Available online at www.sciencedirect.

com

Journal of Ocean Engineering and Science 5 (2020) 101–115

www.elsevier.com/locate/joes

Gimbal orientation study and assessment for the autonomous underwater

vehicles
Sudhir Kumar Chaturvedi a, Sourav Basu b, Saikat Banerjee c,∗
a Department of Aerospace Engineering, UPES, Dehradun 248007, Uttarakhand, India
b Department of Electrical Engineering, Jalpaiguri Government Engineering College, Jalpaiguri 735102, West Bengal, India
c Department of UAV and Remote Sensing, Wingbotics, Kolkata-700086, West Bengal, India

Received 17 May 2019; received in revised form 31 August 2019; accepted 5 September 2019
Available online 19 September 2019

Abstract
Autonomous underwater vehicles are at present being used for scientific, commercial and military submerged applications. In this paper,
a system has been proposed which can be used underwater as remotely operated underwater vehicle for submerged survey for different
purposes. These systems require self-sufficient direction and control frameworks so as to perform submerged assignments. Displaying,
framework identification and control of these vehicles are as yet real dynamic zones of innovative work. This theory is worried about
the plan and improvement of an Autonomous Underwater Vehicle (AUV) specifically proposed for passage into global submerged vehicle
rivalries. The theory comprises of two stages; the first includes the plan and development of the vehicle while the subsequent stage is
worried about the demonstrating and framework identification of the vehicle, just as the reproduction of a control system. The structure and
advancement of the vehicle comprised of actualizing a mechanical and electrical framework, just as the reconciliation of subsystems. The
framework identification of the vehicle parameters comprised of utilizing locally available sensors to perform static and dynamic tests. Least
squares estimation was utilized to gauge the parameters from the pre-researched data obtained.
© 2019 Shanghai Jiaotong University. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license. (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Keywords: IMU sensor; Gimbal stabilization; Kalman filter; MATLAB Simulink; Brushless DC motor.

1. Introduction vehicles, remotely operated vehicles and autonomous under-

water vehicles. The vast majority of the contexts today that
“Underwater” usually used as a term to describe the natu- require a specific degree of accuracy and ability are worked
ral feature of the land underneath the shallow of water. This as Remotely Operated Vehicles (ROVs). Though, the agendas
realm might be ocean, sea, lake, or river. Considering that the that perform repetitive errands are organized as Autonomous
seventy per cent of the earth is enclosed by water, there are Underwater Vehicles (AUVs). ROVs are tethered underwater
still lots to discover. Although there are some underwater ac- vehicles. These vehicles are usually used in deep water ap-
tivities that can be achieved by humans, underwater medium plication such as offshore extraction of oil, and natural gases.
is a very hazardous environment and these activities cannot be ROVs are highly manoeuvrable and functioned by an individ-
performed for a long period of time. For that reason, research ual on the mother ship where communication is achieved via
efforts on underwater robotics have been increased and, un- tether [1]. This tether carries the electrical power and data
derwater vehicles (UVs) are started to be used for a wide from operator to the vehicle or vice versa. Since the data is
variety of tasks. The general name of these vehicles is Un- transferred via tether; ROVs can be controlled directly by the
inhabited Underwater Vehicles or Unmanned Underwater Ve- human operator. By this way, human decision is included in
hicles (UUVs). There are mainly two types of underwater the loop that enables the system to carry on precise operations
in undetermined environments. Alternatively, AUVs don’t re-
∗ Corresponding author. quire such tether connections. Nevertheless, most survey class
E-mail address: [email protected] (S.K. Chaturvedi). AUVs maintain acoustic communication with their human
https://doi.org/10.1016/j.joes.2019.09.003
2468-0133/© 2019 Shanghai Jiaotong University. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license.
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
102 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

Various underwater vehicles are built up to this date. Most

of these vehicles use thrusters for motion and steering. Usage
of thrusters increases costs and because they are exposed to
sea water, they have corrosion problems. Steering of the ve-
hicle and the stability under environmental effects are going
to be provided by internal moving masses. The volume of the
vehicle is going to be changed to control the depth. Internal
moving masses are used to relocate the centre of gravity of the
vehicle; thus, the orientation of the vehicle will be changed.
Change in the volume causes the change in the buoyancy and
it leads to motion in heave direction. The research in this
paper is focused on the design for internal moving masses
mechanism. Conceptual designs are proposed and appropri-
ate one is used for simulation tests to prove validity of the
control for hovering stability under the impact of underwater
hydrodynamic disturbance and steering [3].
Autonomous Underwater Vehicles: Autonomous underwa-
ter vehicles (AUVs) usually have only one thruster that is
located at the backside of the vehicle. A number of sensory
systems are located on the vehicle. These sensors are inertial
measurement units that contain linear acceleration and angu-
lar velocity sensors, pressure-meter to observe vehicle depth,
sonar systems; front sonar to measure the distance from obsta-
cles, ground speed sonar to measure the speed of the vehicle
relative to ground. Global Positioning System (GPS) is used
to learn the exact location of the vehicle at the sea surface.
The positioning through GPS can only be achieved when the
vehicle surfaced. AUVs may also consist of acoustic systems
that are used to learn the exact location of the vehicle in pre-
defined areas. Also, acoustic communication can be used for
data transfer between the base station and the vehicle. AUVs
Fig. 1. Flowchart of the UUV Design (Source: Aytar and Dede, 2009). are supposed to be entirely self-governing, consequently re-
lying on onboard power systems and intelligence.
operator, in order to send the data and status of the vehi- The mechanical autonomy world has arrived at a phase
cle. The limited bandwidth characteristic of acoustic commu- where the remotely operated vehicle (ROV) industry is very
nication makes the live streaming of the visual information entrenched with heaps of ROVs having been made and sent
and real-time control not possible. Both AUV and ROV sys- since the foundation of this industry. The requirement for au-
tems can be modified by the addition of a robotic manip- tonomy in robots and vehicles, be that as it may, is ending
ulator. These systems are called as Unmanned Vehicle Ma- up increasingly more a pervasive issue by and large and sit-
nipulator System (UVMS). It would be distinguished that by uations around the world. The capacity to convey between
this addition, dexterity of the vehicle is increased while the the administrator and the vehicle is one of the fundamental
control problems arise. Fig. 1 outlines the design flow of a components affecting whether a vehicle is to be structured as
UUV.However this methodology is also applicable for floating an ROV or as a self-sufficient vehicle. One of the conditions
structures like ship and submarines too. in which communication is especially obliged is submerged.
Complying with the design criteria, after the first two de- Submerged vehicles are assuming an imperative job in sub-
cisions on inclusion of a manipulator and configuring the sys- merged investigation and enabling people to investigate in-
tem as an AUV or an ROV, the design flow remains the same credible profundities. ROVs have been applied in the offshore
for all UUVs. The designer must investigate the working con- business since the late 1960s and are settled and concentrated
ditions of the system for possible additions of fault tolerance vehicles for deepwater missions. Notwithstanding, their im-
features. After the selection of sensory information, naviga- peratives, for example, the requirement for an interchanges
tion and localization equipment, dynamic prototypical of the tie and a control stage have constrained their utilization and
system is developed. The dynamic model also includes the abilities. this paper consists a technique for efficient adaptive
environment model with hydrodynamic effects. Later con- control of submerged robots with changing body weight and
trol architecture design is initiated. This design phase in- elements [4]. The body weight changes when the robot grabs
cludes simulation studies and the verification of the developed or drops off a payload. Our methodology joins balance control
controller and finally, the examination of the system through and buoyancy control. Specifically, we portray the design of
experimental studies [2]. two equipment modules: (1) a buoyancy control component
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 103

