances. A fuzzy approximator with double-
inertia including added mass; f u, f v, f w, f q and f r are complex input and single-output is proposed and satisfies
^
unknowns and can be defined as supx ∈ R f (x) − f (x) < ε with any ε > 0 [36, 37].
−1
f u = m11 (m22vr − m33wq − d11u + du)
3.1 Fuzzification
−1
f v = m22 ( − m11ur − d22v + dv)
Let each input have seven types of fuzzy subsets, namely positive
−1
f w = m33 (m11uq − d33w + dw) (4) big (PB), positive medium (PM), positive small (PS), zero (ZE),
−1
negative small (NS), negative medium (NM), and negative big
f q = m55 ((m33 − m11)uw − d55q + dq) (NB). In terms of underactuated AUV motion, a fuzzy
−1
f r = m66 ((m11 − m22)uv − d66r + dr) approximator adopts a double-input and single-output structure and
includes fuzzification, fuzzy inference, and defuzzification. The
where dii(1 ≤ i ≤ 6) are the hydrodynamic damping uncertainties. inputs are the velocity tracking errors ue, qe, re and the

IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370 365
© The Institution of Engineering and Technology 2019
 17519578, 2020, 5, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-its.2019.0347 by Readcube (Labtiva Inc.), Wiley Online Library on [09/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1 Fuzzy rules
NB NM NS ZE PS PM PB
NB NB NM NS ZE PS PM PB
NM NB ZE PS PB PB PB PB
NS NB NS ZE PM PM PB PB
ZE NB NS NS PS PM PM PB
PS NB NM NS ZE PS PM PB
PM NB NM NM NS PS PS PB
PB NB NB NM NM ZE PS PB

corresponding change rate u̇e, q̇e, ṙe. Also, the outputs are the ẋi = xi + 1 + f i(xi) + λiζi(x1, …, xi) + di(x, t)
estimations for f u, f q, f r. xn = f (x) + b(x, t)u + λnζn(x) + dn(x, t) (11)
The universe for surge velocity can be expressed as
η = x1, 1≤i≤n
E = −6, 6 , EC = −6, 6 , U = −7, 7 ;
where x = [x1, x2, …, xn] ∈ Rn is the system state, u ∈ R is the
and the universe for angular velocities can be expressed as input, η ∈ R is the output; and λi is an unknown parameter.
Similar to the traditional backstepping design, the DSC
E = −1, 1 , EC = −0.1, 0.1 , U = −1, 1 procedure contains the following steps.
Step 1: Define the first dynamic surface as
where E and EC denote the input deviation and deviation change
rate, respectively. And U denotes the output. z1 = x1 − x1d (12)
Note that a fuzzy approximator cannot be available without
regularisation for the inputs. Therefore, the inputs are converted to where x1d is the reference signal.
the internal universe −1, 1 , where the quantisation factor can be The time derivative of z1 along (11) is
obtained. Also, the membership function is designed using triangle
distribution. ż1 = x2 + f 1(x1) + λ1ζ1 + d1 − ẋ1d (13)
3.2 Fuzzy inference To stabilise (12), a virtual control law χ̄ 2 is designed as
The fuzzy rules of the approximator are established in Table 1, ^
which takes into consideration rapidity, stability, and accuracy. χ̄ 2 = − k1z1 − f 1(x1) − λ1ζ1 − a12z1 /2ε + ẋ1d (14)
Based on the tracking errors ue, qe, and re, we design the fuzzy
rules to estimate the complex unknowns f u, f q and f r. ^
where k1 and ε are positive constants. λ1 is the estimation for λ1.
The definition (E is NB) means that velocities are much higher Importing χ̄ 2 to a first-order filter with time constant ϖ2 > 0, we
than the desired value and the definition (EC is NB) means that the get the filtered control signal x2d as
proposed controllers are driving velocities upward, thereby the
output should be set NB to turn velocities downward quickly. ϖ2 ẋ2d + x2d = χ̄ 2, x2d(0) = χ̄ 2(0) (15)
Moreover, we set the output to be NM or PM to make velocities
upward and asymptotically converge to the desired value without
Step i: Define the ith dynamic surface as
chattering.
By virtue of expert experience, the process of fuzzy inference
zi = xi − xid (16)
including 49 fuzzy rules can be obtained. For instance, the first
fuzzy rule can be described as
where xid is the reference signal.
If E is NB and EC is NB, then U is NB . Then, the corresponding virtual law is
^
3.3 Defuzzification χ̄ i + 1 = − kizi − f i(xi) − λiζi − ai2zi /2ε + ẋid (17)

