Design and control of an Unmanned Surface Vehicle for

Environmental Monitoring Applications.

J.L. Villa1 , J. Paez2 , C. Quintero2 , E. Yime1 , and J. Cabrera1

Abstract—Unmanned Surface Vehicles (USV) have been used countries Autonomous vehicles start to perform the most
in a wide range of applications, in particular in environmental dangerous tasks in military and civil applications [?].
monitoring. In this work we propose the design and control of an This paper is organized as follows. In the second section
USV which is intended to be used for monitoring environmental
variables of the main inlet waters located in Cartagena de Indias, we provide a general overview of the Naval design. In section
Colombia, such as Chambacú, Cabrero and San Lázaro. three We describe the main components of the Guidance,
The paper shows the results obtained in the naval design, focused Navigation and Control (GNC) system. In section fourth, We
on the hull design, and the control strategy for unmanned describe the dynamical model of the system and the control
navigability which is tested in simulation. design.

I. INTRODUCTION II. NAVAL D ESIGN

Unmanned Maritime Vehicles (UMV) are divided into two In this section We introduce the main steps of Naval Design.
types: Unmanned Surface Vehicles (USV) and Unmanned First, The Hull Design is presented. Second, the propulsion
Underwater Vehicles (UUV). USV are a type of boat or jet design is described and finally We describe some issues in the
ski surfing the water surface or navigating semi-submerged. Hull BUilding Process.
Unmanned Underwater Vehicles (UUV) generally refers to A. Hull design
those vehicles without human crew, with either automated
Naval design takes the form of the so-called ship design
navigation, such as Autonomous Underwater Vehicles (AUV)
spiral where a higher precision is progressively achieved at
or Remotely Operated Vehicles (ROV). AUV are usually more
each cycle or phases. The ship design process starts with
smaller than their USV counterparts, [?]. ROVs have the
deciding the mission requirements. According to the mission
main property of having a cable, called umbilical cord, which
and expected functionality, ship parameters and characteristic
connects the vehicle with a surface boat, allowing in this way
are defined. The sequence continues with the determination
communication, power transmission and interchange of data
of the hull dimension (length, beam and draft), followed by
or control commands [?].
arrangement of hull and machinery as well as determination
Maritime autonomous vehicles have been widely used in of preliminary lines and body plan, powering, type of engine,
offshore applications such as Oil and Gas Industry, where and structural arrangement. This process continues until a
they have gained great popularity during last decades [?],[?]. final setting of the list of materials, specifications, and cost
Recently, ROVs have been used in other applications, such as estimation.
estuarine and fluvial environments [?], military development The figure 1 shows the body lines plan for 1.3m length boat.
[?], detecting contaminants [?], and inspection of underwater A Naval Design software is used for hydrostatic and stability
structures post-disaster [?]. analysis, hull arrangement and internal distribution of weights
The advantages of this technology over manned vehicles and centers.
are that USV are not constrained due to restrictions imposed
by a human crew member, such as temperature, space, or
movements. They are designed based on the specific mission,
besides they are not limited by extremely hazardous environ-
ments for humans, such as nuclear or waste contaminated
environments with very heavy sea conditions [?]. In some

*This work is supported by Universidad Tecnológica de Bolı́var under con-

tract FI1506T2001 ”Design and Implementation of an Autonomous Operation
System for a boat”
1 J. Villa, E. Yime and J. Cabrera are with Faculty of Engineering,
Fig. 1. Perspective and lines plan
Universidad Tecnológica de Bolı́var, Km 1, Via Turbaco, Cartagena, Colombia
jvilla,eyime,jtovar at unitecnologica.edu.co
2 J. Paez and C. Quintero are Master Students in the Engineering Faculty,
Universidad Tecnológica de Bolı́var, Km 1, Via Turbaco, Cartagena, Colombia B. Propulsion Design
cristianutb, jheiferpaez at gmail.com One of the most important stage of a ship design process
is the estimation of power requirement. For achieve this, it
978-1-5090-3787-2016/$31.00 2016
c IEEE
is necessary the estimation of ship resistance and propulsion
efficiency. Several approaches are used to evaluate ship
resistance, from experimental to empirical and numerical
methods. In this case, we use the empirical estimation method
developed by Holtrop and Mennen through a regression
analysis of random ship model tank experiments,[?]. In the
implementation of this project 90 watts is required of ship
powering to reach a froude number 0.8 of pre-planning.
Considering the maximum ship displacement of 17 kg then
two electric motors with gear reduction (type 500) are used
with a maximum power of 60 watts each, eliminating the use
of water cooling system.