and (2) a balance control mechanism. The buoyancy control must determine the effects of drag in real-time. Therefore,
system gives a power-efficient intends to make up for ad- empirical models must be relied upon. While these models
ditional weight. It utilizes a cylinder to dislodge water. The cannot predict with great accuracy all the interactions of an
equalization control framework gives an efficient approach to unsteady, three-dimensional flow regime over a self-propelled
controlling the robot’s focal point of mass and by and large di- underwater body, they can capture the essential elements of
rection. It modifies the robot’s focal point of mass by moving flow and provide an effective tool to approximate the hydro-
an inside battery. Both of these frameworks spare significant dynamic forces that act on the body. The viscous drag can
vitality during the robot’s main goal and require power during be expressed by the following standard empirical model:
modification as it were. We additionally depict robot control 1
calculations that utilization buoyancy and equalization con- Fdrag = − ρACDu|u| (1)
2
trol to reaction to an adjustment in weight efficiently without
expanding the thruster power and previous researched data. 1 
τdrag = − ρA CDi ri3 ω|ω| (2)
2
2. Viscosity where ρ is the density of the surrounding fluid, A is the
matrix of projected cross-sectional areas of the body, CD is
Kirchhoff’s equations rely on the assumption that the fluid the drag coefficient of the body and ri is the mean distance to
is inviscid and slips freely over the surface of the body. In the centre of rotation. A body with three planes of symmetry
reality, viscosity has an appreciable effect on the flow regime. has a diagonal matrix of drag coefficients CD and so there
This plays a significant role in the dynamics of the system, is no coupling between the different degrees of freedom.
so much so that for bodies at large depths the contribution of The obligation of suitable standards for CD is essential to
potential damping is negligible in comparison. In a viscous realistically model drag.
fluid, a layer of fluid adheres to the solid boundary and
friction between the adjacent layers of fluid forms a thin 3. Basic gyro mechanical system
boundary layer over the surface. The relative flow velocity
adjusts rapidly from zero at the surface of the body to the A gyroscope is a device for measuring or maintaining ori-
velocity of an effectively potential flow just outside the entation. based on the principle of conservation of angular
boundary layer. The velocity gradient of the boundary layer momentum. The essence of the device is a spinning wheel on
gives rise to viscous stresses that act on the surface of an axle. The device, once spinning, tends to resist changes to
the body and retard its motion through skin friction. The its orientation due to the angular momentum of the wheel. In
boundary layer grows in the streamwise direction as the physics, this phenomenon is also known as gyroscopic inertia
flow is deflected away from the body to avoid accumulation. or rigidity in space.
For non-planar surfaces, a pressure gradient exists along the This proposed project’s mechanical frameworks comprise
boundary layer due to the centrifugal effect of flow around a of the vehicle structure, upper hull, actuators, and outside the
curved body. In the case of an adverse pressure gradient the enclosure. The upper hull and outside enclosure in areas are
flow decelerates as the pressure increases along the body. In a using for protecting the electronic parts and shielding them
sufficiently adverse pressure gradient, the fluid cannot diffuse from water, while the structure gives mounting focuses and in-
fast enough into the boundary layer and the flow separates surance to the majority of the sensors and walled in areas. The
from the body with the direction of flow over its surface re- edge defines the positions and directions of each mechanical
versed beyond the separation point. This forms a low-pressure segment in the vehicle, keeping up the basic trustworthiness
wake that gives rise to a retardant force on the body that and unbending nature of the vehicle and securing sensitive
is known as pressure drag. The effects of viscous flow are parts. This casing will go about as a phenomenal stage for
highly dependent on the type of flow regime. Since boundary some extreme changes in the general vehicle’s structure. The
layers are composed of both laminar and turbulent regions, casing accentuates simplicity of assembling and incorporation,
it is first necessary to determine the appropriate type of flow convenience, alternatives for modification, and in reverse sim-
regime [5]. Turbulent boundary layers generally exert more ilarity [6]. The vast majority of the segments on the vehicle
frictional drag on a surface than laminar boundary layers, but must be in a bad way straightforwardly to the vehicle, requir-
nevertheless often results in a smaller overall drag as they are ing no additional mounting highlights, which spares weight,
less prone to separation. An indicator of the dominant type lessens unpredictability, holds parts all the more safely, and
of flow regime is the Reynolds number Rn = ulν , where u is makes gathering and dismantling simpler. All parts were put
the velocity of the body, l is its characteristic length and ν with the end goal that the focal point of mass.
is the kinematic viscosity of the fluid. Whilst the technology Likewise, with air, underwater pressure is brought about
exists to analytically solve the Navier Stokes equations and by the weight of the medium, for this situation water,
determine the exact effects of the hydrodynamic interactions, following up on a surface. absolute signifying the complete
such calculations require sophisticated CFD packages and pressure and encompassing being of a relativistic sort. Adrift
powerful computers and even then, can take several days to level, weight because of air is 1atm. For each 10 m of pro-
solve. Therefore, with current technology, these equations fundity, weight increments by about 1atm and subsequently,
are not suitable for application onboard AUVs since they the total weight at 10 m submerged is 2atm. Albeit straight
104 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

in nature, the expansion in weight as profundity increments

is significant and submerged vehicles must be basically fit
for withstanding a moderately enormous measure of weight
on the off chance that they are to endure.
By performing different underwater moves of the vehicle
at full speed previously by other researchers, it was seen that
the vehicle was entirely steady, with roll and pitch develop-
ments being negligible. This was an immediate aftereffect of
the symmetrical idea of the vehicle, just as the inborn meta-
centric correcting minute that went about as an aloof roll
and pitch control framework. Since roll and pitch points are
unimportant, at that point, therefore, roll and pitch speeds can
likewise be viewed as irrelevant. Fig. 2. The orientation of the gimbal (Source: Basu Himali Banerjee and
To incorporate the dynamic of the gyros in the future mod- Chaturvedi 2019).
els of the system we need to have a model of the gyroscope
compatible with the rest of the models used. In this case, an
as the quaternions, but can lead to gimbal locks and are not
attempt to obtain a transfer function with angular velocity as
as efficient. A possible future work could be to develop a
input was done. From the transfer function, it is straightfor-
stabilization control based on quaternions. However, assuming
ward to get the state-space representation if required. For this
the real angles and angular velocities will not deviate much
particular case a second-order transfer function with cut-off
about the desired ones the behaviour of both systems is rather
frequency of 50 Hz was chosen.
similar.
ωr2 Basic coordinate frame transformations: For the 2-axis
TFGYRO = (3)
s2 + 2ζ ωr s + ωr2 gimbal problem three different coordinate frames are used.
The base or platform frame, the outer gimbal frame and the
where
inner gimbal frame. All frames are related by transformation
ωr : resonance frequency = 50Hz matrices, in this case, the sum of a translational and a ro-
ζ : damping factor = 0.7 tational matrix. Fig. 2 shows the gimbal structure with the
different coordinate frames axis depicted in different colours
4. Gimbal along with their angular relations [7].
The relation between the platform frame and the outer gim-
A gimbal, also called a gimbal ring, is a mechanical device bal frame is expressed by the sum of two matrices. One. Rθ P0 ,
consisting of two or more rings mounted on axes at right an- represents a rotational transformation and the other one, TPO ,
gles to each other. An object mounted on a three-ring gimbal represents a translational transformation.
⎡ ⎤
will remain horizontally suspended on a plane between the cosθPO sinθPO 0
rings regardless of the stability of the base. Gimbals have a RθP0 = ⎣−sinθPO cosθPO 0⎦ (4)
wide range of practical uses including aerospace applications. 0 0 1
A gimbal may be used to keep objects level in unstable en- ⎡ ⎤
vironments. Gimbals are also extremely valuable in shipboard 1 0 0
and aircraft environments when measuring instruments such TPO = ⎣0 1 0 ⎦ (5)
as chronometers and compasses must be kept level with the 0 0 Tpoz
horizon. Gimbals may also be used for aerospace navigation,
Coordinates in Outer Frame = [RθPO + Tpo]. Coordinates in
as they can be set to provide a stable measurement from a
Planiform Frame the same situation happens for the relations
specific reference point such as the earth or sun regardless as
between outer and inner gimbal frame where now Rθ PI rep-
to their actual position in space. In this specific application,
resents the rotational transformation and TOI represents the
our gimbal will hold the cameras and instruments that instead
translational transformation.
the compasses will not be kept in horizontal position but will ⎡ ⎤
point to a fix position in the inertial frame, or will perform cosθOI 0 −sinθOI
tracking of an object in that frame. Therefore, the selection RθOI = ⎣0 1 0 ⎦ (6)
of the gimbal configuration is of special importance. sinθOI 0 cosθOI
⎡ ⎤
1 0 0
4.1. Gimbal dynamics and kinematics
TOI = ⎣0 1 0 ⎦ (7)
0 0 TOIZ
Gimbals employed in aerospace navigation utilize Euler
angles to orient an object such as a spacecraft. This work Coordinates in Inner Frame = [RθPI + TOI ]. Coordinates
will also use Euler angles to develop the dynamics of the in Platform Frame because our interest is focused on angular
system. Euler angles are more intuitive, although not as robust velocities, angular accelerations and angles, we can drop the
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 105