It is the inverse process of fuzzification. The outputs using fuzzy ^

inference cannot be directly employed by the control unit. The real where ki > 0 and λi is the estimation for λi.
universe to the internal universe can be transformed based on a Step n: Define the nth dynamic surface as
^ ^ ^
membership degree, and the outputs f u, f q and f r are further
zn = xn − xnd (18)
constructed. However, it has difficulties to get the point of
maximum membership in the output universe. By virtue of the The time derivative of zn along (11) is
weighted average method in engineering application, the outputs
are proposed as
żn = f (x) + λnζn + bu + dn − ẋnd (19)
n
^ ∑i = 1 xi μ(xi)
f ( ⋅ )(E, EC) = (i = 1, …, 49) (10) Finally, the actual control law is proposed as
n
∑i = 1 μ(xi)
^ ^
u = β( − f (x) − knzn − λnζn − an2 zn /2ε + ẋnd) (20)
where μ(xi) is the membership function.
^ ^
where β = 1/b. β and λn are the estimations for β and λn,
4 Adaptive fuzzy-based DSC (AF-DSC) respectively.
In this section, the tracking controllers are employed using the AF- The adaptive law can be designed as
DSC scheme [38].
Consider an uncertain non-linear system

366 IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370
© The Institution of Engineering and Technology 2019
 17519578, 2020, 5, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-its.2019.0347 by Readcube (Labtiva Inc.), Wiley Online Library on [09/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
˙^ Based on (14), the time derivative of χ2 can be expressed as
λ1 = c1z1ζ1
˙^ ^ ^
λn = cnznζn (21) χ̇ 2 = − k1ż1 + ẍ1d − λ1ζ1(x1) − λ(∂ζ1 /∂x1)ẋ1 − a12z1 /2ε
(31)
˙^
β = − csgn(b)zn x1d −a12z1 ẋ1 /2ε − (∂ f 1 /∂x1)ẋ1

where ci and c are positive constants. Substituting (29) into (31) yields
Based on AUV dynamics (3) and control laws (20) and (21), we ^
can get the control inputs τu, τr and τq as follows: χ̇ 2 ≤ γ2(z1, z2, y2, λ1, k1, x1d, ẋ1d, ẍ1d) (32)

