Design of Robust Depth Controller for an AUV

with Disturbance
Ravishankar P. Desai and Narayan S. Manjarekar
2020 IEEE 17th India Council International Conference (INDICON) | 978-1-7281-6916-3/20/$31.00 ©2020 IEEE | DOI: 10.1109/INDICON49873.2020.9342326

Department of Electrical and Electronics Engineering

BITS Pilani, K.K.Birla Goa Campus
South Goa, INDIA.
email: [email protected], [email protected]

Abstract—A linearized decoupled depth plane model is used etc. control scheme is designed with and without time-
to formulate the control problem of an autonomous underwater varying/timeinvariant disturbance, with and without modelling
vehicle (AUV). This paper describes the design and simulation uncertainties. Therefore defined control scheme may vary as
of a depth plane control for a diving autopilot system for MAYA
AUV. An effective control strategy on pitch angle is inevitable to per user demand, shape and size of the vehicle, ocean envi-
achieve the specified depth of the vehicle. The resonant controller ronmental condition, desired motion and mission requirement
is designed by introducing a pole assignment technique to control and its multitude operation.
the pitch angle with disturbance. The proportional controller is The modelling of an underwater vehicle plays a very es-
used to command and regulate the depth of the vehicle. The sential role in precise control of the underwater vehicle. The
robustness analysis is done using the proposed control scheme
with achieved performance. The efficacy of the proposed control dynamics of the vehicle can be obtained mathematically based
scheme to track the sinusoidal reference input and reject the on computer-aided design (CAD), strip theory, computational
sinusoidal disturbance is reported with and without uncertainties fluid dynamics (CFD) and tow-tank test etc. The basic prin-
in depth plane model using the computational simulation tool. ciple of modelling is applicable for any shape and size of
Keywords—AUV, resonant controller, reference tracking, distur- the vehicle but their parameter may vary depending on the
bance rejection and robustness.
ocean environment. The modelling detail of the underwater
vehicle presented in [1] and a linearized decoupled model of
I. I NTRODUCTION
the underwater vehicle presented in [2]. The decoupled model
In marine technology, an AUV enhances its importance of REMUS AUV to control the depth is fully described with
when human safety is prime important. Nowadays AUV used its linear control scheme and results are presented through
for multipurpose and multitude operation such as commercial, simulation and experiment in [3]. The two separate actuation
military, research and exploration, and extra-curricular inter- system problem is addressed for depth control and controller
ests. An AUV is a robotic device i.e. operated through the sea is designed through complementary actuation approach in [4].
by the propulsion system, piloted and controlled by an onboard The steady-state error problem is considered in [5] depth
computer. The ocean vehicle invited attention because of its loop and proposed two solutions through modification in a
nonlinear dynamics, structural and nonstructural uncertainties, control structure and adding switching integrator to eliminate
the dependency of the model parameter on the environmental the steady-state error and improvement in transient response. A
condition and external disturbances in the operating environ- robust H2 optimal control method is designed and the simula-
ment. AUV design requirement consists, pre-programming is tion result is compared with P+PD controller in the presence of
essential as per mission need, for communication capability model uncertainties and output disturbance rejection in [6]. In
human intervention is inevitable with the requirement of [7] proposed improved active disturbance rejection controller
fool proof control, navigation and guidance (CNG) systems (ADRC) for depth control by implementing the hyperbolic
on board, to match the requirement of mission accuracy. It tangent function in an extended state observer (ESO) and
accomplishes the backup of every mission to mothership if any its convergence and stability performance is compared with
mission fails to produce the transmission data. In addition to conventional ADRC using gain adjustment function in ESO.
the accomplishment of AUV design, always there is a design The performance of the proposed controller does not depend
challenge and its composed of; system shape, its configura- on modelling accuracies is presented in [8] through adaptive
tion, dynamic equations, interactive (coupling) effects between fuzzy logic controller (FLC) coupled with phase lead compen-
the vehicle and other accessories namely manipulator, tether sator. It took minimum computation time to achieve the desired
etc., accurate positioning, trajectory tracking, development of depth of the vehicle is proposed in [9] through PD with FLC
control scheme, sensor integration, sensor fusion and their compare with PID controller. In the presence of hydrodynamic
availability with the cost. parameter uncertainties, the robust H∞ controller is designed
The depth motion control is one of the inevitable con- to track the set point and reject disturbances proposed in [10].
trol schemes in an underwater vehicle. Hence, for depth The free pitch angle diving behaviour problem of AUV is
control a linear, nonlinear, robust, adaptive and intelligent considered in [11] and adaptive neuro-fuzzy sliding mode