C. Hull Building Process

Fig. 3. Final Hull Ship
The materials used for construction of the hull are
fiberglass and resin orthophthalic. Fiberglass fabrication
process involves the ship molding to shape and support
materials until they solidify. Hull sections from body plan
are used for mounting base molding. Multiple layers of mat
and roving are built into a laminate and after the resin in
all the layers has set up hard and cured, using the so called
hand lay-up technique. Finally the solid hull part can be
removing from the mold. The figure 2 shows the transverse
ship sections in the process of building the wood-base mold
and figure 3 shows the resulting fiberglass hull.
Fig. 4. GNC signal Flow

is still one of the most difficult issues because although many

investigations have been brought forward [?], [?], there are still
difficulties in the recognition of the objectives [?] and obstacle
avoidance [?]. There are also other areas of work which are
still being investigated, such as control and maneuverability
in extreme conditions [?].
In consequence, it is common to use semi-autonomous
vehicles because the human supervision is the easiest way to
increase reliability [?], unfortunately this does not eliminate
the risk since according to statistics the human errors are
present in most collisions at sea. For this reason there are
works like [?] where guidance is based on international
regulations for preventing collisions at sea.
Fig. 2. Building Hull In this case, the most part of the GNC system is supported
by a Raspberry Pi 2 and Navio+, see figure 5. A real-time
Linux operating system (RTLinux) is used running over the
integrated features of a Broadcom BCM2836 SoC micropro-
cessor.
III. GUIDANCE NAVIGATION AND CONTROL
A. Navigation
In unmanned vehicles navigation, guidance and control sys-
tems (GNC) are responsible for obtaining the current location The selection of the Navigation sensors for a USV are
of the vehicle relative to a reference system (Navigation), to determined by their performance on the environment where
determine the path and speed to follow (Guidance) which is they will operate:
transformed into commands for a Control System [?], [?]. In this case the most cost-effective packet found is the
In the most advanced applications, the guidance system Navio+, a system with the following sensors:
must have enough intelligence to achieve safe navigation with-
out human intervention, identifying environmental conditions • MPU-9250 is a multi-chip module (MCM) consisting of
[?], objectives, and detect and avoid obstacles. But this feature a single QFN package with a 3-Axis gyroscope, 3-Axis
 IV. CONTROL DESIGN
The dynamical model for control design is a simplified
version of the simulation model and uses the number of
degrees of freedom (DOF) needed to carry out its mission, in
this case 3. This is a classic model of maneuverability, where
the forward movement and yaw deviation is analyzed under
the assumption that the boat is moving to a positive velocity
U without the effect of the waves.

A. Modeling
From the point of view of dynamics, a marine vehicle can be
considered as a rigid body with six degrees of freedom, three
Fig. 5. UART connection integrated circuits coordinates to describe the movements of translation and three
for the rotation movements. When the vehicle movements are
analyzed it is convenient to define two coordinate systems,
accelerometer and 3-axis magnetometer [?]. one fixed ground n and one fixed to the vehicle b, whose
origin usually coincides with the center of gravity (CG) if it
• The MS5611-01BA is a high resolution altimeter sensor is located in the principal plane of symmetry. Choosing an
with SPI and I2 C bus interface. This barometric pressure orthogonal coordinate system parallel to the principal axes of
sensor is optimized for altimeters and variometers with inertia reduces the complexity of the equations.
an altitude resolution of 10 cm [?]. Classic models of maneuverability have been divided into
holistic models and modular models, an example of the firsts
• The NEO-6 module series is a family of stand-alone are the called equations of Abkowitz [?], which takes into
GPS receivers featuring the high performance u-blox account the effect of independent variables until third order. In
6 positioning engine. These receivers offer numerous this model the dynamic of the boat is seen as a closed block
connectivity options in a miniature 16 x 12.2 x 2.4 mm where the effects of propulsion, rudder and hull shape are
package [?]. part of a single system. While modular approaches divide the
model into subsystems that can be studied and analyzed sepa-
Navio+ is an embedded system and supports SPI commu- rately. In general, mathematical models describe the geometry,
nication with Raspberry Pi 2. Navio+ controls all the sensor inertial characteristics and interaction with the environment
and answer with the requested measured data when needed. through the hydrodynamic coefficients [?], which can be found
for compact or modular models using the following methods:
B. Guidance
• Theoretical methods.
In this work is used Mission Planner, figure 6; a widely • Numerical methods.
used ground station application for the ArduPilot open source • Experimental methods.
autopilot project. • Interpolation systematic data series.
Some Mission Planner features are:
Below are shown the equation of motion using the Fossen
• Setup, configure, and tune a vehicle for optimum perfor- nomenclature:
mance.
• Monitor a vehicles status while in operation. (MRB + MA )ν̇ + (CRB (ν) + CA (ν))ν + D (ν) ν = τ (1)
• Operate a vehicle in FPV (first person view). Where:
MRB Is the mass matrix and inertia.
MA Is the added mass matrix and inertia.
CRB (ν) Is the Coriolis matrix due to ship rotation about the
fixed coordinate system.
CA (ν) Is the Coriolis matrix due to added mass in a rotating
coordinate system.
D(ν) Is the damping matrix.
τ It is the vector of control inputs.