^ ^ Finally, it follows that

τu = β1( − f u − k1z1 + u̇d)
^ ^
τr = β2( − k2z2 − f r + ṙd) (22) χ̇ i + 1 ≤ γi + 1(O, ẍ1d) (33)
^ ^
τq = β3( − k3z3 − f q + q̇d)
Theorem 1: Consider an underactuated AUV with dynamics (3)
with adaptive law and (4) under the complex unknowns. Select the first-order filter
(14) and the control law (21) with the adaptive law (22). There
˙^
β1 = − ρ1sgn(b1)z1ud exist optimal parameters such that all errors in the closed-loop are
˙^
global asymptotically stable, and the AUV tracks the reference
β2 = − ρ2sgn(b2)z2ψ d (23) trajectory from any initial conditions.
˙^
β3 = − ρ3sgn(b3)z3θd Proof: Consider the following Lyapunov function candidate:
n n−1 n
where k1, k2 and k3 are positive constants; ρ1, ρ2 and ρ3 are positive 1 ~2
constants. z1 = u − ud, z2 = r − rd, z3 = q − qd, where ud = vP and V=
2 ∑ Vis + ∑ Vi y
2 2
( + 1) + ∑ Viλ + b β /c
2
(34)
i=1 i=1 i=1
rd, qd can be formulated using (5), (6) and (15). Also, the
^ ^ ^
estimations f u, f q, f r can be derived by a fuzzy approximator. where V is = zi and V i(y + 1) = yi + 1 are the ith dynamic surface and the
~
filter error, respectively. V iλ = λ i is the estimation error for system
~ ^
5 Stability analysis parameters and satisfies λ i = λi − λi.
To analyse the stability, we define the filter error Taking the time derivative of V is along (25), and using
zi di ≤ (zi2ai2 /2ε) + ε/2 , we get
yi = xid − χ̄ i (24)
~
V̇ is ≤ − kizi2 + zi zi + 1 + yi + 1 + ziλ iζi + ε/2 (35)
Then, the corresponding dynamic surface error can be deduced as
and V ns for the nth dynamic surface
żi = xi + 1 + f i(xi) + λiζ(xi) + di − ẋid
~
= zi + 1 + x(i + 1)d + f i(xi) + λiζ(xi) + di − ẋid (25) V̇ ns ≤ − knzn2 + znλ nζn + ε/2 (36)
~ 2
= zi + 1 + zi + 1 − kizi + λ iζ(xi) − ai zi /2ε + di
Furthermore, we get
Using (17) and (18), we get ~ ~
V̇ iλ = − (1/c)λ i(cziζi) = − V isλ iζi (37)
~
xi + 1 = zi + 1 + yi + 1 − kizi − f i(xi) + λ iζ(xi)
(26) and
−ai2zi /2ε − yi /ϖi
V̇ i(y + 1) = yi + 1ẏi + 1 ≤ −yi2+ 1 /ϖi + 1 + yi + 1 γi + 1(O, ẍ1d) (38)
For i ≥ 2, one can finally obtain
Substituting (35)–(38) into (34), one obtains
^ ^
z1, …, zi + 1, y2, …, yi + 1, …, λ1, …, λi,
xi + 1 = Bi (27) n n−1
ϖ2, …, ϖi, k1, …, ki, x1d, ẋid V̇ ≤ − ∑ kizi2 + ∑ zi zi + 1 + yi + 1 − yi2+ 1 /ϖi + 1
i=1 i=1
Substituting (27) into (25) yields ~ ˙^
+nε/2 − b β β /c + yi + 1γi + 1
~ (39)
żi = zi + 1 + yi + 1 + λ Bi(z1 + x1d, B1, B2, …, Bi − 1) + di n n−1
(28) ≤ − ∑ kizi + 2
∑ zi zi + 1 + yi + 1 − yi + 1 /ϖi + 12

−kizi − ai2zi(z1 + x1d, B1, B2, …, Bi − 1)/2ε i=1 i=1

~ ^
Then + yi + 1γi + 1 + nε/2 − b β sgn(b)zn βu

żi ≤ Mi(O) Substituting (20) into (39), one can get

(29)
dζi /dt ≤ Bi(O)

where Mi and Bi are the functions with respect to O. Also, O is

defined as
^ ^
z1, …, zi + 1, y2, …, yi + 1, …, λ1, …, λi,
O= (30)
ϖ2, …, ϖi, k1, …, ki, x1d, ẋid

IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370 367
© The Institution of Engineering and Technology 2019
 17519578, 2020, 5, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-its.2019.0347 by Readcube (Labtiva Inc.), Wiley Online Library on [09/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
n−1
V̇ ≤ ∑ zi zi − 1 + yi + 1 − yi2+ 1 /ϖi + 1 + yi + 1γi + 1
i=1
~ n
+nε/2 + zn(żn − β λn) − ∑i = 1 kizi2
n−1
(40)
≤ ∑ zi zi − 1 + yi + 1 − yi2+ 1 /ϖi + 1 + yi + 1γi + 1
i=1
n
−knzn2 + (n + 2)ε/2 − ∑i = 1 kizi2