978-1-7281-6916-3/20/$31.00 © 2020 IEEE

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.
controller (ANFSMC) is proposed with model uncertainty
OE
and disturbance. The resonant controller is designed and Rudder

ϕ
verified experimentally for highly resonant damping modes XE ѱ
θ
YE
Fin

with robust to perturbations of the flexible manipulator in

ZE
[12]. The effectiveness of the resonant controller is proposed
in [13] for smart structure with robustness. The details about rg
O
design resonant controller can be found in [14]. In this work,
a linearized decoupled model of depth system of MAYA Rudder
Fin

AUV is considered and proposed a linear control technique

to achieve desired depth. The control scheme is designed Xo
Surge: u, X
Roll : p, K Yo
such that to achieve desired vehicle depth with rejecting Sway: v, Y
Pitch : q, M
Zo
sinusoidal disturbance and tracking sinusoidal reference signal. Heave: w, Z
Yaw : r, N

The controller structure is represented in cascade form and Fig. 1. AUV Schematic
each loop can be controlled independently for specific purpose.
Firstly, the resonant controller is designed for inner-loop called
a pitch control loop to provide a sufficient pitch angle in the
for communication. To linearized the depth system following
presence of sinusoidal disturbance for desired depth. The pitch
assumtions are considered for MAYA AUV [15].
loop controller utilizes the pitch angle measurement provided
by stern plane deflection and guarantees that the stability of 1) the forward vehicle velocity u0 is constant
closed-loop. The proportional controller is designed for precise 2) to hold the depth plane, the sway velocity should be zero
depth position of the vehicle. i.e. v = 0
The paper is briefed as follows. Section II represents mod- 3) the rotation about x-axis, angular velocity and roll angle
elling of a linearized depth plane model, section III covers the should be zero i.e. p = φ = 0
design of resonant controller and proportional controller and 4) the rotation about y-axis, vehicle angular velocity and
section IV presents the simulation result and its discussion. yaw angle should be zero i.e. r = ψ = 0
Finally, the conclusion is drawn in section V. The reduced EOM for depth system is given by,
• Surge
II. D EPTH P LANE M ODEL OF AN AUV
mu̇ + mqu = −(W − B) sin θ + CX u2 + CXu̇ u̇ + T (3)
The AUV has 6-degree of freedom (DOF) and manoeu-
vrable in three dimensions (3D). The AUV motion described- • Heave
with two co-ordinate frame called as body-fixed frame (X0 −
Y0 − Z0 ) and reference earth-fixed frame (XE − YE − ZE ) mẇ + mqw = (W − B)cosθ + CZẇ uw + CZq̇ uq
are dipicted AUV schematic as shown in Fig 1. +u2 CZδs δs + CZẇ ẇ + CXq̇ q̇ (4)
A 6-DOF nonlinear equation of motion (EOM) represented
as [1]: ż = −u sin θ + wcosθ (5)
η̇ = J(η)ν (1) • Pitch
M v̇ + C(v)v + D(v)v + G(η) = τ + τd (2) Iy q̇ = BzCB sin θ + CMw uw + CMq uq + u2 CMδs δs
∵ v is a linear and angular velocities, η is the position and +CMẇ ẇ + CMq̇ q̇ (6)
attitude, τd is the enviornmental disturbances, τ is the control
θ̇ = q (7)
input, G (η) is the gravitational/buoyancy forces and moments,
D(v) is the damping, C(v) is the coriolis and centripital where W is weight and B is buoyancy of the vehicle and
terms, M  is the mass. The highly coupled dynamics of AUV BZGB = −(ZG − ZB ) ∗ W . The linearized EOM for the surge
converted into lightly interacting subsystem i.e. speed, steering by neglecting interaction is given by,
and diving system [2]. The subsystem presented with its state
veriable as follows: (m − CXu̇ )u̇ = 2CX u0 u + T (8)
1) Speed system- u(t) The simplified rigid-body EOM in heave (w, Z) and pitch (q,
2) Steering system- v(t), r(t), ψ(t) M) [2] is given by,
3) Depth system- w(t), q(t), θ(t), and z(t)
In this paper depth subsystem is considered. The deflection of m(ẇ − u0 q) = Z (9)
stern planes causes, corresponding change in pitch movement Iy q̇ = M (10)
and angle of the vehicle results vertical force creates the pitch
of the vehicle that action causes submerge or rise of the The moment caused by the vertical distance between the centre
vehicle. For depth control AUV is to be able to submerge to of gravity ZG and buoyancy ZB is B̄Gz = ZG − ZB must
a specified depth and to ensure that again at periscopic height be modelled, distant from the external moments and forces