Translational Matrix that has no influence on how the rate

vector of one coordinate frame relates with the other. The
outer frame rotates around the z-axis and the inner frame
around the y-axis so the angles θ PO and θ OI will not stay
fixed but will vary in time. The derivative of θ PO depends on
the different velocities between the platform coordinate frame
angular velocity (ωP ) and the outer coordinate frame angular
velocity (ωO ) around the z-axis. It is clear from here that
ωO cannot be solved explicitly v.1thout taking into account
the forces that produce the relative angular velocity between
both frames of reference. The same goes for the inner coor-
dinate frame angular velocity (ωI ), but in this case between
the outer coordinate frame angular velocity (ωO ) and the in-
ner coordinate frame angular velocity (ωO ) around the y-axis.
As mention above the derivative of θ PO represented by equa-
tions is
θ˙PO = ωOz (t ) − ωPz (t ) (8)
Creating the auxiliary matrix
⎡ ⎤ Fig. 3. Gimbal free body diagram for Inner axis reaction torques on the
0
external joint axis of rotation.
˙ PO = ⎣0 ⎦
 (9)
θ˙PO
The Euler’s moment equations are
We can get the relation between the platform frame and
the outer gimbal frame as M = h˙ I = h˙ B + ω ∗ h (16)
˙ PO
ωO (t ) = RθPO (t )ωP (t ) +  (10) Where M represents the applied moment and h is the angular
momentum. The subscript I express a derivate in the inertial
following an identical procedure frame and the subscript B a devote in the object frame. If the
θ˙OI = ωIy (t ) − ωOy (t ) (11) principal axes of inertia coincide with the coordinate frame,
which is our case, performing the vector product give us three
⎡ ⎤ scalar equations
0
˙ OI
 = ⎣θ˙OI ⎦ (12) MX = IX ω˙ X + ωY ωZ (Iz − IY ) (17)
0
MY = IY ω˙ Y + ωX ωZ (IX − Iz ) (18)
And the relation between the outer frame and the inner
gimbal frame is Mz = IZ ω˙ Z + ω˙ X wY (IY − IX ) (19)
˙ OI
ωI (t ) = RθOI (t )ωO (t ) +  (13) The following equations that will be used to develop the
Finally, we need to calculate the angular acceleration rate gimbal dynamics model follow closely the treatment given in
of the inner frame. These angular acceleration rates will be with major differences in the model of the friction.
used in the next section to develop the gimbal dynamics
4.3. Equations of motion
model. This model will be the basis for the rest of the work
and will be used to perform the simulations and to analyse
In this section, the gimbal equations of motion are summa-
different controllers. Therefore, differentiating Eq. (13) we
rized. They have been derived from the Euler moment equa-
get,
tions for general rigid body 6 degrees of freedom motion by
ω˙ I (t ) = RθOl (t )ω˙ o(t ) + RAUX ωo(t )θ˙OI + 
¨ OI (14) means of the application of the kinematic constraints to de-
fine the joint axes. Euler’s equation
M circumstances that the
Where,
⎡ ⎤ summation of the moments, , about a body is equal to
−sinθOI 0 → cosθOI the rate of change of its angular momentum, [H˙ ]InerT ial .
RAUX = ⎣0 0 0 ⎦ (15)
cosθOI 0 −sinθOI 
M
= [H˙ ]InerT ial = [H˙ ]local + [ ] × [H ] (20)
4.2. Euler’s moment equations The gimbal is broken up into two autonomous forms, Pan
and Tilt and is characterized by the free body diagram illustra-
The gimbal dynamics model can be derived from the tions presented in Figs. 3 and 4 Gimbal free body diagrams.
torque relationships about the inner and outer gimbal body
axes based on rigid body dynamics. TOz = TF riction + TDrive + [RI O
TI ]Z (21)
106 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

motions of the Outer/Pan axis results in the following:

⎡ ⎤ ⎡ ⎤
TOX IX ω˙ X + ωY ωZ (IZ − IY )
⎣TOY ⎦ = ⎣ IZ ω˙ Z + ωX ωY (IY − IX ) ⎦
TOZ 1
IZ
∗ { −ω X ω Y ( IY − IX ) + ( T F riction + TDrive )}
⎡ ⎤ ⎡ ⎤
TOX TI, OX
+ ⎣ TOY ⎦ + ⎣ TI, OY ⎦ (26)
− I1Y ∗ TOZ Gravity −1/IZ ∗ TI, OZ IO

TOx , TOy are reaction torques of the gimbal onto the base.
For the scope of this project, it is assumed that the inertia
of the base, or host aircraft, is much larger than the gim-
bal allowing us to ignore any base disturbances caused by
the gimbal’s reaction torques. The term TGravity represents the
mass imbalance torques of the gimbal due to the force of
gravity. To simplify the gimbal dynamics, it is assumed that
centre of gravity of the inner (tilt) axis lies on the inner axis
of rotation and that the centre of gravity of the outer axis lies
Fig. 4. Gimbal free body diagram in which TIx , TIz are torques exerted by on the outer axis of rotation [8–10]. This assumption requires
the inner/tilt axis onto the outer/pan axis. that the real gimbal system be balanced with counterweights.
Applying the CG constraint to the outer axis requires the in-
ner axis CG to lie not only on its axis of rotation but also
TIy = TF riction + TDrive (22) along the outer axis of rotation. This implies that these two
rotational axes intersect putting an additional constraint on the
Assuming alignment of both sets of body axes principal mechanical design. In carefully aligning the CG locations, the
inertia axes the gimbal moment equations can be written in torque induced from gravity can be cancelled out significantly
matrix form. Inner/Tilt axes: simplifying the dynamics and the control system complexity.
⎡ ⎤ ⎡ ⎤
 TIX TIX
4.4. Inner gimbal dynamics
MI = ⎣TIY ⎦ = ⎣TFriction + TDrive ⎦
TIZ TIZ
⎡ ⎤ ⎡ ⎤ We will begin analysing the dynamics of the inner gimbal.
IX ω˙ X + ωY ωZ (IZ IY ) TIX The sum of the kinematics torques about the inner gimbal is
= ⎣IY ω˙ Y + ωX ωZ (IX IZ ) ⎦ + ⎣TIY ⎦ (23)
IZ ω˙ Z + ωX ωY (IY IX ) TIZ Gravity MIx (t ) = TOIx (t ) − TGGlx (t ) (27)