Using ki = 2 + α and 1/ϖi + 1 = 1 + (ςi2+ 1 /2ε) + α, it follows that

n−1
V̇ = ∑ zi2+ 1 + yi2+ 1 /2 + zi2 − 1 + α + αi2+ 1 /2ε yi2+ 1
i=1
n Fig. 2 WL-II
+yi2+ 1γi2+ 1 /2ε + (n + 2)ε − (2 + α) ∑ zi2
i=1
(41)
n
≤ − α ∑i = 1 zi2 + yi2+ 1 + (n + 2)ε
n−1
− ∑i = 1 1 − γi2+ 1 /ςi2 = 1 αi2+ 1zi2+ 1 /2ε

Thus, it can be concluded that V̇ ≤ 0 for any small ki, ϖi + 1 and ε.

Furthermore, the tracking errors converge to zero and the closed-
loop system is stable. This completes the proof. □

6 Simulation studies
In this section, simulation studies are carried out on WL-II (Fig. 2) Fig. 3 Trajectory tracking performance
to illustrate the efficiency and superiority of the proposed AF-DSC
scheme. The simulated environment is conducted on a PC with the
solver: ode3 (Bogacki-Shampine), the fundamental sample time:
0.1, and the simulation time: 150 s.
The underactuated AUV length is 1.46 m, and weight is 45 kg.
Model parameters of this vehicle are presented in [39]:
m11 = 47.52 kg, m22 = 94.05 kg, m33 = 94.12 kg, m55 = 13.52 kg, and
m66 = 13.38 kg. Unknown terms are given by

d11 = 13.5 + 6.44 u

d55 = 23.76 + 3.89 q (42)
d66 = 27.2 + 4.9 r
Fig. 4 Comparisons on separate trajectories
To verify the robustness of controllers, we assume that the AUV
suffers from white noise disturbances with a normal distribution.
The simulated disturbances are constructed to act on the surge,
sway, and yaw axes, respectively, whose corresponding energy
spectral densities are 1, 0.05, and 0.1.
The reference trajectory is given by the helix

xd = Asin(ωt)
yd = Acos(ωt) (43)
zd = − ωt

where A = 10 and ω = 0.1. The initial AUV dynamics and

kinematics are as follows:
Fig. 5 Comparisons on tracking errors
T T
u, v, w, q, r = 0, 0, 0, 0, 0 guaranteed. Comparisons on separate trajectories are given by
(44)
x, y, z T
= 0, 10, 0 T Fig. 4. In comparison with the DSC scheme, the AF-DSC scheme
can achieve superior tracking performance in the presence of
The reference velocity is defined such that ud = vP, and the model uncertainties and time-varying disturbances. Moreover, the
reference angle θd, ψ d can be obtained from (5) and (6). AF-DSC scheme has higher tracking accuracy and faster response.
Comparisons on tracking errors and attitude errors are presented in
The controller parameters of the AF-DSC scheme are selected Figs. 5 and 6, respectively. Tracking errors using the AF-DSC
as follows: k1 = 0.3, k2 = 1.5, k3 = 0.5, ϖ2 = 0.2, ρ1 = 0.5, ρ2 = 8, scheme can converge to zero faster and without chattering
and ρ3 = 2. compared with the DSC scheme. Note that the DSC scheme can
Figs. 3–8 show the simulation results for tracking performance also control AUV to track the reference trajectory with an
using the AF-DSC scheme and DSC scheme. Fig. 3 shows that the overshoot about 10% under the complex unknowns. Fig. 7 plots the
underactuated AUV can track the reference trajectory accurately comparisons on velocity responses, which reveals that AUV can
with the proposed AF-DSC scheme, thereby good performance is converge to the reference velocities. Fig. 8 plots the comparisons

368 IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370
© The Institution of Engineering and Technology 2019
 17519578, 2020, 5, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-its.2019.0347 by Readcube (Labtiva Inc.), Wiley Online Library on [09/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
compensates the lumped disturbances and contributes accurate
tracking performance. Further research will extend this result to the
coordinated tracking of multiple AUVs in the presence of state and
input constraints.

