Development of an Autonomous

Underwater Profiler for Coastal Areas

João M. Monteiro∗ , Nuno A. Cruz∗†
∗ Faculty of Engineering, University of Porto, Portugal
† INESC TEC, Porto, Portugal

[email protected], [email protected]

Abstract—One of the most common ways of collecting ocean Section VI describes the syntax of the available maneuvers
data is to deploy sensors from the surface, allowing to understand for the operation of the vehicle during a mission, and Section
the variation of water properties with depth. Autonomous vertical VII provides the first results from our experimental validation.
profilers are robotic vehicles that replace human operators in this
task. They form a particular class of autonomous underwater Lastly, Section VIII presents the concluding remarks and
vehicles that move predominantly along the vertical axis, typically future research directions.
with reduced control on the horizontal axis. This paper describes
a propeller driven autonomous underwater profiler, optimized
II. R ELATED W ORK
for shallow waters. The vehicle has no fins or other control Vertical profiles are important for studying the variation
surfaces, and uses four independent thrusters to provide both of water properties with depth, for instance to obtain the
vertical and horizontal motion, including hovering in the water temperature-depth relationship typically used in oceanography.
column. The paper describes the main subsystems, including the
hardware implementation, the software structure, and the motion Autonomous profilers have been developed to replace human
controllers, with experimental data from the first trials. operators in obtaining these data, and they have been designed
Index Terms—Autonomous Underwater Vehicles, Autonomous both for deep waters [1], and also for shallow depths [2].
Underwater Profilers, Vertical Profilers. Although their main operation mode is profiling in the vertical
direction, they have also been developed with the ability to
I. I NTRODUCTION maintain a fixed distance above seabed [3]. These operational
Monitoring the oceans is critical to understand our planet characteristics are directly related to some of the design
and data on water properties need to be gathered to better aspects of existing solutions, mainly in terms of propulsion
understand oceans phenomena. One of the ways of collecting systems and motion control.
these data is through vertical profiles of the water column that A. Propulsion
allow to understand variation of water properties with depth. To drive a profiler in the vertical direction there are two
Taking into account the difficulties of obtaining measurements propulsion systems commonly used: adjusting the buoyancy or
by human operators, various autonomous profiling vehicles using thrusters. Buoyancy driven profilers move by changing
were developed to perform these tasks in a systematic and their submerged volume which causes the vehicle to either
more reliable way, with minimal human intervention. sink or rise [4], [5], [1], [6], [7], [8]. The main advantage of
An underwater profiler is thus a vehicle that moves predom- this type of propulsion is that it only uses energy to change
inantly along the vertical axis, and typically drifts along the its buoyancy state, making them extremely energy efficient
horizontal plane with the currents. Underwater profilers are a on great descents [1]. However, since the buoyancy of the
particular class of Autonomous Underwater Vehicles (AUVs), vehicle depends on water density, this type of propulsion is not
with a design that minimizes drag in the vertical direction. suitable for shallow water environments where water density
This paper presents the development of a new autonomous varies greatly and affects the performance of the vehicle [5].
profiler for coastal areas. The main objective was to acquire Thruster driven profilers use propelled thrusters to move the
profiles up to 200m of depth, while maneuvering also in vehicle [2], [9], [3]. The vertical performance of the thruster
the horizontal plane. The vehicle has no fins, and uses four is not affected by the water density as much as the buoyancy,
independent thrusters to provide both vertical and horizontal since it’s easier to control descent rate with the appropriate
motion, including hovering in the water column. thruster control [3]. The main disadvantage of using a thruster
The paper is organized as follows: Section II provides is that it is an active propulsion system that constantly needs
an overview of the main efforts that have been made to energy to drive the profiler, making it inefficient for dives in
develop underwater profiling vehicles. Section III summarizes greater depths.
explicitly the main requirements for the development of this
new vehicle, while Section IV details the components of the B. Motion Control
system, including the hardware implementation and the soft- To perform the required maneuvers, a profiler must run
ware structure. Section V presents the hydrodynamic model algorithms that controls the propulsion system. In the buoy-
of the vehicle and the controllers designed for the maneuvers. ancy driven profiler described in [5], a task implements a