Solving for the unknowns the equations of motion of the

Inner/Tilt axes results in the following: Mly (t ) = TeI (t ) − TGGIy (t ) − TfI (t ) (28)
⎡ ⎤ ⎡ ⎤
TIX IX ω˙ X + ωY ωZ (IZ − IY )
⎣ω˙ Y ⎦ = ⎣ 1 ∗ {−ωX ωz (IX − IZ ) + (TF ricT ion + TDrive )}⎦ MIz (t ) = TOIz (t ) − TGGIz (t ) (29)
IY
TIZ IZ ω˙ Z + ωX ωY (IY − IX ) With
⎡ ⎤
TIX TeI : Elevation stabilization control torque.
+ ⎣− I1Y ∗ TIY ⎦ (24) TfI : Friction torque about the y-axis
TIZ Gravity rGGIx , τ GGly , τ GGIz : Gravity gradient torques about each
gimbal axis.
Moment equations for the Outer/Pan axis written in matrix TOIx , TOIz : Torques exerted by cuter gimbal on inner gim-
form are shown below: bal
⎡ ⎤ ⎡ ⎤
 TOX IX ω˙ X + ωY ωZ (IZ IY ) The elevation stabilization torque TeI -produced by the mo-
MO = ⎣ TOY ⎦ = ⎣IY ω˙ Y + ωX ωZ (IX IZ )⎦ tor attached to the y-axis of the inner gimbal −consists of
TF ricT ion + TDrive IZ ω˙ Z + ωX ωY (IY IX ) two parts. One has the main function of cancelling the distur-
⎡ ⎤ ⎡ ⎤
TIX TI,OX bances and therefore, to nullify the angular velocity ωIy this
+ ⎣TIY ⎦ + ⎣TI,OY ⎦ (25) will be called TeIω and the other has to control the inner gim-
TIZ Gravity TI,OZ IO bal in order to move up to the desired angle, this part will be
called TeIθ . The gimbal-motor system has been built in a way
Note that the inner axis reaction torques are accounted for to avoid or minimize all nonlinear of behaviour of friction. It
in the [T]IO term. Solving for the unknowns the equation of has proved to be a realistic assumption to consider the friction
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 107

as the sum of a linear component, dependant on the relative

angular velocity between the inner and outer gimbal y-axis 3μ
Gy = (Iz − lx )sin (2θ )cos(∅ ) (36)
and a nonlinear component called coulomb friction. The lin- 2R03
ear component of the friction is proportional to the viscous 3μ  
friction coefficient Klf . The Coulomb friction has a constant Gz = Ix − Iy sin (2θ ) sin (∅ ) (37)
2R03
value, Ref and its direction depends on the sign of θ˙OI . This
constant can only be determined empirically [11–13]. These are the gravity gradient movement components of G.
The outer gimbal will exert torques around the x and z- The gravity moment vector G should be expressed in terms
axis of the inner gimbal. This torque will be necessary to of the angles of the inner gimbal system of reference. This
produce the same angular displacement experienced by the can be achieved by the body axes angular rates relative to
outer gimbal due to the fixed relationship between these axes. the reference frame together with knowledge of the initial
One important external moment is the gravitational mo- conditions of the Euler angles relative to the reference frame
ment. An asymmetric body subject to a gravitational field will [16,17].
experience a torque tending to align the axis of least inertia If we combine Eqs. (17)–(19) and (27)–(29) and replace
with the field direction. For the following development, we the general inertia matrices and angular velocities by the in-
assume that the gimbal is at a distance R0 from the Earth’s ertia matrices and the angular velocities of the inner frame,
centre of mass. The reference frame will be defined as fol- we get
lows: the origin of the reference frame moves with the centre  
TOIx (t ) = IIx ω˙ Ix + ωIy ωIz IIz − IIy + TGGIx (t ) (38)
of mass (cm) of the inner gimbal. The zR -axis points towards
the cm of the earth (the subscript “R” stands for reference).
The xR -axis is perpendicular to the zR axis in the direction of IIy ω˙ Iy = ωIx ωIz (IIx − IIz ) + Tel (t ) − TGGIy (t ) − TfI (t ) (39)
unmanned vehicles velocity. The YR axis is perpendicular to
 
both, the ZR and xR -axis. The aircraft’s axis frame is defined TOIz (t ) = IIz ω˙ Iz + wIx ωIy Ily − IIz + TGGIz (t ) (40)
by xB , yB and zB where each axis coincides with the inertia
axes (the subscript stands “B” denotes body) [14,15]. The first and third equation of having no practical use since
The Euler angles are defined as the rotational angles about the gimbal rotation around the x and z-axis are will depend
the body axes as follows: φ, about the xB -axis; φ, about the completely on the rotation of the outer gimbal. Therefore, the
yB -axis; and ψ, about the zB -axis, assuming an initial align- torques will be the necessary ones to accomplish this. This
ment between the reference and the body frame. leaves us with the second equation which contains the deriva-
The gravity gradient vector is defined as tive of one of the variables to be controlled (ωIy ). Expanding
 T the terms of the friction torques and the control torques we
G = Gx Gy Gz (30) get
The force exerted on a mass element due to gravity is Ily ω˙ Iy = −ωIz ωIx (IIx − IIz ) + TeIw (t ) + Telθ (t ) − TGGIy (t )

 
μdm − Kvf θ˙OI (t ) − Kcf sgn θ˙OI (t ) (41)
dF = − r (31)
|r |3 The goal is to control the elevation axis -inner gimbal y-
Where r = R + ρ is the distance from the earth’s centre of axis -and the cross-elevation axis-inner gimbal z-axis. There-
mass to the mass dm. Since p < R0 the moment about the fore, the variables to be controlled are ωIy and ωIz . Because
centre of mass of the body becomes we do not have direct control over the cross-elevation axis
μdm we have to do it indirectly through the azimuth axis -outer
dG = p∗dF = − ρ∗r (32) gimbal z-axis.
|r |3
Using the relations stated in Eqs. (6) and (14) we get,
where ρ is the radius vector from the body’s centre of mass
to a generic mass element dm. ωIx = cos θOI ωOx − sin θOI ωIz (42)
Because of p < < R0 . 1/r3 approximated as
 ωIz = sin θOI ωOx + cos θOI ωOz (43)
1 1 3R · ρ
≈ 1 − (33)
r3 R03 R02
ωOx = cosθPO ωBx + sinθPO ωBy (44)
Integration of Eq. (32) over the entire body of mass, to-
gether with Eq. (33), leads to
3μ ωOy = − sin θPO ωBx + cosθPO ωBy (45)
G= ∫ [R · ρ][ρxR]dm (34)
2R05 M The first three equations give us
After calculating the scalar and vector products we get the cos2 θOI ωOx − sinθOI ωIz +sin2 θOI ωOx ωOx − sinθOI ωIz
final results ωIx = =
cosθOI cos θOI
3μ   cosθPO ωBx + sinθPO ωBy − sinθOI ωIz
GX = 3
Iz − Iy sin (2φ)cos2 (θ ) (35) = (46)
2R0 cosθOI
108 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

We also know that In the same manner as before. the azimuth stabilization
torque Taz − produced by the motor attached to the z-axis of
θ˙OI = ωIy (t ) − ωOy (t ) (47)
the outer gimbal-consists of two parts: disturbance cancella-
Using the relations stated in Eqs. (46) and (47) it is possi- tion and pointing. Because stabilization is required about the
ble to represent Eq. (41) in terms of the controlled variables inner z-axis. called cross elevation axis, Taz is used to in-
and the base disturbances. directly control the angular velocity ωIz . This fact poses a
  difficult problem due to the high nonlinearities that appear in
IIy ω˙ Iy = −Kvf ωIy + Kvf cosθPO ωBy − sinθPO ωBx
  the dynamics of & the system [18].
ωIz cosθPO ωBx +sinθPO ωBy − sinθOI ωIz (IIx −IIz ) As before, the friction can be divided into viscous friction,
−
cos θOI proportional to θ˙BO and KOf . and the Coulomb friction, that
 has a constant magnitude equal to Kfc and whose direction
− Kcf sgn wIy − cosθPO ωBy − sinθP0 ωBx + TeIw (t )
depends on the sign of θ˙BO .
+ TeIθ (t ) − TGGIy (t ) (48)
The Outer and Inner torque vectors are defined as
Eq. (48) is useful to simulate and model the real system 
but for control purposes, it is more practical to use the infor- M0 = MOx , MOy , MOz (54)
mation of the angular velocities directly on outer gimbal axis