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.
described by linear damping, the stern plane deflection effects,
and hydrodynamic added mass, yeilds
Z = CZ ẇ ẇ + CZ q̇ q̇ + CZw u0 w + CZq u0 q + CZδs u20 δs (11)

M = CM ẇ ẇ + CM q̇ q̇ + CM q u0 q + CM δs u20 δs
+CM w u0 w − mg B̄Gz sin θ (12)
In steady state θ0 = q0 = φ0 = 0 suggests dropping ∆
notation in the following relations Fig. 2. A control structure of depth system

θ0 = q (13)
TABLE I
ż = −θu0 + w (14) L INEARIZED COEFFICIENT OF MAYA AUV

Parameter Iy CM q̇ CM q CM δs ¯ z
BG u0
Both simulation and ocean trials indicateds that the heave
Value +9.92 -18.46 -34.53 -23.51 -3.59 +1.50
velocity w during diving is negligible (< 0.05)m/s. Thus, we
neglect the term w and ẇ. Rearranging the expression from
the above equation (3) to (14) with preceding discussion into
state-space form C. Resonant Controller Design
To track the sinusoidal reference input and reject sinusoidal
      
q̇ A11 A12 0 q B11
 θ̇  =  1 0 0   θ  +  0  δs disturbance is often required in mechanical (e.g. robotics)
ż 0 −u0 0 z 0 closed-loop control system. The resonant controller has better
   capability to reject disturbances and zero steady-state error
1 0 0 q for sinusoidal waveform tracking. It also provides infinite
Y =  0 1 0  θ  (15) gain at a certain frequency with no gain and phase shift
0 0 1 z at other resonant frequency. For collocated systems resonant
CM q u0 CM δs u20 controller guarantees unconditional closed-loop stability. In
∵ A11 = Iy −CM q̇ ; A12 = − IyB̄Gzω
−CM q̇ ; B11 = Iy −CM q̇ this paper, we assume that sinusoidal signal with known pa-
III. D ESIGN OF C ONTROLLER rameter is θd (t) = Am sin(w0 t) and for disturbance rejection;
the sinusoidal disturbance signal is Di (t) = Dm sin(w0 t)
A. Problem Statement
with unknown magnitude Dm and known frequency w0 is
The design problem stated as a linearized dynamic model considered. The bandwidth (BW) of the closed-loop control
of AUV depth system is used to capture vehicle peculiarities system plays an important role as per disturbance rejection
such as control synthesis to achieve control objectives as sta- and noise suppression point of view. Because for slow varying
bilization, desired trajectory tracking and disturbance rejection disturbances will take longer time to response hence need to
using linear control techniques. decrease the BW whereas for high measurement noise (for
B. Structure of Controller the low-quality sensor) require to increase the BW. Hence,
for good tracking performance the complementary sensitivity
The diving system consists of three state variable pitch
function should be |T (jw)| = 1 at w = w0 and to reject
angle (θ), pitch angular velocity (q), and depth position (z)
disturbance the sensitivity function should be |S(jw)| = 0
and two control loops as shown in Fig. 2. called pitch and
at w = w0 . Considering the sinusoidal reference signal to
depth control system. These two loops are controlled using
be tracked and for fixed frequency w0 the disturbance to be
a resonant controller and proportional controller respectively.
rejected. The fundamental control action usually performed by
The desired depth Zd can be achieved by regulating the (θ)
resonant controller is
from the resonant controller and providing commanded signal
i.e (θ) from the proportional controller to resonant controller, Am w 0
θd (s) = (18)
based on the difference between the desired depth and actual s2 + w02
depth i.e. eθ = Zd − Z. The Table 1. presents the linearized ∵ wo is the frequency (must be reject and/or followed)
coefficient of the MAYA AUV [15]. The open-loop transfer Considering the sinusoidal disturbance signal to be rejected
function of the pitch control loop and depth control loop of and for unknown magnitude Dm and fixed frequency w0
an AUV referring to Fig. 2, using eq. (15) and Table 1. can the reference signal to be tracked. The sinusoidal disturbance
be written as: signal is
−1.8635 Dm ω0
Gθ (s) = 2 (16) Di (s) = 2 (19)
s + 1.8247s + 0.1266 s + ω02
−1.5
Gz (s) = (17) Considering the block diagram depicted in Fig.2 for pitch
s control loop of the vehicle the linear time-invariant (LTI)

