Você está na página 1de 6

In Proceedings of the 9 th IFAC Conference on Manoeuvring and Control of Marine Craft (MCMC2012), Arenzano, Italy, September, 2012.

Vector-product based Collision Estimation and Detection in e-Navigation.

L. P. Perera * and C. Guedes Soares *

*Centre for Marine Technology and Engineering (CENTEC), Instituto Superior Tecnico,Technical University of Lisbon Portugal (Tel: +351 21 841 7468; email:prasad.perera@mar.ist.utl.pt and guedess@mar.ist.utl.pt).

Abstract: This study focuses on the formulation of collision detection facilities among vessels that can be integrated into an e-Navigation strategy in maritime transportation. The detection of potential collision situations by relative motions of vessels that consist of state and parameter uncertainties in vessel maneuvering is considered in this study. A two vessel collision situation is presented and an extended Kalman filter is used to estimate the relative bearing and relative course-speed vectors between vessels. The collision detection process consists of cross and dot products among vessel velocity, bearing and heading vectors. Finally, a collision/near-collision situation is simulated and successful results on the detection of a potential collision situation with respect to vessel maneuvering are illustrated in this study.

keywords: e-Navigation, Ship Collision Detection, Extended Kalman Filter, Vector product, Curvilinear motion model, Vessel Relative Motion.

1. E-NAVIGATION

e-Navigation is a global vision that consists of proper integration of modern technologies not only to prevent collision and to enhance safety and security, but also to improve operating efficiency in maritime transportation. That could also be contributed towards vessel operational cost reduction (Mitropoulos, 2006). The e-Navigation vision was presented by the International Maritime Organization (IMO) and the International Association of Marine Aids to Navigation and Lighthouse Authorities (IALA) (IMO, 2007), which consists of the integration of modern technologies in real-time, the implementation of intelligent decision making capabilities (Perera et al., 2012) and the limitation of human subjective factors in maritime transportation. The e- Navigation concept is formulated by IALA is categorized as (IMO, 2007):

”The harmonized collection, integration, exchange, presentation and analysis of marine information onboard and ashore by electronic means to enhance berth to berth navigation and related services for safely and security at sea and protection of the marine environment”

The e-Navigation strategic plan was established by the IMO sub-committees on Safety of Navigation (NAV), Communications Security (COMSEC) and Standards of Training & Watchkeeping (STW), that has entrusted on the Norwegian Coastal Administration to coordinate the overall work program. That is formulated as a collaborated network of traffic information among vessels and shore based authorities. However, many independent research groups have also been working on advanced topics of e-Navigation to formulate better navigation platforms. Therefore, the main contribution this study can be categorized as filtering of

collision risk among vessels to enhance safety and security in maritime transportation.

The advantages in e-Navigation can be further elaborated as:

reduction of navigational errors, increase awareness of vessel situations, accumulation of traffic monitoring facilities, and reduce shipping costs (Ward and Leighton, 2010). Therefore, e-Navigation reduces navigators responsibilities and helps during the decision making process in maritime transportation. As core modern technologies are available, the major challenges on e-Navigation are to properly integrate them and to simplify interactions among navigators and shore based monitoring authorities. However, the integration among modern technologies is still under developed in maritime transportation, where a considerable amount of navigation functionalities is based an human interaction. As the congested sea routes force vessels to perform specific types of manoeuvres under varying sea conditions, the improvements in maritime transportation with the integration of modern technologies are demanded, where the human interaction alone cannot succeed.

To overcome some of the above challenges, an experienced navigator should have been placed on the bridge for a long period to satisfy vessel operational and navigation requirements (ie. to avoid potential hazard conditions of grounding, collision or near collision and weather damaged). However, that is not a realistic approach in ship navigation. As for the reported data 75-96% of marine accidents and causalities are caused by some type of human errors (Rothblum et al., 2002), where even an experienced navigator could make wrong navigation decisions that could result in human casualties and environmental disasters (i.e. the vessels Tricolor and Kariba collision in 2002 near north-south shipping route in English Channel (Kerckhof et al., 2004),

The Erika in 1999, The Prestige in 2002 (Eide et al., 2007) etc.