instead of getting the data from the base. A second way to M1 = MIx , MIy , MIz (55)
represent these dynamics is in terms of the controlled vari-
ables and the outer gimbal angular velocities as they will Applying the Euler’s moment equation to both vectors
be seen by the controller. Later on, it will be shown that the 
outer gimbal angular velocities will not be known exactly due Mo = h˙ + ω ∗ h = [Io ω˙ O + ωO ∗ Io ωO ] (56)
to sensor dynamics and noise measurements. Representing
Eq. (48) in terms of the outer gimbal angular velocities give 
us, [M1 ]0 = Rθ TOI MI = Rθ TOI h˙ I + ωI ∗ h = Rθ TOI [II ω˙ I + ωI ∗ II ωI ]
ωIz (ωOx − sinθOI ωIz )(Ilx − IIz ) (57)
IIy ω˙ Iy = −Kvf ωIy −
cos θOI Combining with Eq. (54) we can derive the rigid body
+ Kvf (ωOy ) − Kcf sgn[ωIy − ωOy ] + Telw (t) torque dynamics for the outer gimbal body as
+ Telθ (t ) − TGGly (t ) (49)
MOT = [IO ω˙ O + ωO ∗ IO ω0 ] + Rθ TOI [II ω˙ I + ωI ∗ II ωI ] (58)
It can be seen that using the outer gimbal angular veloci-
ties not only simplifies the model but reduce the number of From Eq. (58) and the symmetry property which states that
sensors used, Rθ −1
OI = Rθ OI we can expand the first term of the right hand,
T

4.5. Outer gimbals dynamics IO ω˙ O = I0 Rθ TOI ω˙ I − IO Rθ TOI RAUX ω˙ oθ˙OI − Io Rθ TOI 

¨ OI (59)

We use the Euler’s moment equations to get the rigid body With this result, Eq. (58) can be expressed as,
torque dynamics for the outer gimbal. The total torque vector 
about the outer gimbal axis is Io Rθ TOI + Rθ TOI II ω˙ I + {(ω0 ∗ IO ω0 ) + Rθ TOI (ωI x II ωI )
MO = MOT − [MI ]O (50) − IO Rθ TOI RAUX ω˙ O θ˙OI − IO Rθ TOT 
¨ OI = MOT (60)
Expanding this equation, we get the sum of the kinematics As we are interested in the cross-elevation axis we shall
torques about each axis of the outer gimbal consider only the third element of the vector shown in Eq.
MOx (t ) = TBOx (t ) − TGGOx (t ) − [MIx (t )]0 (51) (60). Solving for the first term of the left hand we have,
 
MOy (t ) = TBOy (t ) − TGGOy (t ) − Mly (t ) 0 (52) Io Rθ TOI + Rθ TOI II ω˙ I 3
⎡⎡ ⎤ ⎡ ⎤⎤
IOx cos θOI 0 −IOx sin θOI ω˙ Ix
MOz (t ) = Taz (t ) − TGGOz (t ) − TfO (t ) − [MIz (t )]0 (53) = ⎣⎣ 0 IOy 0 ⎦ ⎣ω˙ Iy ⎦⎦
[Mlx (t)]0 , [Mly (t)]0 , [MIz (t)]Q are the torque’s matrices of the IOz sin θOI 0 IOz cos θOI ω˙ Iz
⎡⎡ ⎤⎡ ⎤⎤ 3
inner gimbal referred to the outer gimbal frame. As in the IIx cosθOI 0 −IIz sinθOI ω˙ Ix
previous case we have, + ⎣ ⎣0 IIy 0 ⎦⎣ω˙ Iy ⎦⎦ (61)
IIx sinθOI 0 Ilz cosθOI ω˙ Iz 3
Taz : Azimuth stabilization control torque.
Tfo : Friction torque about the z-axis.
TGGOx , TGGOy , TGGOz : Gravity gradient torques about each
= (IOz sin θOI ω˙ Ix + IOz cosθOI ω˙ Iz ) + (IIx sin θOI ω˙ Ix + IIz cos θ0I ω˙ Iz )
outer gimbal axes.
TBOx , TBOy : Torques exerted by the base on outer gimbal. = os θOI (Ilz + IOZ )ω˙ Iz + sin θOI (IIx + IOz )ω˙ Ix (62)
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 109
⎡ ⎤
In this case []3 denotes the third element of the vector. As −Kv f /Iiy 
we can see below, the third element of the last term of the +⎣ 0 ⎦sgn ωiy −ωoy
right hand is equal to zero. 0
⎡⎡ ⎤⎡ ⎤⎡ ⎤⎤ ⎡ ⎤
IOx 0 0 cosθOI 0 −sinθOI 0 1/Iiy 

¨ OI = ⎣⎣0 IOy O ⎦⎣0
Io Rθ TOI  1 0 ⎦⎣θ¨OI ⎦⎦ + ⎣ 0 ⎦ Telω + Telθ − TGGly + Kvf ∗ ωoy
3
0 0 IOz sinθOI 0 cosθOI 0 0
⎡ ⎤ 3 ⎡ IIx −IIz ⎤
0 − Ily
= ⎣IOy θ¨OI ⎦ = 0 (63) ⎣
+ 0 ⎦ωIz (ωox − sin θoI ωIz )/cosθθI (69)
0 0
3
⎡ ⎤
After replacing the results obtained above for the cross-  x1
elevation axis dynamics equation we get ysensor = 0 1 0 ⎣x z ⎦ (70)
x3
cos θOI (IOz + IIz )ω˙ Iz
 The gimbal used for this project has a special property: its
= sin θOI (IOz + IIx )ω˙ Ix − (ωo ∗Iωo ) + Rθ TOI (ωI ∗ II ωI ) 3 axɛs of inertia are symmetrical. We can see that due to this

+ Io Rθ TOI RAU X ω0 θ˙OI = [MOT ]3
3
(64) symmetry the last term in the state equation will be cancelled.
Also due to this symmetry, the gravity gradient torque will
From Eq. (14) the angular acceleration about the inner x- disappear (see Eqs. (35)–(37)). Furthermost, this representa-
axis can be obtained. tion is useful for adding measurement noise at the output of
Substituting into 64, expanding the cross products terms the gyroscopes. simplifying the state equation and adding the
and substituting the kinematics torques leads to, measurement noise we obtain,
⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ 1 ⎤
Ir ω˙ Iz = sinθOI (Ioz + IIx )ω˙ Ox + (IOz ωIx − sin θOI IIx ω˙ Oz )θ˙OI x˙ 1 −Kvf /IIy O 0 xI IIy
⎣x˙ 2 ⎦ = ⎣0 0 1 ⎦⎣x 2 ⎦ + ⎣0 ⎦
− cosθOI [ωOx ωOy (IOz − IOx ) − sinθOI ωIy ωIz IIz
x˙ 3 2500 −70 −2500 x3 0
+ ωIy ωOx IIy − cosθOI ωIx ωIy IIx ] ⎡ −Kc f ⎤
 
