Model-Based Development and Simulation For Robotic Systems With Sysml, Simulink and Simscape Profiles
Model-Based Development and Simulation For Robotic Systems With Sysml, Simulink and Simscape Profiles
Model-Based Development and Simulation For Robotic Systems With Sysml, Simulink and Simscape Profiles
DOI: 10.5772/55533
© 2013 Rahman and Mizukawa; licensee InTech. This is an open access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
www.intechopen.com Mohd Azizi Abdul Rahman and Makoto Mizukawa: Int Model-Based
J Adv RoboticDevelopment
Sy, 2013, Vol.and
10,Simulation
112:2013 1
for Robotic Systems with SysML, Simulink and Simscape Profiles
approaches to computerized model‐based approaches [3]. aspects need a new conceptual framework for multi‐
In addition to MBSE, a graphical‐based language for system integration in the later stages of development.
enabling systems engineering activities, the OMG Recent contributions to the development of robot systems
SysML™ [4] has been standardized based on UML [5] in that use the potential of SysML language are presented in
support of the MBSE framework. Using MBSE in systems [11] for the design of a space robotics system, in [12] for
design, engineers can solve systems engineering the software development process of a mobile robot and
problems by expressing multiple views that can in [13] for a SysML‐based robot systems design for
transform stakeholders’ requirements into detailed manipulation tasks. We seek to investigate the potential
specifications with the aid of models [6]. uses of SysML in the field of robotics, control and systems
engineering.
In this paper, we propose a comprehensive system‐level
design workspace for robotics in conjunction with the SysML was developed as a general‐purpose modelling
simulation of continuous system dynamics. The primary language that is only capable of descriptive semantics.
modelling languages are introduced, such as SysML™ Executable semantics are required for the design of the
for systems engineering activities, Simscape™ [7], the simulation and its analysis. Associating executable
declarative language for physical systems modelling and semantics with SysML descriptive semantics can be
also Simulink® [8], Matlab’s extension product that realized by developing an extension to the UML model
allows the modelling and simulation of continuous element [14]. Johnson et al. [15‐16] proposed a formal
processes using block diagrams. For this study, the UML technique for modelling continuous system dynamics in
modelling tool, the IBM® Rational® Rhapsody® SysML through language mapping between SysML
software [9] is used to design system‐level models based constructs and the declarative Modelica [17] language.
on the SysML profile. The simulation environment is set The mapping method uses the extension of SysML model
to Simulink. In brief, the system model of the robot is first elements through the stereotypes mechanism rather than
described using SysML to represent the continuous‐time formulating totally new language constructs. As a result,
dynamics behaviour of the system. A new set of compatible SysML constructs are established for selected
stereotypes is created in order to provide distinctive Modelica elements in representing the continuous
model semantics in support of Simulink‐ and Simscape‐ dynamics modelling constructs, as well as their
compatible models. A subsystem that contains simulation model in SysML. Venderperren and Dehaene
components, for example, actuators, sensors, and the base [18], and Kawahara et al. [19] discussed a co‐simulation
platform is then modelled based on a physical system approach using UML/SysML and Matlab/Simulink, in
modelling approach using a Simscape language. Lastly, which a software‐based coupling tool is used to combine
all the models are integrated and synchronized with the the UML/SysML model and the simulation model in
system‐level simulation model and be executed in the Simulink.
Simulink window automatically.
For mechatronic system behaviour, Turki and Soriano
The remainder of the paper is organized as follows. Some [20] proposed a method for representing continuous
related works are reviewed in section 2 which consider dynamics behaviour using a bond graph based on the
continuous behaviour modelling and systems model extension of SysML activity diagrams. Furthermore, Cao
integration of different languages. Section 3 describes the et al. [14] recently introduced a mixed‐behaviour system‐
proposed method with the aid of a case study. The level design using SysML, and Stateflow and Simscape
modelling of the continuous system dynamics and the for system‐level simulation. In addition, a model
whole model composition are presented in greater detail semantic extension is carried out based on the stereotypes
in section 4. Section 5 briefly discusses model integration mechanism [21]. Based on the above information, we
and how the models are kept in sync, including believe that the system‐level design and simulation
implementation and simulation results. Last but not least, technologies which use the potential of SysML, Simulink
section 6 gives a summary of the proposed method and and Simscape could offer promising solutions for robotics
describes some possible future works. and its applications.