2. SHIP COLLISION DETECTION

Automatic Radar Plotting Aid (ARPA) and Automatic Identification System (AIS) have been developed to provide digital navigation aids for vessel detection and tracking. However, vessel states estimation, navigation trajectory prediction and collision detection facilities have been underdeveloped in maritime transportation. These facilities can be used for the prediction of vessel advance and future collision risk among vessels ahead of time. However, the assessment of collision risk is a continuous process to guarantee navigation safety. As illustrated in literature, the mathematical analysis of collision detection in maritime transportation can be divided into three categories (Imazu, 2006): closest point approach method (CPA : 2D method), predicted area of danger method (PAD : 3D method) and probabilistic method.

The CPA method consists of the calculation of the shortest distance between two vessels and the assessment of the collision risk, which has been predicted with respect to a one vessel domain. However, this method is not sufficient to evaluate the collision risk, since it does not consider vessel size, shape, course, and speed variations. An extensive study of the CPA method with respect to a two vessel collision situation has been presented by Kwik (1989). The PAD method consists of plotting one vessel’s possible trajectories as an inverted cone and the other vessel trajectories as an inverted cylinder. Hence, the region of both objects intersection is categorized as the Predicted Area of Danger. However, in this method vessel size, shape, course, and speed variation can be integrated into the geometry of the objects. In the probabilistic method, vessel collision situations are identified as probabilistic situations that are based on prior statistical data on maritime collisions. Further details on these probabilistic methods are presented by Rawson et al. (1998). Even though these prior data provides a realistic base in maritime collision, it is unreasonable to use for future collision predictions, because ship structures, operational procedures, and navigation equipments will be improved continuously (Pedersen, 2010).

However, collision conditions in these methods are simplified by assuming that vessels are moving in straight line motions and vessel estates/parameters are in deterministic conditions (i.e. fixed values in vessels speed and course conditions). Even though, these assumptions may be realistic for land and air transportation, maritime transportation is often under stochastic state/parameter conditions because of varying sea conditions(ie. variation in the vessel speed and course conditions). Therefore, the detection of collision situations under maritime transportation is a complicated process that needs advanced tools and technologies. This study proposes a methodology to detect collision and near collision situations by estimating a relative navigation trajectory, a relative course-speed vector and a relative bearing vector among vessels. The methodology can be used as a part of the e-

in

maritime transportation.

Furthermore, vessel manoeuvring under stochastic states/parameter conditions is considered in this study to implement realistic navigation conditions in maritime transportation. One should note that the study is limited to a two vessel situation, however this can be developed for a multi-vessel situation by accumulating multiple two vessels’ collision situations (Perera et al., 2012). Collision detection with respect to a two vessel collision situation consists of following steps: the observation of both vessels’ positions; the estimation of both vessels’ velocities, accelerations and navigational trajectories; the calculation of a relative navigation trajectory, a relative course-speed vector and a relative bearing vector of a one vessel with respect to another vessel position and orientation. The above calculations have been done by vector dot and cross product among respective vectors. The final calculation consist of the collision risk and the relative navigation trajectory of a one vessel with respect to one vessel heading, which can be used for decision making process of collision avoidance in maritime transportation.

Navigation

platform

that

is

used

to

avoid

collisions

In general, vessel navigators’ monitor collision situations by observing relative bearing of other vessels in open sea. The unchanged relative bearing of a vessel can lead to a collision situation. However, this requirement alone cannot predict accurate collision risk and that should not be used in decision making process under complex navigation conditions. Therefore, the observation of a relative navigation trajectory,

a relative course-speed vector and a relative bearing vector

among vessels should be used to improve the detection of collision situations as proposed in this study. The collision risk between two vessels is categorized as a vector dot product between a relative course-speed vector and a relative bearing vector between vessels. One should note that the relative navigation trajectory of a vessel with respect to

another vessel position can be calculated considering both vessel positions (Perera and Guedes Soares, 2012). However, the relative navigation trajectory of one vessel with respect to another vessel orientation (ie. heading) will complicate the calculation process due to vessel heading changes. Therefore,

a comprehensive mathematical process that consists of vector dot and cross product is considered in this study.