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.
causal plant which is described (the dynamic of pitch model leads to,
of an AUV) in the linear time domain is given by, Acl = Adcl (26)
θ(s) Gθ (s) then, from eq. (26) a linear equation by comparing the
= (20)
θd (s) 1 + Gθ (s)Gd (s)Gc (s) coefficient is written as,
∵ Gθ (s) is the pitch loop transfer function, Gc (s) is the pitch a1 + l0 = Ad3 : s3
loop controller, Gd (s) is the delay of computation device.
a1 l0 + w02 + bC3 = Ad2 : s2
where input is the stern plane deflection and the output is
the desired pitch angle. a1 w02 + w02 l0 + bC3 α1 = Ad1 : s1
1) Controller Implementation: In the design of the resonant a1 w02 l0 + bC3 α0 = Ad0 : s0 (27)
controller, two real poles on the real axis at the sinusoidal
component (harmonics) to track and/or reject is considered. 2) Steady State Error Analysis: For steady-state error, the
The second-order transfer function that is used to describe the closed-loop system must be stable and satisfy limt→∞ eθ (t) =
dynamics of pitch control loop with the form, 0. The pitch model Gθ (s) of an AUV is controlled by a unity
feedback system by an LTI causal controller is,
b b
Gθ (s) = = (21) δs (s) = Cin (s)eθ (s) (28)
s2 + a1 s + a0 (s + a1 )(s + a0 )
Here, assuming a0 > a1 . The following resonant control where the pitch control loop error signal is,
structure [15] is chosen from eq. (21) for the second-order
eθ (s) = θd (s) − θ(s) (29)
system is,
C3 s3 + C2 s2 + C1 s1 + C0 The output signal is,
Cin (s) = (22)
(s2 + w02 )(s + l0 ) C3 s3 + C2 s2 + C1 s1 + C0
b
θ(s) = eθ (s)
where C3 to C0 are controller gain. The eq. (22) is re-written s2
+ a1 s + a0 (s2 + w02 )(s + l0 )
(30)
by implementing the pole-zero cancellation technique is,
Re-arranging eq. (29) as
C3 (s2 + α1 s1 + α0 )(s + a0 )
Cin (s) = (23) θd (s) = θ(s) + eθ (s) (31)
(s2 + w02 )(s + l0 )
Here, to make the analytical solution simple a pole-zero then from eq. (30)
cancellation technique is used. The cancellation is based on b(C3 s3 + C2 s2 + C1 s1 + C0 )
the pole dynamic response and it will be rejected through the θd (s) = [1 + ]eθ (s) (32)
(s2 + a1 s + a0 )(s2 + w02 )(s + l0 )
resonant controller when it reappears as input disturbance in
the closed-loop system. and using eq. (18) the eq. (29) become,
From eq. (21) and eq. (23) the open-loop transfer function Am w0 (s2 + a1 s + a0 )(s + l0 )
is written as, eθ (s) = (33)
(s2 + w02 )(s2 + a1 s + a0 )(s + l0 ) + x(s)
b(C3 (s2 + α1 s1 + α0 ))
Gθ (s)Cin (s) = (24) ∵ x(s) = b(C3 s3 + C2 s2 + C1 s1 + C0 )
(s + a1 )(s2 + w02 )(s + l0 ) To check the steady-state tracking error applying final value
The closed-loop transfer function with pole-zero cancellation theorem to eq.(33), we obtain
is
lim eθ (t) = lim seθ (s) = 0 (34)
θ(s) bC3 (s2 + α1 s1 + α0 ) t→∞ s→0
=
θd (s) bC3 (s + α1 s + α0 ) + (s + a1 )(s2 + w02 )(s + l0 )
2 1 3) Disturbance Analysis: The transfer function between
(25) disturbance acting on pitch loop and the output of the pitch
The pole placement technique is used to design the controller loop is written as
to reject the sinusoidal disturbance with its frequency of w0 .
θ(s) Gθ (s)
The selection of desired closed-loop pole is based on the basis = (35)
of control effort/energy, a time constant to avoid overshoot Di (s) 1 + Gθ (s)Cin (s)
and speed of closed-loop response i.e. reference tracking, Using eq. (19) the eq. (35) is
robustness, noise attenuation and disturbance rejection.
Dm w0 b(s + l0 )
The actual closed-loop characteristic polynomial from eq. θ(s) = (36)
(25) is (s2 + w02 )(s2 + a1 s + a0 )(s + l0 ) + x(s)
∵ x(s) = b(C3 s3 + C2 s2 + C1 s1 + C0 )
Acl = bC3 (s2 + α1 s1 + α0 ) + (s + a1 )(s2 + w02 )(s + l0 )
To check the steady-state tracking error applying final value
and desired closed-loop characteristic polynomial theorem to eq.(36), we obtain