+ cos θOI Taz − TGGz − KOv f θ˙OI − Kocf sgn θ˙OI  Iiy 
× Telω + Kv f ∗ ωOy +⎣ 0 ⎦sgn ωIy − ωOy (71)
(65)
0
This equation could be expanded further to represent the ⎡ ⎤
whole dynamics only in terms of the controlled variables and  x1
ysensor = 0 1 0 ⎣x2 ⎦ + e (72)
the base disturbances [19,20]. x3
Amplified inner gimbal dynamics: All variables are mea-
sured by sɛnsors which have their own dynamics plus noise Where e is the measurement noise of the sensor. This noise
added at the output. Therefore, the actual variables cannot be is assumed to be white noise with covariance Ecov .
known exactly. To deal with this, an augmented state-space e (t ) ∼ N (0, Ecov ) (73)
representation of the system is done. From Eq. (3) we know
It can be seen from the state-space equation that cancelling
that the dynamics of the sensors are described as
the nonlinear term would leave as with a near system and
ysensor 2500 therefore able to apply a controller for LTI systems. e.g. PID
TFsensor = = 2 (66)
ωm s + 70s + 2500 controller [21]. Two important obstacles are thɛ fact that the
Were the input ωm is the angular velocity to be measured. If paranιeler Kef is unknown and need to be estimated in real-
we use a state-space representation time, and as stated before that we do not have the precise
    values of the variables ωIy and ωOy . lt is clear the necessity
x˙1 0 1 x˙1 0
= + ω (67) to develop an observer to be able to apply negative feedback
x˙2 −2500 −70 x˙2 2500 to counteract the torque produced by this term. This observer
 will be developed in the following section [22].
 x˙1
ysensor = 1 0 (68)
x˙2
4.6. An overview of Kalman filtering
Now we are ready to combine Eqs. (67) and (49) to get
a state-space representation of the inner gimbal dynamics in- The Kalman filter is essentially a set of mathematical equa-
cluding the sensor dynamics. This representation of the com- tions that implement a predictor-corrector type estimator that
plete system will be used afterwards to develop the observers is optimal in the sense that it minimizes the estimated er-
and controllers for the system, ror covariance when some presumed conditions are met. Al-
⎡ ⎤ ⎡ ⎤⎡ ⎤ though not all the required conditions are met, because the
x˙1 −Kv f /Iiy 0 0 x1 system is only an approximation to the real nonlinear sys-
⎣x˙2 ⎦ = ⎣ 0 0 1 ⎦⎣x 2 ⎦ tem, the Kalman filter has proven to be flexible enough to be
x˙3 2500 −70 −2500 x3 implemented [23,24].
110 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

The Kalman filter estimates a process by using a form on the estimation of Xk−1 , comes straight from Eqs. (81) and
of feedback control: the filter estimates the process state at 82 Consequently,
some time and then obtains feedback in the form of (noisy)  
xˆ k|k−1 = f Xk−1|k−1 , uk−1 , 0 (83)
measurements. As such, the equations for the Kalman filter
fall into two groups: time update equations and measurement Before presenting the whole set of equations we need to
update equations. The time update equations are responsible define the Jacobian matrices of the partial derivatives of f and
for projecting forward (in time) the current state and error co- c with respect to X, w, and e. These matrices will be used
variance estimates to obtain the a priori estimates for the next later in a similar way to how matrices Ad and Cd where used
time step. The measurement update equations are responsible in the discrete Kalman filter.
for the feedback-i.e. for incorporating a new measurement ∂ f[i]  
into the a priori estimate to obtain an improved a posteriori Ak[i,j] = xˆ k−1|k−1 , uk−1 , 0 (84)
∂ X[j]
estimate [25]. A summary of the main equations is shown be-
low, the estimated error covariance to be minimized is defined ∂ f[i]  
Wk[i,j] = xˆ k−1|k−1 , uk−1 , 0 (85)
as, ∂ w [j ]
  T 
Pk = E xk − xˆ k xk − xˆ k (74) ∂ c[i]  
Ck[i,j] = xˆ k|k−1 , 0 (86)
The first step is the time update calculations. The equations ∂ X[j]
used are, ∂ e[i]  
Ek[i,j] = xˆ k|k−1 , 0 (87)
xk|k−1 = Ad xˆ k−1|k−1 + Bd uk (75) ∂ X[j]
We are now ready to define properly the lime update equa-
Pk|k−1 = Ad Pk−1|k−1 AdT + Gd Wcov Gdr (76) tions
 
For the measurement update, we need first to compute the xˆ k|k−1 = f Xk−1|k−1 , uk−1 , 0 (88)
Kalman gain and then to correct the estimated states and to
compute the new covariance matrix.
Pk|k−1 = Ak Pk−1|k−1 AkT + Wk Wcov WkT (89)
 −1
Kk = Pk|k−1 CTd Cd Pk|k−1 CTd + Ecov (77) and the measurement update equations
ek|k−1 = yk − yˆk|k−1 (78) Kk = Pk|k−1 CTd (Ck Pk|k−1 CTk + Ek Ecov EkT )−1 (90)
  
xˆ k|k = xˆ k|k−1 + Kk ek|k−1 (79) xˆ k|k = xˆ k|k−1 + Kk yk − c xˆ k|k−1 , uk , 0 (91)

  −1
 −1
Pk|k = Pk|k−1 − Pk|k−1 CTd Cd Pk|k−1 CTd + Ecov Cd Pk|k−1 Pk|k = Pk|k−1 − Pk|k−1 CTd Cd Pk|k−1 CTd + Ek Ecov EkT Cd Pk|k−1
(80) (92)
It is quite common for most applications to use the steady- The system in Eq. (71) was discretized considering the
state Kalman gain instead of the time-varying gain. unknown parameter Kc as a variable of the system and using
Extended Kalman filter: To overcome the few drawbacks the approximation
of the steady-state an alternative method is proposed. Instead
of using a linear approximation compatible with the tradi- ˙ = Xk − Xk−1
X (93)
T
tional Kalman filter the extended Kalman filter for the nonlin-
ear system will be developed. A Kalman filter that linearizes where T is the sampling period, For T small enough, this
about the current mean and covariance is referred to as an discretization method is almost equivalent to the zero-order-
extended Kalman filter. or EKF. In something alike to a Tay- hold method. The Jacobian matrices of this discretized system
lor series, we can linearize the estimation around the current are,
estimate using the partial derivatives of the process and mea- ⎡ ⎤
1 − TKvf /IIy −Tsign (x1k−1 − ωk−1 ) 0 0
surement functions to compute estimates even in the face of ⎢0 1 0 0 ⎥
non-linear relationships [26-27]. The nonlinear system will be Ak = ⎢⎣0
⎥
⎦
0 1 T
represented as Tωr 2
0 −T2ς ωr 1 − Tωr
2

Xk = f (Xk−1 , uk−1 , wk−1 ) (81) (94)

⎡ ⎤
yk = c ( Xk , ek ) (82) g11 0 0 0
⎢0 g22 0 0 ⎥
As same as with the discrete Kalman filter, the EKF pro- ⎢
Wk = ⎣ ⎥ (95)
cess is divided into two steps: time update and measurement 0 0 g33 0 ⎦
update. The time update equation, for estimation of Xk based 0 0 O g44
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 111

Fig. 5. Simulink model of three-axis gimbal circuit.

Ck = [0 0 1 0] (96) While IMUs comprise of accelerators and rate gyros, the

modelling performed is just reliant on the angular positioning
of the gimbal framework and is accordingly the main piece
Ek = 1 (97) of the inertial sensor that is modelled as shown in Fig. 5.
The parameters g11 , g22 , g33 , g44 will be used in the design And then an inertial measurement unit (IMU) sensor block is
process to tune how the noise w affect each of the states of attached to the circuit just with gravitational power capacity
the system. Using the matrices from Eqs. (94)–(97) and the of it just to detect the unseemly arrangement of the equip-
Eqs. (88)–(92), an S-function can be evaluated to perform the ment circuit and give the error input to the Arduino so the
Extended Kalman Filter algorithm [28-30]. servo system can keep up the axes according to prerequisite
[32,33]. Furthermore, that is the means by which the 3-axis
5. Gimbal and motor control camera gimbal circuit will be in working condition. The noisy
signals or the distortions will be eliminated through Kalman
The idea of driving the venture is to give the degree of filtering in the control unit which is pre-programmed with this
freedom (DOF) for three axes (x, y, and z) to the gimbal sys- feature of noise elimination. Also, when the structure of the
tem. What’s more, to get that sort of adaptability to the axes control framework for the camera gimbal is done, it will be
the circuit needs to get joined with moving components, for started in the equipment part to check whether its working
example, brushless DC motor to give the 360 degrees of free- or not. On the off chance that all the equipment segments
dom (DOF) for three axes (x, y, and z) to the gimbal system. and control procedure is fitting for one another then it will
Therefore three 3 dc motor would be executed to get a gimbal demonstrate the output in the screen for each engine for every
system. From the Figure it very well may comprehend that axis. What’s more, from the output qualities of the motors,
pin3(D), pin5(D), and pin6(D) are utilized for the information it will be straightforward the effectiveness, deferral and posi-
signal for the three motors of the gimbal, on the grounds that tion of motors. What’s more, it very well may be controlled
the engines of the circuit is intended to get fluctuating I/p according to the prerequisite of the application [34].
signals at consistently during the running method of gimbal
circuit, and digital pin 3, 5, 6 of Arduino ready to produce 5.1. Hardware attachments
pulse-width modulated (PWM) signals. The fundamental goal
is to assess the framework execution dependent on the sensor The dynamic adjustment of payload or equipment installed
selection; therefore, the sensor is modelled first [31]. in gimbal is creating for quite a while for capturing stable
112 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