It is assumed that the vessel positions are determined by conventional AIS and GPS systems. However, there are many challenges that can be observed in these measurements:

firstly, the AIS and GPS position signals can be associated with sensor noise and/or errors that degrade the vessel position accuracy; secondly vessels are manoeuvring under varying sea conditions, where vessel kinematics and dynamics are associated with stochastic state/parameter conditions. Those conditions are categorized as state/parameter uncertainties in this study. Hence, a proper mechanism to estimate vessel states (ie. position, velocity and acceleration) has been presented. An extended Kalman filter (EKF), one of the well known estimation algorithms to overcome the previously discussed challenges and to estimate the accurate vessel states is considered in this study. One should note that the state estimation is based only on vessels’

position measurements. The main contribution of this study can be categorized as estimating a relative navigation trajectory, a relative course-speed vector and a relative bearing vector and calculating the collision risk between vessels, wild could be used in the e-Navigation formulations.

3. ESTIMATION OF COLLISION VECTORS

A two-vessel collision situation is presented in Figure 1. One

vessel is located in position O (x o (t), y o (y)) and another

vessel that is heading for a collision with vessel O is located

at position A (x a (t), y a (t)). Vessel O speed and course are

represented by V o (t) and χ o (t) respectively. Vessel A speed and course are represented by V a (t) and χ a (t) respectively. The x and y velocity vectors for vessel O and A are presented

by v xo (t), v yo (t), v xa (t) and v ya (t) respectively. The normal and tangential acceleration vectors for vessel O and A are presented by a no (t), a to (t), a na (t) and a ta (t) respectively. The relative course-speed vector of vessel A with respect to vessel

O is represented by V i,o (k). The estimated bearing vector of

vessel O and vessel A with respect to XYZ coordinate system

is presented by B o (k) and B a (k) that are calculated by discreet-

time instants. The vessel O heading vector is presented by H o (k). The relative bearing vector of vessel A with respect to vessel O is presented by B ao (k).

with respect to vessel O is presented by B a o (k). Fig. 1. A two

Fig. 1. A two vessel collision situation

3.1

System Model

To capture the vessel manoeuvring trajectory, a continuous- time curvilinear motion model is proposed in this study; a

standard continuous-time curvilinear motion model for vessel

O can be written as:

χ &

o

(t) =

v xo

(t)

=

a

&

(t)

V (t) cos

; V

(t)

no

V

o

o

o

(t)

(χ

o

=

a

to

(t)) v

;

(t)

yo

(t)

=

V

o

(t)

sin

(χ

o

(t))

(1)

The vessel motion model presented in (1) can be summarized into a system model that can be written as:

x & (t)

x

=

f(x (t))

x

+

E [w

w

x

E[w (t)]

x

=

0

,

x

(t)

(t);w

x

(t)]

=

[Q (t)]

o

(2)

where the system states, x x (t), and the system function, f(x x (t)), can be further described as:

x

x

(t) =

x

v

y

v

a

a

o

(t)

xo

(t)

o

(t)

yo

to

no

(t)

(t)

(t)

,

f(x

x

(t))

=

v

a

v

a

0

0

xo

to

yo

to

(t)

(t)f

(t)

(t)f

vxo

vyo

v (t) f vxo xo vyo = , f = 2 2 v (t) +
v
(t)
f vxo
xo
vyo
=
,
f
=
2
2
v
(t)
+ v
(t)
xo
yo

+

a

a

no

no

(t)f

(t)f

vyo

vxo

v (t) yo 2 2 v (t) + v (t) xo yo
v
(t)
yo
2
2
v
(t)
+
v
(t)
xo
yo

(3)

and w x (t) is a white Gaussian process noise, a to (t) is tangential acceleration, and a no (t) is normal acceleration. The Jacobian of the system function, f(x x (k)) in (3) can be expressed as:

x

0

0

0

0

0

0

x

(

f

(

x

x

(t)