2. Related Research 3. Method
For this section, related works on continuous behaviour This section describes the methodology used in the
modelling, model semantics extension and systems system‐level modelling, simulation and control of a
model integration of different languages are reviewed. wheeled mobile robot. From control engineering
According to Bruninyckx [10], robotics is a field of perspectives, the design of control dynamic systems
integration rather than a fundamental science, so it works well using a model‐based design (MBD) approach
contributes to complex systems integration. In that sense, with a widely‐used design software and simulation, (e.g.,
robotic systems that typically consist of multi‐domain Matlab/Simulink). Simulink uses block diagrams, whose
3.2 An example system where, (x, y) denotes the robot’s position in the X‐Y
reference frame. The measured variables (x, y, θ) will be
The physical system and its behaviour are usually formed compared to the desired input and fed to the controller as
in terms of differential equations. In this paper, a the error term (e.g., e = [ex, ey, eθ]T = [xd‐x, yd‐y, θd‐θ]T). The
simplified closed‐loop system, as shown in Fig.1, is used control algorithm is simplified by the following
as a case study. The system simply consists of two main equations:
subsystems and several components: the robot base (i.e.,
including its body dynamics and kinematics components) de
with embedded actuators and sensors (e.g., both u l K 1 e K 2 e dt K 3 (7)
dt
translational and rotational sensors), a control system and
the desired input in the form of a position in the X‐Y de
u r K 1 e K 2 e dt K 3 (8)
coordinate system and the desired orientation (e.g., xd, yd, dt
θd). The robot base is a symmetrical round‐shaped mobile
platform with two actuated wheels and one passive where, ul and ur denote the controller left‐ and right‐
wheel (not shown in the figure) that is placed in front of output, respectively. K1, K2, and K3 represent the tunable
the platform for stability. coefficients for controller gains. The objective of the
controller is to provide appropriate torques (tl, tr) to be
The physical properties of the robot platform are the mass applied to the robot for motion. Therefore, the robot will
(m), the radius of robot’s base from the centre of the robot be controlled based on the torques applied to the left‐ and
platform (rc), the robot’s wheel radius (rw), the robot’s right‐side of the wheels.
www.intechopen.com Mohd Azizi Abdul Rahman and Makoto Mizukawa: Model-Based Development and Simulation 3
for Robotic Systems with SysML, Simulink and Simscape Profiles
4. System Definition View in SysML method because it is more easily implemented in our
current SysML tool.
The SysML’s block is the most utilized model element for
describing both logical and physical objects of complex Based on the above analysis, we choose to model the
systems [4]. It is the basic construct of the SysML model. system in SysML using a component‐based approach, in
For conceptual system‐level definition, the closed‐loop which the closed‐loop system is first decomposed into
system as shown in Fig.1 is firstly defined in a several components with their associated functions. The
hierarchical way using the SysML block definition extension of the model is then implemented on SysML
diagram (BDD). This is shown in Fig.2 where the blocks to design the Simulink‐ and Simscape‐compatible
‘ClosedLoopSystem’ block is viewed as a basis for the models. Thus, a set of new stereotypes is defined to
simulation of the main structure of the system. As we can associate the SysML constructs with the new model
see, the ‘ClosedLoopSystem’ block is composed of one or element capability. As a result, the design and analysis
more ‘SimulinkSubsystem’, and one or more models of the system are prepared and we apply the
‘SimscapeSubsystem’. Note that the subsystem block is physical system modelling approach to construct an
directly composed to the system block by using the black‐ energy‐flow physical system network. The physical
diamond‐arrow relation. Down to the component level, modelling network is considered as a group of
the ‘SimulinkSubsystem’ is made up of zero or more interrelated subsystems and/or components that are
‘Input’ and one or more ‘Controller’ blocks. The linked to one another through their energy ports [25].