Adcl = s4 + s3 Ad3 + s2 Ad2 + s1 Ad1 + Ad0 lim θ(t) = lim sθ(s) = 0

t→∞ s→0
(37)

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.
 TABLE II
L INEARIZED COEFFICIENT OF MAYA AUV

Pitch Loop C3 α1 α0 l0 w0
(Cin ) -0.1936 -4.1583 -0.2130 1.9277 1
Depth Loop Kp - - - -
(Cout ) +0.07 - - - -

Fig. 3. Analysis of depth model

D. Proportional Controller Design

One of the simplest controllers is the proportional controller
mainly used when we have little information about the system
and there is no demand for steady-state operation. The root
locus method is used to find the proportional gain (Kp ). The Fig. 4. Zd = 1 sin(t), Dp = 2 sin(t), Dd = 1 sin(t)
closed-loop system behaviour is dependent on Kp because as
Kp varies its effect on the pole-zero location of the system
changes. From Fig. (2) the control signal for a proportional
controller is written as sinusoidal disturbance input Di = Dm sin(wo t) = Dm sin(t)
with variation in magnitude of pitch and depth loop for
U (t) = Kp (Zd − Z) = Kp ez (t) (38) disturbance rejection. From Fig. (4) to Fig. (7) shows that
the proposed controller reject the disturbance and track the
The open-loop transfer function is
desired trajectory i.e. Zd = Z with the same frequency
Kp Gθ (s)Cin (s)Gz (s) w0 . An almost zero pitch and depth loop error when pitch
G(s)Cout (s) = (39)
1 + Gθ (s)Cin (s) loop disturbance magnitude greater and less than depth loop
disturbance magnitude but it took high pitch loop control effort
IV. S IMULATION R ESULT
observed in Fig. (4) compare to Fig. (5). Whereas an almost
The effectiveness and efficacy of the proposed control zero pitch loop error when depth loop disturbance magnitude
scheme for sinusoidal trajectory tracking with sinusoidal dis- greater and less than pitch loop disturbance magnitude but
turbance under robustness are discussed in this section. The it took nonzero depth loop error and acceptable pitch loop
simulation observation was carried out for decoupled depth control effort observed in Fig. (6) and Fig. (7). Hence, we
plane model of MAYA AUV with a proposed controller using can say that the proposed control strategy is effective and
computationaltool MATLAB R2019a (9.6.01072779). implementable on pitch angle to achieve the specified depth of
1) Analysis of Open Loop System: Using depth loop vehicle for sinusoidal reference tracking. In Table. 3. the per-
transfer function from eq. (16) and eq. (17) the resultant formance criteria for sinusoidal trajectory tracking and various
eigenvalues of linearized system are obtained as real poles settings of sinusoidal disturbances on pitch and depth loops
(P = 0, −0.0722, −1.7525) with one pole at origin and other are tabulated. It confirms that by introducing high magnitude
two are stable pole. The pole at origin corresponds to the pure disturbance in pitch loop and low magnitude disturbance in