))

a

a

to

to

(t)f

(t)f

vxo

vxo

vyo

vxo

=

1

0

+

0

0

a

a

no

no

(t)f

(t)f

vyo

vxo

vxo

vxo

0

0

0

0

0

0

a

a

to

to

(t)f

(t)f

vxo

vyo

vyo

vyo

0

1

+

0

0

a

a

no

no

(t)f

(t)f

vyo

vyo

vxo

vyo

f

f

0

vxo

0

vyo

0

0

0

f

vyo

f

0

vxo

0

0

(4)

where the respective functions in (4) can be written as:

as

f

vxo

vxo

f

vyo

vxo

v

yo

2

( )

t

=

(

v

xo

2

( )

t

v

+

xo

v

yo

( )

t

2

v

( )

t

)

3 / 2

yo

( )

t

,

=

(

v

xo

2

( )

t

+

v

yo

2

( )

t

)

3 / 2

,

3.2 Measurement Model

f

vxo

vyo

=

f

vyo

vyo

=

v

yo

( )

t

v

xo

( )

t

(

v t

xo

2

( )

v

+

xo

v

2

yo

2

( )

t

( )

t

)

3 / 2

(5)

(

v t

xo

2

( )

+

v

yo

2

( )

t

) 3 / 2

The measurement model observes the actual positions of the vessels. This can be formulated by AIS, GPS and Radar systems. A discrete-time measurement model is considered due to the availability of the vessel position values in discrete time instants. It is assumed that the correlations between the position measurements are negligible. The position measurements of vessel O can be written as:

z(k)

E[w (k)]

=

z

h(x(k))

=

,

0

w (k),

+

E [w (k);w

k

z

z

= 1 2

z

,

(k)]

,

=

[R(k) ]

(6)

where the system states, z z (k), and measurement function, h(x x (k)), can be further described as:

z

z

(k) =

z

z

x

y

(k)

(k)

;

h(x

x

(k)) =

x

o

(k)

0

0

0

y

0

o

(k)

0

0

0

0

0

0

(7)

where z x (k) and z y (k) are the measurements of x and y position values of vessel O, and w z (k) is a white Gaussian measurement noise. The Jacobian matrix of the measurement function, h(x x (k)) in (7), can be written as:

x

x

(

h

(

x

x

(k)

))

=

1

0

0

0

0

1

0

0

0

0

0

0

(8)

The negligible correlations between system noise and measurement noise are assumed as:

E[w

x

(t); w

z

(k )]

=

0

for all k , t

(9)

A similar mathematical formulation has been used for the

motions in vessel A.

3.3 Extended Kalman Filter

The EKF (Gelb et al., 2001) is proposed in this study as the

estimation algorithm for vessel states of positions, velocities and accelerations. The estimated vessel states are used to calculate a relative navigation trajectory, a relative course- speed vector and a relative bearing vector with respect to a one vessel position. Even though, vessel positions are measured with noisy data, the estimation algorithm has improved the position accuracy. The implementation of the EKF is summarized in following section. The error condition

of the system states can be written as:

~

x (k)

x

=

ˆ

x (k)

x

x (k)

x

(10)

where x(t)

is the state error vector and xˆ(t) is the estimated

states of the system. The initial system state conditions can

be considered as:

~

x x

(

0 )~N

(x ˆ

x

(

0

),P(

0

))

(11)

where P(0) is the initial state covariance, describing the uncertainty present on the initial estimates. The state estimation propagation can be written as:

d

dt x

ˆ

x

(k)

=

ˆ

f(x (k))

x

(12)

The error covariance extrapolation can be written as:

d

dt

P(k)

=

ˆ

F(x (k))P(k)

x

+

P(k)F

T

ˆ

(x (k))

x

+

Q(k)

where the function of F(x (k)) can be considered as:

ˆ

x

ˆ

F(x (k))

x

=

x (k)

x

f(x (k))

x

x

x

(k)

= ˆ

x

x

(k)

(13)

(14)

and P(k) is the estimated error covariance with the state estimate update. At each step, after measurement data is available from the sensors, the system state can be estimated as:

x ˆ (k

x

+

)

=

ˆ

x (k

x

)

+

K(k)[z(k)

h

k

(x

ˆ

x

(k

))]