978-0-692-94690-9/$31.00 ©2017 MTS

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 PID controller using the measurements of depth and altitude malfunction. Additionally, it should detect situations that can
above seabed. This controller computes an intermediate depth cause harm to the system. The profiler should then be able to
trajectory using basic filtering to keep the depth changes within detect or measure the following faults and act accordingly:
the dynamic capabilities of the profiler. • Detect water ingress;
To acquire a profile of the water column, the vehicle • Detect overheating inside the pressure case;
described in [9] runs at a constant speed and, as it nears the • Detect the sea bed to avoid collision;
target depth, ramps down the thruster voltage to zero bringing • Report battery status;
the profiler close to the target depth. Then, the profiler ascends
using the positive buoyancy defined on construction. The vehi- IV. I MPLEMENTATION
cle described in [2] operates in a similar way. The Autonomous
A. Mechanical Architecture
Vertical Profiler described in [9] can also hover at a fixed
depth. It uses a controller comprising of Linear Quadratic 1) Weight and Buoyancy: In order to minimize the energy
Regulator (LQR) in combination with a complementary filter, spent in changing the depth of an underwater vehicle, the
as described in [3]. vehicle should be neutrally buoyant. However, a typical safety
measure is to operate underwater vehicles with a little positive
III. M AIN R EQUIREMENTS buoyancy, to ensure that it returns passively to the surface
The list of prototype requirements took into account the fol- in case of a major fault. Since buoyancy depends on water
lowing guidelines: the autonomous profiler construction should density, which can vary significantly in coastal waters, a
follow a modular design and allow low logistics operation typical margin of 1–2% of dry weight is typically employed.
during transportation, deployment and recovery. This system To accommodate different vehicle configurations and to make
should be designed to operate on coastal areas (and shallow final adjustments in weight and buoyancy, the profiler was
waters) without any physical connection with an operator. The designed with free volumes in flooded sections, both at the top
requirements considered for the development of the vehicle and at the bottom, to insert flotation material or lead weights.
were divided in two subsections: a) Operational requirements, The passive attitude of an underwater vehicle is determined
related to the maneuvers that the profiler should perform and by the relative position of the center of mass and center of
its operation; and b) Safety requirements, that should reduce buoyancy. In order to achieve the desired attitude (vehicle
the risk of fault and minimize their impact in the behaviour pointing downwards), all heavier subsystems have been in-
of the vehicle. stalled towards the nose cone, equally distributed around the
vertical axis.
A. Operational Requirements
2) Mechanical Modules: The mechanical design was based
The system should allow a defined set of maneuvers taking on a modular system used for the assembly of autonomous
into account the considered usage scenarios. Therefore, the underwater vehicles [10]. This system has been used to as-
vehicle should be able to perform the following maneuvers: semble multiple configurations of AUVs, such as the MARES
• profile in the same position (estimating drift caused by AUV, in operation for more than 10 years [11]. In the case
currents and returning to reference position); of the profiler, the mechanical sections are revolution shapes
• acquire profiles with arbitrary angles; with 120mm of diameter, that can stack to each other and
• hovering at a fixed depth or fixed height above sea propagate the male/female coupling system. They can be
bottom; designed to carry wet sensors and thrusters and they are fully
• profile within 2 different depths without surfacing. interchangeable, which allows for very easy sensor swapping
Moreover, the profiler should satisfy the following require- and/or repositioning.
ments related with the logistics of a mission: The most critical parts, like the pressure housing, were
• The profiler should require a single operator for launch machined in polyacetal copolymer (POM), a very strong
and recovery (weight in air limited to 30Kg and length technical plastic. Others, like the supports for the thrusters
limited to 1.5m); or the nose of the profiler, were printed using a 3D printer.
• Battery life should be sufficient to perform 10 profiles to Figure 1 presents the final configuration of the profiler, with
200m depth with a velocity of 1m/s along 24 hours; a total mass of 11.3kg and 1.35m long.
• At the end of the mission, the vehicle should communi- The pressure housing has 8 holes in each end cap, to
cate its position to be recovered; accommodate standard bulkhead connectors, to install sensors
• The mission plan should be communicated using a wire- in contact with the water, and to pass cables using cable pen-
less connection; etrators. All underwater cables are routed within the flooded
• The vehicle batteries should be rechargeable without sections, for protection and to minimize drag. Figure 2 shows
opening the pressure housing. the arrangement of interfaces in the top end cap.
The electronics are mounted inside the pressure housing
B. Safety Requirements fixed to an aluminum frame (figure 3). The batteries are
As a first step towards safe operation, the profiler should mounted at the bottom to lower the center of gravity of the
have a positive buoyancy to return to the surface in case of vehicle and are fixed in the tray using 3D printed supports.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 Fig. 3. Profiler prototype and inner frame.