Fig. 6. Circuit diagram of the experiment.

images or data stats for various purposes. These days it turns less over other dc motors. To get ideal proficiency brushless
out to be entirely adaptable to control the gimbal movement DC motors are utilized in this gimbal circuit. By picking up
remotely. The point of this analysis or venture is to make running and controlling a brushless DC engine structure the
another control framework to balance out the 3-axis gimbal investigations of past research papers, the basic components
camera position in MATLAB Simulink. are executed in the re-enactment model.
To get this control system, the boosted control hypothe-
sis for 3-axis developments is required with a smaller scale 6. Results and analysis
controller which is an Arduino UNO for this investigation,
and to plan a motor driving control rationale for 3 axis de- The most significant piece of the venture is to make the
velopment [35]. And, the separating of undesirable sign will control framework for the gimbal. Since the entire gyroscopic
be executed in the Arduino. From Fig. 6, the schematic dia- development will be constrained by the control technology
gram of the control system, a general thought can be gotten which will keep up the strength of the camera gimbal. The
from the trial. From the Figure it tends to be comprehend higher the effectiveness of the control strategy will be, the
that pin3(D), pin5(D), and pin6(D) are utilized for the in- deferral would be less for the framework. One other thing
formation signal for the three motors of the gimbal, on the ought to be as a primary concern that if the mechanical and
grounds that the motors of the circuit is intended to get fluc- other outer loses can be decreased or relieved, the productivity
tuating I/p signals at consistently during the running method of the camera gimbal would be better. And, that is the means
of gimbal circuit, and digital pin 3, 5, 6 of Arduino ready to by which the caught photographs or the recordings would
produce pulse-width modulated (PWM) signals [36]. An in- be better in quality. The control system has been explained
ertial measurement unit (motion sensor) (IMU) sensor block underneath with a square outline.
is attached to circuit just with gravitational power capacity of The full control procedure to make the gimbal circuit stable
it just to detect the unseemly arrangement of the equipment (appeared in Fig. 7), the information was given by the IMU
circuit and give the mistake criticism to the Arduino so the unit is first changed over into a precise dislodging parameter,
servo framework can keep up the axes according to necessity. and from that point forward, the change over information is
Brushless DC motors can create higher torque and unwa- contrasted and axes misalignment of the gimbal and the error
vering quality. Brushless DC engine can be synchronized with calibration feedback is gained. To get the ideal and profi-
the input signal immediately that is the reason the delay is cient output from the IMU sensor and Arduino UNO Kalman
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 113

Fig. 7. Control Strategy of the gimbal circuit.

filtering is required. In this paper 0.0001 to 0.001 s noise, co-

variant parameters’ qualities will be talked about. The appli-
cation on the microcontroller Arduino UNO depends on. For
a linear system, the state space condition can be composed
as
Xk+1 = AXk + BUk (98)
In Kalman filter, (98) will be converted into;
   
A 1 dt A dt
= + u (99)
B k+1 0 1 B k 0 k
Here u is the read from gyroscope and A, B are taken
subjectively just because of execution. The most essential is-
sue of this venture is the assessed error for the covariance
noise. This worth will tell us how much flimsiness is getting Fig. 8. MATLAB simulation with 0.0001 as measurement covariant noise.
by the sensor module. As the Arduino UNO has the broadly
useful ATmega microcontroller so the matrix activity needs
to change over into the basic numerical condition for the de-
crease of the multifaceted nature of estimation. Furthermore,
that is the manner by which the above conditions would be
changed over into rearranged structure given beneath
Ak+1 = Ak − Bdt + uk .dt (100)

Bk+1 = Bk (101)
And Eq. (100) can be written as,
Ak+1= Ak − (uk − Bk )dt (102)
On presentation, tests were taken as an offset the stan-
dard for each sensor while the framework was still and level Fig. 9. MATLAB simulation with 0.001 as measurement covariant noise.
to account static error no doubt in view of temperature af-
fectability, misalignment or diverse different angles for the of motors by the detecting range it has been fabricated with.
accelerometer yield speaking to inclination,
new − bias  = a. tan2 (ax , az ) (104)
1 =  prev + ∗ t (103)
s Here Refer to Eq. (104), is the foundation of the linear
Here “a” is the standard speeding up, a(bias) is the bal- relationship at the o/p side of the sensor. Ф is the roll rate of
anced increasing speed during initialization, and S is the sen- gyroscope, and del.t is the sampling rate. This is the calcu-
sitivity of the sensor. The IMU control the rotational course lation to get the mandatory information for each axes of the
114 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115

Fig. 10. Output curve of a single brushless DC motor(roll) of the gimbal.

Fig. 11. PID controller output for roll axis of a gimbal.

gimbal. And, the sensor had taken the most significant part topic to emphasize and develop for UUVs and AUVs for
to start these kinds of investigations. After all it is the sen- better surveillance or for any other purposes. That is why
sor for which the entire gyroscopic developments have been this project has been initiated for gimbal stability. And from
conceivable to execute as a team with the previously men- the simulation results, it can be easily understood that this
tioned conditions on the grounds that without the assistance process is applicable and efficient for further study in this
of sensors the camera gimbal strength is difficult to acquire. topic. Moreover, from the output curve or the movement of
The camera gimbal direction can be controlled in two unique the motor can be read from the simulation result at it can be
ways, for example, the controller (wi-fi module or Bluetooth) changed as per user’s requirement.
of the circuit can move the heading of the camera according
to necessity or the gimbal can naturally keep up its camera 7. Conclusions
position which will be given pre-customized in the MATLAB
reproduction. At the outcome segment of the recreation, just The undertaking work contained reports a couple of basic
x(roll) axis’ output signature plot has appeared in the output. issues in the field of gimbal adjustment. Another way to
At first the simulation results for the estimated angle from deal with inside steadiness research utilizing a gimbal has
the Kalman filter with 0.0001(Fig. 8) and 0.001(Fig. 9) as of appeared in this paper. The new technique comprises of
measurement covariance noise. a Simulink block and thusly gives us an approx. yield.
And then the output of one gimbal motor has been taken Assessment of the framework, after the investigations of
out to show the output curve (Fig. 10) of the system and how past research papers on the gimbal which depend on the
it acts as per the inclination of the hardware system and how inertial estimation unit information, and just gives us a
the curve varies with angular displacement. The output curve plan to the inward assessment of the gimbal’s security. The
can be obtained with more accuracy if the IMU sensor is used normal improvement of the inward recreation relies upon the
with multiple functions for the feedback to the system which self-ruling mode and gimbal quality. From the assistance of
is shown in Fig. 11. As the roll axis diagram is obtained after various past examines on the gimbal circuit and its strength,
receiving the feedback from a PID controller. the rationale of adjustment in a genuine domain is to set up
From the study of previous research papers on the topic, it the reachable adjustment precision. With the referenced test,
can be understood that the gimbal stability had been a primary the information extends from 2.56° and 1.97°, and a standard
 S.K. Chaturvedi, S. Basu and S. Banerjee / Journal of Ocean Engineering and Science 5 (2020) 101–115 115