The properties of each of the terms of the equation of

motion can be found on [?].

Fig. 6. Mission Planner Test Now, in order to describe all the components of the motion
equation let the {b}-frame coordinate origin be set in the
center-line of the craft at the point CO, such that yg = 0,
then:
 
m 0 0
MRB =  0 m mxg  (2)
0 mxg Iz
 
−Xu̇ 0 0
Fig. 7. Simulink diagram
MA =  0 −Yν̇ −Yṙ  (Yṙ = Nν̇ ) (3)
0 −Yṙ −Nṙ
  are a the propulsive force, actual position and velocity; the
0 0 −m(xg r + ν) output is the acceleration. Hydrostatic forces are not taken into
CRB (ν) =  0 0 mu  (4) account because the movements are in the horizontal plane.
m(xg r + ν) −mu 0 The PID block is part of the closed loop and computes a scalar
  control signal which is transformed by the Compute Force
0 0 Yν̇ ν + Yṙ r
block into a vector using the actual attitude of the ship. The
CA (ν) =  0 0 −Xu̇ u  (5)
integrator block is used to compute the velocity and position
−Yν̇ ν − Yṙ r Xu̇ u 0
  vector from the acceleration.
Xu 0 0 C. Control Simulation
D(ν) =  0 Yν Yr  (6)
0 Nν Nr As shown in Figure 7 the control has been simulated in
Simulink, this is part of the initial tests.
B. Control
When designing motion control systems, the control objec-
tive must be well defined in order to satisfy the requirement
specifications for safe operation of the craft. In this context,
it is important to distinguish between the following three
important control objectives: Set-point Regulation, Trajectory-
Tracking Control and Path-Following Control. This paper
presents a Set-point Regulation control, the most basic guid-
ance system, where a human operator provide a constant input
or set-point and the corresponding controller will then be a
regulator. In our case, the operator will introduce a x, y desired
position and the control will move the ship toward to the
destination.
Motion control systems for marine craft have been an
active field of research since the first mechanical autopilot
was constructed by Elmer Sperry in 1911, see [?]. Modern
control systems are based on a variety of design techniques
such as PID control, linear quadratic optimal and stochastic Fig. 8. Ship displacement
control, H∞ control methods, fuzzy systems, neural networks The figures 8,9 and 10 show that the boat is located in (0, 0)
and nonlinear control theory, And many others. and moves up to (50, 50). This is the expected action.
Although PID control techniques have been replaced in
many applications by LQG and H∞ , PID control is still widely V. CONCLUSIONS
used today. In this paper is use this technique. In this work the design of a USV is presented. Initially, this
In most ship operational conditions, the desired trajectory work is focused on the hull design and the guidance navigation
is slowly varying motion due the 2nd order wave loads, wind and control systems. In order to simulate the control of the
and currents compared to the oscillatory 1st order motion USV, some tests are performed using the simulation model
induced by the waves, so it is necessary to avoid the 1st in simulink showing that the performance is acceptable since
order motion because it causes wear and tear in the thrusters the ship moves up to the desired position. The future work
and propeller systems. It is solved using a model based in includes to calculate the thrust force and improve the whole
space state estimator(normally a Kalman filter), to use a PID model.
controller in cascade with a deadband, notch and/or low-pass
R EFERENCES
filter can solve the problem too, but add an additional phase
lag and nonlinearities to the system. [1] J. Brown, C. Tuggle, J. MacMahan, and A. Reniers, “The use of
autonomous vehicles for spatially measuring mean velocity profiles in
In the figure 7 it is showed a block called Ship model which rivers and estuaries,” Intelligent Service Robotics, vol. 4, pp. 233–244,
contains the Mass, Coriolis and Damping matrices; its inputs Oct. 2011.
 [11] M. Kristan, V. Sulic, S. Kovacic, and J. Pers, “Fast image-based obstacle
detection from unmanned surface vehicles,” CoRR, vol. abs/1503.01918,
2015.
[12] W. Naeem, G. W. Irwin, and A. Yang, “COLREGs-based collision
avoidance strategies for unmanned surface vehicles,” Special Issue on
Intelligent Mechatronics (LSMS2010 & ICSEE2010), vol. 22, pp. 669–
678, Sept. 2012.
[13] InvenSense Inc., MotionTracking device, 1 2014. Rev.1.
[14] Measurement Specialties., Barometric Pressure Sensor, with stainless
steel cap, 10 2012.
[15] u-blo, u-blox M8 concurrent GNSS modules.
[16] M. A. Abkowitz, Lectures on ship hydrodynamics : steering and
manoeuvrability. Lyngby: Hydrodynamics Department, 1964.
[17] J. Newman, Marine Hydrodynamics. Wei Cheng Cultural Enteroprise
Company, 1977.
[18] T. Fossen, Handbook of Marine Craft Hydrodynamics and Motion
Control. Wiley, 2011.
[19] and others, “Gyroscopic navigation apparatus.,” June 21 1911. US Patent
App. 1/255,480.