Fig. 1. Final configuration of the profiler.

Fig. 4. Subsystems of the profiler.

the power consumption of the other subsystems. They can be

recharged using an underwater connector available at the top
part of the profiler. This subsystem represents a significant
volume and weight, therefore it was be placed in the bottom
part of the pressure housing.
2) Processing Unit: The processing unit is a BeagleBone
Black single board computer (SBC) that runs the onboard
software on a Linux distribution. This SBC features a AM335x
Fig. 2. External view of the top end cap of the pressure housing, with sensors, 1GHz ARM processor and has 4 GB of onboard storage.
underwater bulkheads, and cable penetrators. For the selection of the processing unit, the main selection
criteria were: lowest power consumption; reduced size and
weight; and interfaces available to communicate with the other
B. Hardware Architecture components.
3) Propulsion System: The propulsion system is composed
The vehicle is composed by a set of interconnected subsys- by four BlueRobotics T100 thrusters mounted on top of
tems as depicted in figure 4. the vehicle (figure 5), with alternate reversed propellers to
1) Power System: This subsystem, located inside the pres- minimize torque during profiles. Each thruster can provide
sure housing, includes the batteries, voltage converters and up to 25N of force, and they are independently actuated
power buses needed to power all the profiler devices. The with electronic speed controllers installed inside the pressure
batteries are two Li-Ion BA95HC-FL packs, connected to a housing. The independent actuation of the thrusters allows to
BBDC-02R Dual Battery Controller with ATX Power Supply, control the vertical velocity of the profiler, as well as pitch
all from OceanServer. Each battery pack has 95Whr of energy and roll angles, as described later.
at a nominal voltage of 14.4V. They were selected as a tradeoff 4) Navigation System: The navigation system includes the
between the available volume and the desired duration and sensors that allow the profiler to estimate its position, heading

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 divided in 3 main modules as depicted in figure 6. These
modules share a block of shared memory, where the data and
the timestamp associated with each measurement is available.

Fig. 5. Top view of the profiler.

Fig. 6. Onboard software modules.