(15)

where x x (k - ) and x x (k + ) are the prior and posterior estimated system states respectively, and K(k) is the Kalman gain. The error covariance update can be written as:

P(k

+

) =

1

ˆ

[

K(k)H (x (k

x

))] P(k

)

(16)

where P(k - ) and P(k + ) are the prior and posterior error

covariance

system respectively

function

of the

with

the

ˆ

H( x (k))

x

can be written as:

ˆ

H( x (k))

x

=

x (k)

x

h(x (k))

x

x

x

(k)

=

ˆ

x

x

(k)

(17)

The Kalman gain can be written as:

K(k)

= P(k

)H

(

x ˆ

x

(

k

))

H

(

x ˆ

x

(

k

))

P(k

)H

(

x ˆ

x

(

k

))

T

+ R(k)

1

(18)

4. COLLISION DETECTION BY VECTOR PRODUCT

The relative trajectory of vessel A with respect to vessel O is calculated by the relative x and y positions that can be written as:

xˆ

ao

= xˆ

xˆ

a

xˆ

ao

o

;

yˆ

ao

yˆ

= yˆ

ao

a

yˆ

o

(19)

are the estimated relative x and y

positions of vessel A with respect to vessel O. The relative course-speed vector components of vessel A can be written as:

where

and

vˆ

xao

=vˆ

xa

vˆ

xo

;

vˆ

yao

=vˆ

ya

vˆ

yo

(20)

represent the estimated x and y relative

velocity components of vessel A respectively. Hence, vessel A estimated x and y relative velocity components are used for calculating the relative course-speed vector. The estimated relative bearing vector of vessel A with respect to vessel O position can be written as:

where

vˆ

xao

and

vˆ

yao

B ao

(k) = [xˆ

ao

(k)

yˆ

ao

(k)

0]

(21)

where the unit vector of (21) can be written as:

B ( k ) = B ( k ) B ao ao ao A unit
B
(
k
)
= B
(
k
)
B
ao
ao
ao
A unit vector G
(k )
ao

(

k

)

2

=

[ b

aox

(

k

)

b

aoy

(

k

)

0]

that is perpendicular to vector B

ao

(22)

(k)

can be written considering the vector cross product as:

G

o

=

Z (k)

B

ao

(k)

=

[

b

aoy

(k)

b

aox

(k)

0]

(23)

where the unit vector in Z axis can be written as:

Z (k ) = [0

0

1]

(24)

Hence, a vector with the vessel domain magnitude and vector

G

o

(k)

direction can be written as:

G

o

(k)

=

R

vd

G

o

(k)

(25)

The boundaries of vessel domain in vessel O with respect to

vessel A is identified by two vectors of which can be written as:

,

F

oa

(k)

and C

oa

(k)

F

C

oa

oa

(

(

k

k

)

)

G

=

=

k

(

k

)

o

G

o

(

)

B

ao

B

(

k

)

ao

(

k

)

(26)

where the unit magnitude vectors of (18) can be written as:

F

oa

C

oa

(

(

k

k

)

)

= F ( k ) F ( k ) oa oa 2 = C (
=
F
(
k
)
F
(
k
)
oa
oa
2
=
C
(
k
)
C
(
k
)
oa
oa
2

(27)

The estimated relative course-speed vector of vessel A with respect to vessel O can be written as:

V

ao

(k)

=

[vˆ

xao

(k )

vˆ

yao

(k)

0]

(28)

where the unit vector of (28) can be written as:

V k

ao

( )

=

V

ao

(

k

( k ) v ˆ yao ( k ) 0 ] (28) where the unit vector

V

ao

(

k

)

2

(29)

Hence, the collision risk between two vessels is calculated as:

and

. This condition can be mathematically formulated as

a vector product of:

vector V

C

oa

(k)

oa

(k)

is located between the vectors of F

oa

(k)

(

F

oa

(

C

oa

(

(

k

k

)

)

V

oa

V

oa

(

(

k

k

)

)

))

Z

(

(

k

)

Z

(

k

0

))

0

(30)

Hence, the vessel A relative course-speed vector with respect

to the relative bearing vector is categorized as the collision

risk (CR) between vessels and that can be written as:

CR (k ) = B

oa

(k ) V

oa

(k ) 0

(31)

The relative motion of vessel A with respect to the heading of

vessel O is further discussed in this section; that can be used

in the decision making process of collision avoidance. A

vector perpendicular to vessel O heading can be written as:

K

o

(

k

) =

H

o

(

k

)

Z

(32)

where the unit heading vector of vessel O can be written as:

H

o

( k

) =

H

H o o 2
H
o
o
2

= [

h

ox

h

oy

0]

(33)

where the unit vector of (32) can be written as:

K

o

( k

)

= K

o

(

k

)
)

K

o (

k

)

2

= [

k

ox

k

oy

0]

(34)

The heading of vessel O can be transformed for the X axis considering the vector summation of:

o

X (k)

=

H

o

(k)

+

o

K

o

(k)

(35)

(36)

(37)

can be derived

(38)

(39)

where the unit vector in X axis can be written as:

X (k ) = [1

0

0]

and heading vector can be written as:

H

o

( k

)

= R

vd

H

o

(

k

)

=

[

h

ox

h

oy

0]

From (35) the constants of as: and (k) o (k) o ( k ) =
From (35) the constants of
as:
and
(k)
o (k)
o
(
k
)
=
h
(
k
)
+
(
k
)
k
(
k
)
o ox
o
ox
0
=
h
(
k
)
+
(
k
)
k
(
k
)
oy
o
oy
That can also be written as:
(
k
)
= h
(
k
)
h
(
k
)
k
(
k
)
k
(
k
)
o
ox
oy
ox
oy
(
k )
=
h
(
k
)
k
(
k
)
o
oy
oy

A unit vector perpendicular to the vessel A relative bearing

can be calculated as:

J

o

(

k

) =

B

ao

(

k

)

Z

(40)

Hence, the vessel A position with respect to the vessel O heading can be formulate in re-orientation of:

B

o

(k)

=

R

vd

B

ao

(k)

+

o

(k)J

o

(k)

(41)

where, the respective unit vector of (41) can be written as:

B

o

( k

)

= B

o

(

k

)
)

B

o (

k

)

2

(42)

The estimated relative bearing vector of vessel A with respect

to the vessel O heading can be written as (see Figure 1):

P

o

(

k

)

=

sgn(

o

)

B

ao

(

k

) 2 B o
)
2 B
o

(

k

)

(43)

5. COMPUTATIONAL SIMULATION

Computational simulation of a two vessel near collision situation is presented in Figure 2. The figure represents vessel O and vessel A actual (Act.), measured (Mea.) and estimated (Est.) navigation trajectories. The vessel position measurements are generated by adding sensor noise into the actual trajectory. As presented in the figure, one near collision point can be noted. A zoomed view of the same trajectories around the near collision point is also presented. A relative navigation trajectory of the same situation with respect (w/r) to vessel O position is presented in Figure 3. One should note that the intersection of two trajectories should not be detected as a collision point because each vessels can pass the collision point in different time intervals. However, this confusion can be clarified by observation of relative navigational trajectories; where a relative trajectory of vessel A is propagated near vessel O initial position and intercept the vessel domain that must be a near collision situation. In Figure 3, each estimated relative position of

vessel A consists of a relative course-speed vector that is presented by an arrow. This representation is an important for detecting a collision situation even under near collision

conditions.

a collision situation even under near collision conditions. Fig. 2. Navigation trajectories vessel A and vessel

Fig. 2. Navigation trajectories vessel A and vessel O

Fig. 2. Navigation trajectories vessel A and vessel O Fig. 3. Relative trajectory vessel A w/r

Fig. 3. Relative trajectory vessel A w/r vessel O position

A relative navigation trajectory of the same situation with

respect to vessel O heading is presented in Figure 4. The

2 (rad) is

assumed in this simulation. One should note that the near

collision situation is further clarified in Figure 4, where each estimated relative position of vessel A is presented by a dot. This representation is important for the decision making process of the collision avoidance process. Furthermore, the collision risk with respect to the relative distance between vessels is presented in Figure 5, where the collision risk can

vessel O heading is varying from 0 (rad) to

risk can vessel O heading is varying from 0 (rad) to be observed long before the

be observed long before the collision situation that will help

navigator to make proper collision avoidance decisions.