‘SimscapeSubsystem’ is made up of one ‘Body’, one or
more ‘Actuator’, one or more ‘Sensor’ and zero or more 4.1 SysML‐Simulink Model Compatibility
‘Kinematics’ blocks.
In this subsection, the definition of the SysML‐Simulink
«block» compatibility model is illustrated with regards to the
ClosedLoopSyst em
input and controller components.
::SimulationContext::
4.1.1 SimulinkBlock (SB) Stereotype
subsys1 1..* subsys2 1..*
«block» «block» This section briefly describes the concept of the SysML
SimulinkSubsyst em SimscapeSubsystem
model extension in defining the SysML‐Simulink
comp1 * comp2 1..* comp3 1 comp4 * comp5 1..* comp6 1..*
compatible model. As mentioned earlier in section 4, it is
«block» «block» «block» «block» «block» «block»
based on the stereotype mechanism that tailors the
Input Controller Body Kinematics Sensor A ctuator SysML meta‐model to create a new model element. The
concept of meta‐modelling is briefly described in [24].
Figure 2. Conceptual system definition view The element that needs to be stereotyped is a standard
SysML block taken from the meta‐model. Fig.3 shows the
The SysML descriptive model needs to be defined as definition of a new stereotype extended to support
executable models, which can then be translated into the Simulink models. It introduces two blocks associated
analysis model for use in the simulation tool. Toward this with a specialization relation (the solid arrow), known as
end, SysML model definition and its semantic extension the extension relationship, (e.g., <<extend>>). It should
should be done in order to support the behavioural be noted that the standard SysML block now becomes a
design models and the integration of a simulation model. special block that happens to be a Simulink reference
For extension of the SysML model, we use the stereotype model representation. Specifically, the semantics of the
mechanism [21, 24]. The stereotype mechanism is widely SB, when it is applied to a SysML block, will be referred
adopted to semantically extend the SysML model and is a to a Simulink model encapsulation. Therefore, the SB
powerful way to define a new SysML model element by represents a wrap model in Simulink. Apart from
tailoring the UML meta‐model [24]. Stereotypes are semantic extension, the SB stereotype declares tagged
visually spotted by looking at the keyword within a set of values with which to indicate auxiliary properties of the
double chevrons (e.g., <<stereotype>>). Simulink model. For example, the ‘URI’ property that
provides a link to the Simulink model files. Optionally,
A model semantic extension can be carried out using two tagged values can be used to add information to the
techniques [14]: either the “heavy‐weight” or the “light‐ model base, specific to the domain or platform [26].
weight” approach. By comparison, the first approach
develops totally new constructs for the SysML profile, Fig.4 shows a BDD that illustrates the “Input” and
while the second technique simply creates a new profile “Controller” blocks. As we can see, both components
of stereotypes based on the UML’s meta‐class for apply the SB stereotype. Equations (7)‐(8) are properly
extending the SysML model element [20]. However, the described in the “Controller” block (shown by the use of
first method requires a distinctive tool to support the new the stereotype, <<Algorithm>>) along with its gain
construct [21]. For this reason, we apply the second parameters, error variables (shown by the stereotype,
www.intechopen.com Mohd Azizi Abdul Rahman and Makoto Mizukawa: Model-Based Development and Simulation 5
for Robotic Systems with SysML, Simulink and Simscape Profiles
Matlab class declaration method, includes member between the function setup process and the variable
classes such as nodes, inputs and/or outputs, parameters, initialization process. For example, the constraint “R.w
and variables for both domain and component files. Note == theta.der”, represents the time derivative of the
that for the domain file, only variables and sometimes orientation angle, which gives a solution to the angular
parameters (optional) are declared so that their function is velocity variable, on the other hand, the constraint
to establish a domain rather than a detailed component. “theta == init_theta”, declares the initial condition
In contrast, the implementation section includes only the setup for the orientation angle. In Fig.6, the kinematics,
setup function and equations for the component file. In the actuator and sensor blocks are stereotyped with
setup function, the component is prepared for simulation <<Comp>> similar to the ‘Body’ block. The kinematics for
and it is executed once for each component during the the system is described based on equations (5) and (6).