and attitude. In the current configuration, the profiler uses a Each module has a different purpose: the sensors drivers
BlueRobotics Bar30 pressure sensor to estimate depth, with a module is responsible to acquire the data from all the sensors
maximum range of 300m. For linear and angular accelerations, available and write it to the shared memory, the data logger
the profiler has a tiny ITG3200/ADXL345 6DOF IMU inside logs the data and the timestamps during the operation of
the pressure housing. To obtain the absolute position at the the profiler and the mission control and supervision module
surface, an Amaryllo GPS receiver has been potted in epoxy is responsible to read and execute the mission plan, control
and placed in the topmost part of the profiler, as indicated in the vehicle and detect possible faults on the operation of the
figure 1. system.
5) Scientific Data System: A set of payload sensors can
be installed in the profiler to acquire the desired scientific V. DYNAMIC M ODEL & C ONTROLLERS
data. When these sensors need to be in direct contact with The 6 degrees of freedom (DOF) model that describes the
water, they have to be installed in one of the end caps, or in movement of the vehicle follows the approach presented in
separate flooded sections. There are still 5 physical ports left to [12]. To properly analyze the system, it is convenient to define
interface sensors with the processing unit, but this number may two coordinate frames: a moving coordinate frame fixed on the
increase by splicing cables and/or adding interfacing boards. body (body-fixed frame) and an inertial reference frame (earth
In order to install a new sensor, both power and communica- fixed frame) as presented in figure 7.
tions need to be routed from the sensor to the relevant port
The motion of an underwater vehicle can be described in
and there are several alternatives in terms of voltage levels
the body fixed frame using a matrix representation as follows:
and interface protocols. In the current configuration only a
water temperature sensor has been installed, the BlueRobotics
M ν̇ + C(ν)ν + D(ν)ν + g(η) = τ (1)
Celsius Fast-Response Temperature Sensor. This allows the
profiler to obtain temperature-depth curves for a given water η̇ = J(η)ν (2)
column.
6) Communication System: The communication system where M represents the inertia matrix, C(ν) is the matrix
supports all data exchange between the profiler and a user or of Coriolis and centripetal terms, D(ν) is the damping matrix,
base station, including the transmission of the mission plan, g(η) is the vector of gravitational forces and moments, and τ
the download of mission data, and the communication of the is the vector of control inputs.
profiler’s position at the end of the mission. In the current
version, the profiler uses a standard USB WiFi transceiver, A. Speed Controller
potted in epoxy close to the GPS receiver, at the top of the In order to perform the defined maneuvers, it is necessary to
profiler. For long range, open ocean operations, it is possible control the speed of the profiler as this is one of the parameters
to replace the WiFi transceiver by an Iridium modem. of the mission plan.
7) Internal Sensors: A set of internal sensors have been The speed controller controls the velocities in z, pitch and
installed inside the pressure housing to monitor possible faults. roll. It was designed using Lyapunov theory, which allows to
These include, for instance, a water ingress sensor and a conclude about the stability of a system and designing control
temperature sensor to detect overheating. laws, overcoming the limitation of linear controllers that only
are guaranteed to work around the selected operating points
C. Onboard Software [13], [14], [15]. Lyapunov theory states that if ”there exists a
The software architecture was implemented to allow an easy scalar function V of the state x with continuous first derivatives
accommodation of new sensors. This software runs in real- such that V (x) is positive definite, V̇ (x) is negative definite
time and has multiple processes to ensure that different tasks and V (x) → ∞ as x → ∞, then the equilibrium point at the
are performed in useful time. The software implemented is origin is globally asymptotically stable” [16].

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 To satisfy (6), we cannot vary the error variable instanta-
neously. Therefore, we define a new error variable α as:
α = v̇r − v̇ref + ke e (7)
That imposes the time derivative Lyapunov function to be:
V̇ = eT (α − ke e) (8)
Substituting v̇r in equation (7) for the expression for v̇ given
in (1) yields:
α = M −1 [(C(vr )+D(vr ))·vr +g(η)+M (v̇ref −ke e)+τ ] (9)
Considering that τ can be actuated using f , to guarantee that
V̇ is negative definite, τ is:
τ = (C(vr ) + D(vr )) · vr + g(η) + M (v̇ref − ke e) (10)
B. Position Controller
The position controller controls the position on the 3 DOF:
depth, pitch and roll. The position controller feeds the speed
controller by passing the speed references to it (see figure 8).

Fig. 7. Reference coordinate frames for the vehicle model.

Fig. 8. Block diagram of the controllers.