integrator from pitch angular velocity q and pitch angle θ. The depth loop and vice versa, the proposed control strategy for
linearized system also fully state controllable and observable. Table. 3. is efficient and effective to reduce the pitch rate zero.
Fig. 3 shows the pole-zero plot, root locus, open-loop and close
loop response of pitch and depth loop for sinusoidal referance
input sin(t).
2) Analysis of Close Loop System: The control structure
of a depth plane system consists of pitch and depth loop
controller. To verify the proposed control scheme performance
the initial conditions of MAYA AUV are considered as [Roll
Velocity (p), Yaw Angle(θ), Heave Position(z)]T = [0, 0, 0]T
and linear velocities are [Surge(u), Sway(v), Heave(w)]T =
[1.5, 0, 0]T with assuming worst-case 1.5sec delay from depth
sensor and the desired close loop poles at Adcl = (s + 4)4 .
The controller setting parameters are tabulated in Table.2. The
1
sinusoidal referance input Zd = Am sin(wo t) = sin(t) is Fig. 5. Zd = 1 sin(t), Dp = sin(t), Dd = 1 sin(t)
2
applied to closed-loop depth plane control system by injecting

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.
 1 Fig. 8. Robustness analysis of depth plane model for case 1
Fig. 6. Zd = 1 sin(t), Dp = sin(t), Dd = 2 sin(t)
2

1 1 Fig. 9. Robustness analysis of depth plane model for case 2

Fig. 7. Zd = 1 sin(t), Dp = sin(t), Dd = sin(t)
2 2

3) Analysis of Robustness: The robustness of proposed C ONCLUSION

controller is evaluated against model uncertainty by: In this work, a linearized decoupled model of depth sys-
• Condition 1:Inserting +15% uncertainty in fin lift moment tem of MAYA AUV is considered. The proposed resonant+
CMδs , -20% in added mass term CMq̇ and hydrostatic proportional controller is designed to achieve the desired
moment CMθ with +10% in velocity to its nominal values vehicle depth by rejecting sinusoidal disturbance and tracking
of depth plane model. Then the transfer function is a sinusoidal reference signal. The efficacy and effectiveness of
the proposed controller are analysed and demonstrated through
−2.2945
Gθ (s) = (40) simulation. The following strength of the proposed controller
s2
+ 2.7689s + 0.1120
is obtained through simulation are:
• Condition 2:Inserting -15% uncertainty in fin lift moment
CMδs , +20% in added mass term CMq̇ and hydrostatic
moment CMθ with -10% in velocity to its nominal values
of depth plane model. Then the transfer function is
−1.1353
Gθ (s) = (41)
s2
+ 1.1625s + 0.1455
Fig. 8 to Fig. 11 shows the robustness analysis of depth plane
system for eq. (40) and eq. (41) with Zd = 1 sin(t) and
variation in PLD and DLD magnitude using Table. 3 con-
firms proposed controller is robust againt model uncertainty
including performance criteria. Fig. 10. Robustness analysis of depth plane model for case 3