deviation from 0.46° and 0.36° for pitch and move parameters [15] M. Quigley, M.A. Goodrich, S. Griffiths, A. Eldredge, R.W. Beard. Pro-
is gotten. The outside direction dependability is improved 3 ceedings of the IEEE International Conference on Robotics and Automa-
or multiple times in contrast with the exploratory tests, con- tion 2005 Apr 18 (pp. 2600–2605). IEEE.
[16] V.S. Doundakov, K.F. Young, inventors; Read-Rite Corp, assignee.
tingent upon the states of the independent submerged vehicle. Head gimbal assembly with a flexible printed circuit having a ser-
Considering the way that a gimbal is least spending device pentine substrate. United States patent US 6,249,404. 2001 Jun 19.
and that utilized innovation is quickly improving this target https:// patents.google.com/ patent/ US6249404B1/ en.
will become too soon. On the off chance that this kind of re- [17] J. Fang, H. Li, B. Han, IEEE Trans. Power Electron. 27 (11) (2011 Dec
6) 4630–4637.
cently gimbal controlling thought is started in the communal
[18] W.K. Zhongchao, Electr. Drive 2 (2005).
intelligence field then we can get a performing multiple tasks [19] J. Osborne, G. Hicks, R. Fuentes, IEEE Control Syst. Mag. 28 (4)
framework, that is by utilizing MATLAB, the controlling (2008) 44–64.
of the gimbal, depiction formulating and submerged control [20] Z.C. Ji, Y.X. Shen, J.G. Jiang, Acta Simul. Syst. Sin. 12 (2003)
agenda and so on can be set up with lesser inconveniences. 1745–1749.
[21] Z.C. Ji, Y.X. Shen, J.G. Jiang, Acta Simul. Syst. Sin. 12 (2003)
Furthermore, this could be introduced in different enormous
1745–1749.
and little extends from image or measurements taking to [22] H. Li, W. Li, H. Ren, IEEE Trans. Power Electron. 32 (3) (2016 May
autonomous submerged observation vehicles. The bore of this 17) 2452–2463.
examination has been under the correlation with the hit and [23] M. Michele, M. Giuseppe. Proceedings of the AIP Conference Proceed-
preliminary strategy which has been explained in the paper. ings 2015 Mar 10 (Vol. 1648, No. 1, p. 780011). AIP Publishing.
[24] H. Li, X. Ning, W. Li, ISA Trans 67 (2017 Mar 1) 348–355.
[25] L. Zhang, Y. Shen, G. Hu, L. Wei, J. Jia, L. Sheng, Microsyst. Technol.
References (2019) 1–8.
[26] S. Webb, J. Ellison, T. Firchau, D. Bloomfield, inventors; FREEFLY
[1] C. Kownacki, Digit. Signal Process. 21 (1) (2011 Jan 1) 131–140. SYSTEMS Inc, assignee. Apparatuses and methods for controlling a
[2] Y. Tsumaki, H. Naruse, D.N. Nenchev, M. Uchiyama. Proceedings of gimbal and other displacement systems. United States patent applica-
the IEEE International Conference on Robotics and Automation (Cat. tion US 14/214,637. 2014 Sep 18. https:// patents.google.com/ patent/
No. 98CH36146) 1998 May 20 (Vol. 3, pp. 2580–2585). IEEE. US20140267778A1/en.
[3] E.B. Aytar, M.İ. Dede. A Survey on Uninhabited Underwater Vehicles [27] H. Yuewei, L. Ming, T. Limei, Y. Jigang. Proceedings of the Twenty-
(UUV). ASME. Seventh Chinese Control and Decision Conference (2015 CCDC) 2015
[4] S.U. Lee, S. Kim. Proceedings of the IEEE/RSJ International Confer- May 23 (pp. 3393–3396). IEEE.
ence on Intelligent Robots and Systems 2006 Oct 9 (pp. 460–465). [28] D. Steyn. Variable Speed Scissored Pair Dual Gimbal Control Moment
IEEE. Gyro for Nano-Satellites (Doctoral dissertation, Stellenbosch: Stellen-
[5] G.J. Knowles, M. Mulvihill, K. Uchino, B. Shea, inventors; QorTek Inc, bosch University).
assignee. Solid state gimbal system. United States patent US 7,459,834. [29] P. Anju, U. Syamkumar, A. Gopinath, B. Jaya, M.N. Namboothiripad.
2008 Dec 2. https:// patents.google.com/ patent/ US20080106170A1/ en. Proceedings of the International Conference on Technological Advance-
[6] S.R. Vadali, S.R. Walker, H.S. Oh, J. Guid. Control Dyn. 13 (6) (1990 ments in Power and Energy (TAP Energy) 2015 Jun 24 (pp. 12–17).
Nov) 1090–1095. IEEE.
[7] S. Basu, S.B. Himali, S.K Chaturvedi, Int. J. Recent. Technol. Eng. 8 [30] M. Lu, S. Wang, W. Zheng, J. Liang, Y. Zhou, N. Xia. Proceedings
(1) (2019) 3049–3053. of the Twenty-Sixth Chinese Control and Decision Conference (2014
[8] T.H. Crocker, B.N. Levitt, inventors; Raytheon Co, assignee. Inertially CCDC) 2014 May 31 (pp. 61–66). IEEE.
stabilized gimbal platform. United States patent US 3,875,488. 1975 [31] H. Tisun, W. Xinsheng, L. Fucheng, X. Yongjiang, Comput. Meas. Con-
Apr 1. https:// patents.google.com/ patent/ US3875488A/ en. trol 2 (2009).
[9] J.S. Bennin, T. Boucher, J.W. Green, G.E. Gustafson, R. Jurgenson, B.D. [32] M. Lu, Y. Hu, Y. Wang, G. Li, D. Wu, J. Zhang. Proceedings of the
Lien, inventors; Hutchinson Technology Inc, assignee. Gimbal flexure IEEE International Conference on Advanced Intelligent Mechatronics
and electrical interconnect assembly. United States patent US 5,645,735. (AIM) 2015 Jul 7 (pp. 518–523). IEEE.
1997 Jul 8. https:// patents.google.com/ patent/ US5645735A/ en. [33] M. Lu, X. Li, Y. Li. Proceedings of the Fifteenth International Confer-
[10] B. Wie, D. Bailey, C. Heiberg, J. Guid. Control Dyn. 24 (5) (2001 Sep) ence on Electrical Machines and Systems (ICEMS) 2012 Oct 21 (pp.
865–872. 1–4). IEEE.
[11] N.S. Bedrossian, J. Paradiso, E.V. Bergmann, D. Rowell, J. Guid. Con- [34] G. Krishnan, K.T. Ajmal. Proceedings of the Annual International Con-
trol Dyn. 13 (6) (1990 Nov) 1083–1089. ference on Emerging Research Areas: Magnetics, Machines and Drives
[12] P. Greiff, B. Boxenhorn, T. King, L. Niles. Proceedings of the TRANS- (AICERA/iCMMD) 2014 Jul 24 (pp. 1–6). IEEE.
DUCERS’91: 1991 International Conference on Solid-State Sensors and [35] B. Kurkcu, C. Kasnakoglu. Proceedings of the Nineth International Con-
Actuators. Digest of Technical Papers 1991 Jun 24 (pp. 966–968). IEEE. ference on Electrical and Electronics Engineering (ELECO) 2015 Nov
[13] N.S. Bedrossian, J. Paradiso, E.V. Bergmann, D. Rowell, J. Guid. Con- 26 (pp. 796–800). IEEE.
trol Dyn. 13 (6) (1990 Nov) 1096–1101. [36] M. Chodnicki, P. Kordowski, M. Nowakowski, G. Kowaleczko, J.
[14] O. Jakobsen, E. Johnson, Infotech@ Aerospace, 2005 Sep, p. 7145. KONES 23 (4) (2016) 49–54.