The profiler is controlled using four thrusters and it was
assumed that the thrusters can instantly apply force to the
vehicle. Although this is not exact in reality, this assumption Starting by defining the errors in depth, pitch and roll as:
was made to simplify the model and it is justified since the ez = zref − z (11)
time constants associated with the actuation are smaller than
the ones associated with vehicle motion [15]. eφ = φref − φ (12)
In order to simplify the determination of the speed con- eθ = θref − θ (13)
troller, the order of the model was reduced eliminating lines
and columns of the matrices that have low influence on the The references for the speed controller are then defined as:
motion of the vehicle. 
dez
To control the velocities in z (w), roll (p) and pitch (q), an w = Kpz ez + Kiz ez dt + Kdz (14)
dt
error vector is defined as e = vr − vref : 
T p = Kpφ eφ + Kiφ eφ dt (15)
e = [w − wref , p − pref , q − qref ] (3)

Then the Lyapunov candidate function V is defined as: q = Kpθ eθ + Kiθ eθ dt (16)
1 T
V = e e (4) where Kpz , Kpφ , Kpθ , Kiz , Kiφ , Kiθ , Kdz , ∈ R+ are propor-
2 tional, integral and derivative gains, respectively.
In which the time derivative is: The integral part of the controller is used to eliminate the
steady state error to the reference. For the depth controller, it
V̇ = eT ė = eT (v̇r − v̇ref ) (5) was also considered a derivative term to prevent overshoot,
since during the descent the integral part of the controller
To guarantee the stability of the system, V̇ must be negative
accumulates error that makes the profiler overshoot.
definite which imposes:
For each maneuver, the controller gains were determined
eT (v̇r − v̇ref ) < 0 ⇒ (v̇r − v̇ref ) < −ke e (6) considering velocities saturation. Since the dynamics of each
maneuver are different, the considered gains were determined
where ke ∈ R, ke > 0. for each maneuver.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 Changing abruptly the position reference of the controllers
makes the vehicle accelerate abruptly, causing significant
peaks in power consumption since the vehicle needs to ac-
celerate quickly to meet the speed reference. To ensure that
the position reference for the controllers is not made abruptly,
the position reference is generated using a ramp with a slope
equal to the desired velocity for the profiler. This makes the
profiler go to the desired speed and position more smoothly.

C. Controller Simulations
Using the model and the controllers described above, the
overall performance was estimated through simulation before
actually trying in the real profiler. Figures 9 and 10 present
the results of the simulation, using the full 6DOF model of
the vehicle, where it was instructed to descent to 3.5 meters Fig. 11. Speed during gotoz maneuver with 20 degrees in pitch and roll.
with a speed of 0.3 m/s (gotoz maneuver).

Fig. 12. Position during gotoz maneuver with 20 degrees in pitch and roll.

Fig. 9. Speed during gotoz maneuver with 0 in pitch and roll.

VI. O PERATION OF THE V EHICLE
To plan a mission on the vehicle, the user must create a
mission file with the name config.ini (as shown in figure 13).
This file must be started with [mission] and the following
commands are available:
• gotoz=<depth> <pitch> <roll> S <descent rate> T
<timeout> — command that allows to profile to a certain
depth (in meters) controlling pitch and roll (in degrees)
and descent rate (m/s).
• surface=<depth> S <ascent rate> T <timeout> —
command to make the profiler surface with a defined
ascent rate (m/s).
• wait=<seconds> — command to wait for a determined
period of seconds. Allows to set intervals between differ-
ent maneuvers. Since the vehicle has a positive buoyancy,
the vehicle waits at surface.
• hover= <depth> D <duration> T <timeout> — com-
Fig. 10. Position during gotoz maneuver with 0 in pitch and roll. mand to hover at fixed depth during the number of
seconds defined in duration.
Figures 11 and 12 present the results where the profiler For each motion command, a timeout must be set, in
was instructed to descent to 3.5 meters with a speed of 0.3 seconds. In case the profiler does not complete the maneuver
m/s (gotoz maneuver) and with 20 degrees in pitch and roll. on time, the mission passes to the next command.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 Fig. 13. Example of a possible mission configuration.

VII. E XPERIMENTAL VALIDATION

In the experimental tests described in this section, data was
logged to a log file named DataLog<time>.txt, with a period
of 250ms. This file was retrieved from the profiler using the
Wi-fi connection and was then processed to obtain the figures
shown. All these tests were conducted in July 2017, in test
tanks at INESC TEC facilities (figure 14).
Fig. 15. Depth, pitch and roll during gotoz to 3.5m.