TABLE III
P ERFORMANCE CRITERIA OF SINUSIODAL TRAJECTORY TRACKING
CONTROL

Case PLD DLD RMS ISE IAE

Case 1 2 sin(t) 1 sin(t) 1.5040 226.3 85.11
Case 2 0.5 sin(t) 1 sin(t) 0.3761 14.14 21.28
Case 3 0.5 sin(t) 0.5 sin(t) 0.5083 25.84 40.90
Case 4 0.5 sin(t) 2 sin(t) 0.8124 66.00 70.03
Case 5 0.5 sin(t) 1 sin(t) 0.3314 10.98 16.84
Case 6 0.5 sin(t) 1 sin(t) 0.3338 11.14 22.09
*PLD and DLD =Pitch and Depth Loop Disturbance
Fig. 11. Robustness analysis of depth plane model for case 4

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.
 1) Effective for closed-loop sinusiodal trajectory tracking
and disturbance rejection with and without uncertainties.
2) It can handle aworst-case computation delay.
3) The design structure is simple, cost-effective and easily
implementable.
The proposed controller confirms through simulation result
that its effectiveness in term of steady-state tracking error,
disturbance rejection and robustness with performance criteria
for sinusoidal trajectory tracking.
R EFERENCES
[1] T. I. Fossen and K. Y. Pettersen, “Guidance and control of ocean vehicles,”
John Wiley and Sons, 1994.
[2] B. Jalving, “The NDRE-AUV flight control system,” IEEE J. of Ocean
Eng., vol. 19, 1994, pp. 497-501.
[3] T. Prestero, “Verification of a six-Degree of freedom simulation model
for the REMUS autonomous underwater vehicle,” M.Sc. Thesis in Ocean
and Mechanical Engineering, MIT, US, 2001.
[4] A. Malerba and G. Indiveri, “Complementary control of the depth of an
underwater robot,” In Proceedings of the 19th IFAC World Congress,
2014, pp. 8971-8976.
[5] S.P. Hsu and T.S. Liu, “Modifications of control loop to improve the depth
response of autonomous underwater vehicles,” J. Math. Probl. Eng., vol.
2014, pp. 1-12.
[6] L. Qiao, S. Ruan, G. Zhang and W. Zhang, “Robust H2 optimal depth
control of an autonomous underwater vehicle with output disturbances
and time delay,” Ocean Eng., vol. 165, 2018, pp. 399-409.
[7] J. Li, X. Zhao and Y. Chen, “Active disturbance rejection controller for
depth-pitch control of an underwater vehicle,” International Journal of
Innovative computing, information and control, vol.13, no.3, 2017, pp
727-739.
[8] A. Nag, S. Patel and S. Akbar, “Fuzzy logic based depth control of
an autonomous underwater vehicle,” International multi-conference on
automation, computing, communication, control and compressed sensing
(iMac4s), 2013, pp. 117-123.
[9] P. Gupta and R. Gupta, “Depth control technique for an autonomous
underwater vehicle system,” International Conference on Communication
and Signal Processing (ICCSP), Melmaruvathur, 2016, pp. 19-22.
[10] A. Nag, S. S. Patel, K. Kishore and S. A. Akbar, “A robust H∞
based depth control of an autonomous underwater vehicle,” International
Conference on Advanced Electronic Systems (ICAES), Pilani, 2013, pp.
68-73.
[11] G. V. Lakhekar, L. M. Waghmare, P. G. Jadhav and R. G. Roy,
“Robust diving motion control of an autonomous underwater vehicle
using adaptive neuro-fuzzy sliding mode technique,” in IEEE Access,
vol. 8, 2020, pp. 109891-109904.
[12] I. A. Mahmood, S. O. Reza Moheimani and B. Bhikkaji, “Precise
tip positioning of a flexible manipulator using resonant control,” in
IEEE/ASME Tran. on Mechatronics, vol. 13, no. 2, 2008, pp. 180-186.
[13] H. Pota, S. R. Moheimani and M. Smith, “Resonant controller for smart
structures,” Smart Mater. Struct., vol. 11, 2002, pp. 1-8.
[14] L. Wang, “PID control system design and automatic tuning using
MATLAB/Simulink,” Wiley-IEEE Press, 2020.
[15] R. Kumar, “Control system design for speed control and diving autopilot
for MAYA AUV,” A Project Report, July 2006.

Authorized licensed use limited to: Carleton University. Downloaded on June 17,2021 at 01:21:05 UTC from IEEE Xplore. Restrictions apply.