Fig. 9. X displacement

Fig. 10. y displacement

[2] R. Murphy, E. Steimle, M. Hall, M. Lindemuth, D. Trejo, S. Hurlebaus,

Z. Medina-Cetina, and D. Slocum, “Robot-Assisted Bridge Inspection,”
Journal of Intelligent & Robotic Systems, vol. 64, pp. 77–95, Oct. 2011.
[3] S. Campbell, W. Naeem, and G. Irwin, “A review on improving the
autonomy of unmanned surface vehicles through intelligent collision
avoidance manoeuvres,” Annual Reviews in Control, vol. 36, pp. 267–
283, Dec. 2012.
[4] R.-j. Yan, S. Pang, H.-b. Sun, and Y.-j. Pang, “Development and
missions of unmanned surface vehicle,” Journal of Marine Science and
Application, vol. 9, pp. 451–457, Dec. 2010.
[5] S. Brizzolara, T. Curtin, M. Bovio, and G. Vernengo, “Concept design
and hydrodynamic optimization of an innovative SWATH USV by CFD
methods,” Ocean Dynamics, vol. 62, pp. 227–237, Feb. 2012.
[6] J. Holtrop and G. Mennen, “A statistical power prediction method,”
International Shipbuilding Progress, vol. 25, no. 290, 1978.
[7] C. Sempere, Tecnologı́a de la defensa: análisis de la situación española.
Instituto Universitario General Gutiérrez Mellado, UNED, 2006.
[8] T. Fossen, Guidance and control of ocean vehicles. Wiley, 1994.
[9] A. Annamalai, R. Sutton, Chenguang Yang, P. Culverhouse, and
S. Sharma, “Innovative adaptive autopilot design for uninhabited surface
vehicles,” in Irish Signals & Systems Conference 2014 and 2014 China-
Ireland International Conference on Information and Communications
Technologies (ISSC 2014/CIICT 2014). 25th IET, pp. 158–163, June
2013.
[10] Shaorong Xie, Peng Wu, Yan Peng, Jun Luo, Dong Qu, Qingmei Li,
and J. Gu, “The obstacle avoidance planning of USV based on improved
artificial potential field,” in Information and Automation (ICIA), 2014
IEEE International Conference on, pp. 746–751, July 2014.