model compilation process whereas the equations section The actuated wheels receive control outputs through the
declares the component equations that are executed signal ports (i.e., tr and tl), and apply the torque to both
throughout the simulation. Therefore, the implementation wheels of the robot via the energy port, T. The sensor can
section is utilized to establish an equation‐based internal be used to detect the state variables changes, and send
behaviour of the system. More detail on the Simscape them as data via the signal port. For the Simscape
model structure, can be found in references [7] and [27]. subsystem block, the stereotype <<SimscapeBlock>>
(see Fig.8) is introduced as an extended SysML block that
Although Simscape is conceptually more specific to the applies the same concept and property for the SB.
physical system theory, it can fit to the general concept Therefore, all Simscape‐related blocks can also apply the
used in Matlab/Simulink. Thus, it allows for Simscape SB stereotype.
models to be combined with other models in the same
«Domain» «Domain»
Simulink environment. Ro tat io na L T r a nsla tiona L
Attributes Attributes
4.2.2 SysML Model Extension for Simscape
«Through» t:real_T=N*m «Through» f :real_T=N
«Across» w:real_T=rad/s «Across» v :real_T=m/s
The SysML block is again semantically extended to
represent the domain and component model in Simscape. «bl ock,Comp»
Since the domain model has no internal behaviour, the B ody
A SysML block has value properties, which are referred «Init» init_s:real_T=m
«Init» init_theta:real_T=rad
to as value types. These are utilized to define system
«Across» w:real_T=rad/s
variables, parameters, constants and other properties of «Across» v :real_T=m/s
component models in Simscape. The stereotypes such as «Across» s :real_T=m
<<Across>> and <<Through>> are created for «Across» theta:real_T=rad
Constraints
The nodes, inputs and outputs of the component are «Setup» through(f ,T.f ,[ ]);
modelled by utilizing the SysML’s flow port. In SysML, «Setup» through(t,R .t,[ ]);
the ‘eneryport’ is defined to replicate the Simscape bi‐ «Setup» ac ross(v ,T.v ,[ ]);
directional conserving port which allows only energy‐ «Setup» ac ross(w,R .w,[ ]);
«Setup» s = init_s ;
based elements to pass through it. It is based on the non‐
«Setup» theta = init_theta;
atomic flow port where multiple items can exchange «Equations» T.v = = s.der;
energy through the ports. As mentioned in 4.2.1, the setup «Equations» R.w = = theta.der;
and equations properties are for the implementation part «Equations» m*T.v .der = = f ;
of the Simscape model. They are represented as «Equations» 2*Ib*rc *R.w.der/wd = = t;
constraints in SysML and describe the syntax with which
the attributes of objects should comply. Thus, the Figure 5. Definition of Simscape domain models and the body
stereotype <<Init>> has been defined to distinguish component
Ports
Ports
Ports
<<extend>>
energyport_R:in_out
energyport_R:in_out energyport_R:in_out
energyport_T:in_out
energyport_T:in_out energyport_T:in_out
signalport_tl:In
signalport_x:Out signalport_A:Out «Stereotype»
signalport_tr:In
signalport_y:Out signalport_S:Out StructuredSimulinkBlock
Constraints
Constraints Constraints
«Setup» across(v,T.v,[ ]);
«Setup» through(f,[ ],T.f);
«Setup» s = init_s; Tags
«Setup» through(t,[ ],R.t);