help navigator to make proper collision avoidance decisions. Fig. 4. Relative trajectory vessel A w/r vessel

Fig. 4. Relative trajectory vessel A w/r vessel O heading

4. Relative trajectory vessel A w/r vessel O heading Fig. 5. Collision risk 6. CONCLUSIONS A

Fig. 5. Collision risk

6. CONCLUSIONS

A methodology for detecting collision risk under

state/parameter uncertainties in vessel manoeuvring is presented in this study. As presented in the simulations the

calculation of a relative navigation trajectory, a relative course-speed vector and a relative bearing vector of a one vessel with respect to another vessel position and orientation

can be used for detecting collision situations ahead to time in

ship navigation. Similarly, this prior knowledge in collision

risk among vessels can be used for decision making process

of maritime transportation.

ACKNOWLEDGEMENTS

This work has been made within the project ”Methodology for ships maneuverability tests with self-propelled models”,

which is being funded by the Portuguese Foundation for Science and Technology under contract PTDC/TRA /74332/2006. The research work of the first author has been supported by a Doctoral Fellowship of the Portuguese Foundation for Science and Technology under contract

SFRH/BD/46270/2008.

REFERENCES

Eide, M.S., Endresen, O., Brett, P.O., Ervik, J.L., and Roang, K., (2007) “Intelligent ship traffic monitoring for oil spill prevention: Risk based decision support building on AIS,” Marine Pollution Bulletin, vol. 54, pp. 145–148. Gelb, A., Kasper Jr. J. F., Nash, Jr. R. A., Price, C. F., and Sutherland, Jr. A. A., (2001) Applied Optimal Estimation. MA. USA: The MIT Press. Imazu, H., (2006) “Advanced topics for marine technology and logistics,” Lecture Notes on Ship collision and integrated information system, Tokyo University of Marine Science and Technology, Tokyo, Japan. IMO, (2007) “Development of an e-Navigation Strategy,” Report of the Correspondence Group on e-Navigation,

NAV/53/13.

Kerckhof, F., Roose, P., and Haelters, J., (2004), “The tricolor incident : From collision to environmental disaster,” Atlantic Seabirds, vol. 6, no. 3, pp. 85–94. Kwik, K.H., (1989), “Calculations of ship collision avoidance manoeuvres: A simplified approach,” Ocean Engineering, vol. 16, no. 5/6, pp. 475–491. Mitropoulos, E.E., (2006) “Aids to navigation in a digital world,” in Keynote address on the 16th IALA/AISM conference, Shanghai, China. Pedersen, P.T., (2010) “Review and application of ship collision and grounding analysis procedures,” Marine Structures, vol. 23, pp. 241–262. Perera, L.P., Carvalho, J.P., and Guedes Soares, C., (2012) “Intelligent ocean navigation and fuzzy-bayesian decision-action formulation.” IEEE Journal of Oceanic Engineering, vol 37, no 2, 2012, pp 204-219. Perera, L.P., and Guedes Soares, C., (2012), ”Detections of potential collision situations by relative motions of vessels under parameter uncertainties.”, Sustainable Maritime Transportation and Exploitation of Sea Resources, E. Rizzuto & C. Guedes Soares (Eds.), Taylor & Francis Group, London, UK, pp 705-713. Rawson, C., Crake, K., and Brown, A., (1998) “Assessing the environmental performance of tanker in accidental grounding and collision,” SNAME Transactions, vol. 106, pp. 41–58. Rothblum, A.M., Wheal, D., Withington, S., Shappell, S.A., Wiegmann, D.A., Boehm, W., and Chaderjian, M., (2002), “Key to successful incident inquiry,” in 2nd International Workshop on Human Factors in Offshore Operations (HFW), pp. 1–6. Ward, N., and Leighton, S., (2010), “Collision avoidance in the e-navigation environment,” in In: 17th Conference of the International Association of Marine Aids to Navigation and Lighthouse Authorities, Cape Town, South Africa.

Interesses relacionados