8 Acknowledgments
This work was supported by the National Natural Science
Foundation of China (grant nos. 51579022, 51879023, and
61673084), the Fundamental Research Funds for the Central
Universities of China (grant nos. 3132019111, 3132019002, and
3132019344), the Research Fund from Science and Technology on
Fig. 6 Comparisons on attitude errors
Underwater Vehicle Technology (grant no. 6142215180102), the
Project of Shandong Province Higher Educational Science and
Technology Program (grant no. J18KA043), the Fund for Liaoning
Innovative Talents in Colleges and Universities (under grant no.
LR2017024), the Fund for Dalian Distinguished Young Scholars
(under grant no. 2016RJ10), and the Liaoning Revitalization
Talents Program (under grant no. XLYC1807013).

9 References
[1] Wynn, R.B., Huvenne, V.A.I., Le Bas, T.P., et al.: ‘Autonomous underwater
vehicles: their past, present and future contributions to the advancement of
marine geoscience’, Mar. Geol., 2014, 352, pp. 451–468
[2] Karimi, H.R., Zapateiro, M., Luo, N.: ‘Adaptive synchronization of master–
slave systems with mixed neutral and discrete time-delays and nonlinear
perturbations’, Asian J. Control, 2012, 14, (1), pp. 251–257
Fig. 7 Comparisons on velocity responses [3] Karimi, H.R.: ‘Robust synchronization and fault detection of uncertain
master-slave systems with mixed time-varying delays and nonlinear
perturbations’, Int. J. Control Autom. Syst., 2011, 9, (4), pp. 671–680
[4] Na, J., Huang, Y., Wu, X., et al.: ‘Adaptive finite-time fuzzy control of
nonlinear active suspension systems with input delay’, IEEE Trans. Cybern.,
2019, pp. 1–12, doi: 10.1109/TCYB.2019.2894724
[5] Wang, N., Karimi, H.R.: ‘Successive waypoints tracking of an underactuated
surface vehicle’, IEEE Trans. Ind. Inf., 2019, p. 1–10, doi: 10.1109/
TII.2019.2922823
[6] Kim, J., Joe, H., Yu, S.: ‘Time-delay controller design for position control of
autonomous underwater vehicle under disturbances’, IEEE Trans. Ind.
Electron., 2016, 63, (2), pp. 1052–1061
[7] Liang, X., Qu, X., Wan, L., et al.: ‘Three-dimensional path following of an
underactuated AUV based on fuzzy backstepping sliding mode control’, Int.
J. Fuzzy Syst., 2017, 20, (2), pp. 640–649
[8] Myint, M., Yonemori, K., Lwin, K.N.: ‘Dual-eyes vision-based docking
system for autonomous underwater vehicle: an approach and experiments’, J.
Intell. Robot. Syst., 2018, 92, pp. 159–186
Fig. 8 Comparisons on control inputs
[9] Wang, F., Wan, L., Li, Y.: ‘A survey on development of motion control for
underactuated AUV’, Shipbuild. China, 2010, 51, (2), pp. 227–241
[10] Do, K.D., Pan, J., Jiang, Z.P.: ‘Robust and adaptive path following for
underactuated autonomous underwater vehicles’, Ocean Eng., 2004, 31, (16),
pp. 1967–1997
[11] Do, K.D., Pan, J.: ‘Practical control of underactuated ships’, Ocean Eng.,
2010, 37, (13), pp. 1111–1119
[12] Wang, N., Pan, X.: ‘Path-following of autonomous underactuated ships: a