For the second test, the profiler was commanded to dive

to 0.4 m depth with an angle of -30 degrees in pitch and
30 degrees in roll. It is possible to observe in figure 16 that
the profiler exceeds its depth target. This happens because the
forces needed to provide a rotation of the vehicle, also make
the profiler dive deeper. Once the angles in pitch and roll are
reached, the profiler goes to the desired position.

Fig. 14. Profiler at the surface in one of the test tanks at INESC TEC.

In the following figures, the target depth of the profiler is

represented together with measured data, to verify that it con-
trols its speed, following the reference during the maneuver.
All the tested maneuvers are preceded by a wait command of 5
seconds to allow the vehicle to reach a steady state. Moreover,
the profiler waits 2 seconds to initialize the Electronic Speed
Controllers resulting in a delay of 7 seconds before the profiler
starts a maneuver.

A. Go To z Test
To test the Go To z maneuver, two different tests were
performed: one where the profiler dives to a defined depth Fig. 16. Depth, pitch and roll of the profiler when diving up to 0.4m.
completely vertical (ie, pitch and roll are both zero) and
another where the dive is made with an angle in pitch and
roll different than zero. The maneuver is considered completed B. Mission
when the profiler is within 10 cm of the target and the angle To test a sequence of maneuvers, the vehicle was instructed
error is lower than 2 degrees. to hover at three different depths: 1.5, 2.5 and 3.5 meters
For the first test, the profiler was instructed to dive down followed by a surface with an ascent rate of 0.5 m/s. Figure 17
to 3.5 meters. The results show that the profiler achieves its shows the results and it is possible to confirm that the vehicle
target depth with a controlled speed while, at the same time, hovers at the desired depths and then, after completing the
controlling pitch and roll angles close to zero (figure 15). maneuvers, rises with a controlled speed to the surface.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.
 In order to increase the usefulness of the profiler at sea,
there are a few possible improvements. The first is to add
more sensors, like a down-looking camera, an altimeter to
measure altitude above sea bottom, and water quality sensors.
Another is to design an user interface for mission planing,
to start a mission and to analyze retrieved data. An Iridium
module would also be useful to expand the communication
capabilities of the profiler at the surface. Considering the
modular architecture of the developed system, the profiler is
ready for a swift integration of all these subsystems.

ACKNOWLEDGMENT
This work is financed by the ERDF – European Regional
Development Fund through the Operational Programme for
Competitiveness and Internationalization – COMPETE 2020
Programme, and by National Funds through the FCT –
Fig. 17. Hover at three different depths in one mission. Fundação para a Ciência e a Tecnologia (Portuguese Foun-
dation for Science and Technology) within project POCI-01-
0145-FEDER-006961.
VIII. C ONCLUSION AND F UTURE W ORK
This paper presents the development of a new autonomous R EFERENCES
profiling vehicle for coastal areas. A fully functional proto- [1] E. Petzrick, J. Truman, and H. Fargher, “Profiling from 6,000 meter with
type was developed using a systems engineering approach, the apex-deep float,” in 2013 OCEANS - San Diego, Sept 2013, pp. 1–3.
[2] B. Ward and T. Fristedt, “Air-sea interaction profiler: Autonomous upper
following three different phases in an iterative process: design, ocean measurements,” in 2008 IEEE/OES US/EU-Baltic International
implementation, and validation. Symposium, May 2008, pp. 1–8.
The development of the prototype focused on the physi- [3] P. K. Maurya, E. D. Sa, A. C. Dubey, N. Dabholkar, and A. Pascoal,
“Autonomous hovering profiler,” in 2016 IEEE/OES Autonomous Un-
cal implementation, fulfilling the main requirements: it is a derwater Vehicles (AUV), Nov 2016, pp. 268–272.
portable vehicle capable to perform profiles up to 200m of [4] R. E. Davis, L. A. Regier, J. Dufour, and D. C. Webb, “The autonomous
depth and the software implementation allows the vehicle to lagrangian circulation explorer (ALACE),” Journal of Atmospheric and
Oceanic Technology, vol. 9, no. 3, pp. 264–285, 1992.
autonomously operate, executing a mission plan and gathering [5] A. Schwithal and C. Roman, “Development of a new lagrangian float
data from its set of sensors. The modularity of the vehicle for studying coastal marine ecosystems,” in OCEANS 2009 - EUROPE,
allows to easily add new sensors, using the ports available May 2009, pp. 1–6.
[6] X. Xu and C. Yang, “A pneumatic-driven airdropped portable under-
in the endcaps. These can be integrated in the vehicle’s water profiler for rapid deployment,” in OCEANS 2015 - MTS/IEEE
software by implementing a device driver and incorporating Washington, Oct 2015, pp. 1–5.
the measurements in the shared memory and the logfile. [7] Y. Zhu, X. Xu, J. Wang, C. Yang, Q. Li, and M. Cai, “A hybrid
underwater profiler used for persistent monitoring,” in OCEANS 2015 -
To design and validate the motion controllers, a 6 DOF MTS/IEEE Washington, Oct 2015, pp. 1–5.
model was considered, describing the profiler dynamics and [8] V. Viswanathan and T. Taher, “Buoyancy driven autonomous profiling
kinematics. This model was used to tune and test the con- float for shallow waters,” in OCEANS 2016 MTS/IEEE Monterey, Sept
2016, pp. 1–6.
trollers before deployment in the test tanks. The performance [9] E. Desa, R. Madhan, N. Dabholkar, S. Prabhudesai, G. Navelkar,
of the profiler during the experimental tests demonstrate the A. Mascarenhas, S. Afzulpurkar, M. Phaldesai, and P. K. Maurya,
adequacy of the controllers and the accuracy of the model. “In situ profiling of eastern arabian sea coastal waters using a new
autonomous vertical profiler,” IEEE Journal of Oceanic Engineering,
Taking into account the results presented above, it is possi- vol. 38, no. 1, pp. 43–54, Jan 2013.
ble to conclude that the profiler is already able to perform basic [10] N. A. Cruz, A. C. Matos, and B. M. Ferreira, “Modular building blocks
maneuvers and can be used as a platform to collect ocean data. for the development of AUVs – from MARES to TriMARES,” in Proc.
Int. Symp. Underwater Tech. UT’13, Tokyo, Japan, Mar. 2013.
However, the developed system can still benefit from further [11] N. A. Cruz and A. C. Matos, “The MARES AUV, a modular autonomous
testing to improve overall performance. For example, the robot for environment sampling,” in OCEANS 2008, Sept 2008, pp. 1–6.
parameters used for the hydrodynamic model can be refined [12] T. I. Fossen, Guidance and control of ocean vehicles. Chichester ; New
York : Wiley, 1994.
based on experimental tests and identification techniques, to [13] A. P. Aguiar and J. P. Hespanha, “Trajectory-tracking and path-following
yield a more accurate model of the vehicle. This improved of underactuated autonomous vehicles with parametric modeling uncer-
model will result in simulations closer to reality, that can be tainty,” IEEE Transactions on Automatic Control, vol. 52, no. 8, pp.
1362–1379, Aug 2007.
used to improve the performance of the controllers. [14] L. Lapierre and D. Soetanto, “Nonlinear path-following control of an
One limitation of the profiler is the rotation in yaw caused AUV,” Ocean Engineering, vol. 34, no. 1112, pp. 1734 – 1744, 2007.
by the torque of the propellers. To overcome this effect, it is [15] B. Ferreira, M. Pinto, A. Matos, and N. Cruz, “Control of the MARES
autonomous underwater vehicle,” in OCEANS 2009, MTS/IEEE Biloxi,
possible to replace the IMU for a 9 DOF version, and measure 2009, pp. 1–10.
the yaw angle. The rotation could then be independently [16] J.-J. E. Slotine, W. Li et al., Applied nonlinear control. Prentice-Hall
controlled, granting full control of the profiler heading. Englewood Cliffs, NJ, 1991.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on September 01,2021 at 08:27:28 UTC from IEEE Xplore. Restrictions apply.