translation-rotation cascade control approach’, IEEE/ASME Trans.
Mechatronics, 2019, pp. 1–10, doi: 10.1109/TMECH.2019.2932205
[13] Xu, J., Wang, M., Qiao, L.: ‘Dynamical sliding mode control for the trajectory
tracking of underactuated unmanned underwater vehicles’, Ocean Eng., 2015,
105, pp. 54–63
[14] Yan, Y., Yu, S.: ‘Sliding mode tracking control of autonomous underwater
vehicles with the effect of quantization’, Ocean Eng., 2018, 151, pp. 322–328
[15] Yildirim, S.: ‘Design of neural network control system for controlling
trajectory of autonomous underwater vehicles’, Int. J. Adv. Robot. Syst., 2014,
Fig. 9 Complex unknowns and fuzzy estimations 11, (7), pp. 1–17
[16] Lakhekar, G.V., Waghmare, L.M.: ‘Adaptive fuzzy exponential terminal
sliding mode controller design for nonlinear trajectory tracking control of
on control inputs. It is obvious to observe that the inputs are autonomous underwater vehicle’, Int. J. Dyn. Control, 2018, 6, (6), pp. 1690–
bounded and consistent with the velocity responses. However, the 1705
force and moments using AF-DSC have smaller chattering than [17] Wang, N., Sun, J.C., Er, M.J.: ‘Tracking-error-based universal adaptive fuzzy
control for output tracking of nonlinear systems with completely unknown
that using DSC, thereby contributing to less abrasion for AUV dynamics’, IEEE Trans. Fuzzy Syst., 2018, 26, (2), pp. 869–883
actuators and higher energy efficiency. Fig. 9 illustrates the [18] Londhe, P.S., Patre, B.M.: ‘Adaptive fuzzy sliding mode control for robust
complex unknowns and their fuzzy estimations, whose results trajectory tracking control of an autonomous underwater vehicle’, Intell. Serv.
show good approximation performance for model uncertainties and Robot., 2019, 12, pp. 87–102
[19] Elmokadem, T., Zribi, M., Youcef-Toumi, K.: ‘Trajectory tracking sliding
time-varying disturbances. mode control of underactuated AUVs’, Nonlinear Dyn., 2016, 84, (2), pp.
1079–1091
7 Conclusions [20] Elmokadem, T., Zribi, M., Youcef-Toumi, K.: ‘Terminal sliding mode control
for the trajectory tracking of underactuated autonomous underwater vehicles’,
This study addressed the three-dimensional trajectory tracking for Ocean Eng., 2017, 129, pp. 613–625
[21] Valenciaga, F.: ‘A second order sliding mode path following control for
an underactuated AUV under the complex unknowns including autonomous surface vessels’, Asian J. Control, 2014, 16, (5), pp. 1515–1521
model uncertainties and time-varying disturbances. By virtue of the [22] Bi, F.Y., Wei, Y.J., Zhang, J.: ‘Position-tracking control of underactuated
DSC technique, the design process of robust controllers is autonomous underwater vehicles in the presence of unknown ocean currents’,
simplified in comparison with the traditional backstepping. In IET Control Theory Applic., 2010, 4, (11), pp. 2369–2380
addition, an adaptive fuzzy logic approach is developed, which

IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370 369
© The Institution of Engineering and Technology 2019
 17519578, 2020, 5, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-its.2019.0347 by Readcube (Labtiva Inc.), Wiley Online Library on [09/07/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[23] Liang, X., Qu, X., Hou, Y., et al.: ‘Three-dimensional trajectory tracking [32] Wang, N., Er, M.J., Sun, J.: ‘Adaptive robust online constructive fuzzy control
control of an underactuated autonomous underwater vehicle based on ocean of a complex surface vehicle system’, IEEE Trans. Cybern., 2016, 46, (7), pp.
current observer’, Int. J. Adv. Robot. Syst., 2018, 15, (5), pp. 1–9 1511–1523
[24] Wang, N., Sun, Z., Jiao, Y., et al.: ‘Surge-heading guidance based finite-time [33] Jiang, B., Karimi, H.R., Kao, Y., et al.: ‘A novel robust fuzzy integral sliding
path-following of underactuated marine vehicles’, IEEE Trans. Veh. Technol., mode control for nonlinear semi-Markovian jump T–S fuzzy systems’, IEEE
2019, 68, pp. 8523–8532 Trans. Fuzzy Syst., 2018, 26, (6), pp. 3594–3604
[25] Xu, J., Wang, M., Qiao, L.: ‘Backstepping-based controller for three- [34] Karimi, H.R.: ‘A sliding mode approach to H∞ synchronization of master-
dimensional trajectory tracking of underactuated unmanned underwater slave time-delay systems with Markovian jumping parameters and nonlinear
vehicles’, Control Theory Appl., 2014, 31, (11), pp. 1589–1596 uncertainties’, J. Franklin Inst., 2012, 349, (4), pp. 1480–1496
[26] Mu, D., Wang, G., Fan, Y., et al.: ‘Fuzzy-based optimal adaptive line-of-sight [35] Jiang, B., Karimi, H.R., Kao, Y., et al.: ‘Adaptive control of nonlinear semi-
path following for underactuated unmanned surface vehicle with uncertainties Markovian jump T–S fuzzy systems with immeasurable premise variables via
and time-varying disturbances’, Math. Probl. Eng., 2018, 2018, pp. 1–12 sliding mode observer’, IEEE Trans. Cybern., 2018, 12, pp. 1–11
[27] Wang, N., Su, S., Pan, X.: ‘Yaw-guided trajectory tracking control of an [36] Patre, B.M., Londhe, P.S., Nagarale, R.M.: ‘Fuzzy sliding mode control for
asymmetric underactuated surface vehicle’, IEEE Trans. Ind. Inf., 2019, 15, spatial control of large nuclear reactor’, IEEE Trans. Nucl. Sci., 2015, 62, (5),
pp. 3502–3513 pp. 1–11
[28] Wang, N., Deng, Z.: ‘Finite-time fault estimator based fault-tolerance control [37] Ho, H.F., Wong, Y.K., Rad, A.B.: ‘Adaptive fuzzy sliding mode control with
for a surface vehicle with input saturations’, IEEE Trans. Ind. Inf., 2019, pp. chattering elimination for nonlinear SISO systems’, Simul. Modelling Pract.
1–9, doi: 10.1109/TII.2019.2930471 Theory, 2009, 17, (7), pp. 1199–1210
[29] Wang, N., Sun, Z., Yin, J., et al.: ‘Fuzzy unknown observer-based robust [38] Li, Y., Tong, S., Liu, Y.: ‘Robust adaptive fuzzy dynamic surface control of
adaptive path following control of underactuated surface vehicles subject to uncertain nonlinear for systems’, 2008 Chinese Control and Decision Conf.,
multiple unknowns’, Ocean Eng., 2019, 176, pp. 57–64 Yantai, China, July 2008, pp. 5019–5024
[30] Wang, N., Karimi, H.R., Li, H., et al.: ‘Accurate trajectory tracking of [39] Liang, X., Qu, X., Hou, Y., et al.: ‘Three-dimensional path following control
disturbed surface vehicles: a finite-time control approach’, IEEE/ASME of underactuated autonomous underwater vehicle based on damping
Trans. Mechatronics, 2019, 24, (3), pp. 1064–1074 backstepping’, Int. J. Adv. Robot. Syst., 2017, 14, (4), pp. 1–9
[31] Yu, J., Ma, Y., Yu, H.: ‘Adaptive fuzzy dynamic surface control for induction
motors with iron losses in electric vehicle drive systems via backstepping’,
Inf. Sci., 2017, 376, pp. 172–189

370 IET Intell. Transp. Syst., 2020, Vol. 14 Iss. 5, pp. 364-370
© The Institution of Engineering and Technology 2019