«Setup» across(w ,R.w ,[ ]); «Setup» theta = init_theta;
«Setup» s = init_s;
«Equations» f = = (tr+tl)/rw ;
«Equations» t = = (tr - tl);
«Equations» theta.der = = R.w ; DiagramForSimulink:RhpString
«Setup» theta = init_theta; «Equations» s.der = = T.v;
«Equations» T.v = = s.der; «Equations» A = = theta;
MatlabExePath:RhpString
«Equations» R.w = = theta.der;
«Equations» x = = s*cos(theta);
«Equations» S = = s;
Parameters:RhpString
«Equations» y = = s*sin(theta);
SampleTime:RhpString=0.01
SolverName:RhpString
Figure 6. Definition of kinematics, actuator, and sensor SolverType:RhpString
components StartTime:RhpString=0.0
StopTime:RhpString=10.0
4.3 System‐level Model Composition
4.3.1 StructuredSimulinkBlock (SSB) Stereotype
«block,StructuredSimulinkBlock»
create a new model element with which to represent the ::SimulationContext::
composition model of the main structure (i.e., the entire
subsys1 1..* subsys2 1..*
closed‐loop system). The Simulink and Simscape models
«block,Simulink Block» «block,SimscapeBlock»
need to be encapsulated as parts of the main structure SimulinkSubsyst em SimscapeSubsystem
block. To this end, a new class model is created to
represent another Simulink sub‐model. Fig.7 introduces comp1 * comp2 1..* comp3 1 comp4 * comp5 1..* comp6 1..*
the definition of the SSB which is extended from the «block,Comp»
Input
«block,Comp»
Controller
«block,Comp»
Body
«block,Comp»
Kinematics
«block,Comp»
Sensor
«block,Comp»
A ctuator
standard SysML block applicable to a new class model.
In this case, the SSB applies the tagged value Figure 8. A remodelled BDD with the new stereotypes applied
www.intechopen.com Mohd Azizi Abdul Rahman and Makoto Mizukawa: Model-Based Development and Simulation 7
for Robotic Systems with SysML, Simulink and Simscape Profiles
ibd [«StructuredSimulinkBlock» block] ClosedLoopSystem [Simulation Context]
energyport_R «energyBind»
signalport_x signalport_y
theta:real_T y:real_T x:real_T 1..* comp5:Sensor
dataport_x dataport_x
energyport_T
energyport_R
dataport_y dataport_y
dataport_A signalport_A
dataport_theta
dataport_S signalport_S
Figure 9. IBD showing the connection of components and subsystems.
In Fig.8, the ‘ClosedLoopSystem’ block is stereotyped «block,S tructuredS imulinkBlock» «block,spsC onv erter»
ClosedLoopSystem
with the SSB. The IBD represents the internal connection S_PS
1..*
of the system behaviour. Furthermore, the IBD shows ::SimulationContext:: sps Ports
every single component of the subsystem as part of the in:In
out:Out
properties of the ‘ClosedLoopSystem’ block. By Tags
definition, the IBD clearly illustrates the logical view of DiagramForSimulink:RhpString «block,pssC onv erter»
the entire system composition. However, it may be seen MatlabExePath:RhpString PS_S
Parameters:RhpString=step-size
as a static view unless the system behaviour can be 1..* Ports
SampleTime:RhpString=0.01 pss
executed. Toward this end, the simulation model should SolverName:RhpString=ode15s
in:In
out:Out
be designed for a complete execution of the model. SolverType:RhpString=variable-step
StartTime:RhpString=0.0
«block,solv er»
StopTime:RhpString=10.0 solver
4.3.2 SysML‐based Simulation Model 1
Solver
Ports
A simulation can be used to anticipate the behaviour of energyport:in_out
scope
the system to evaluate its performance. In SysML, a
construct for modelling simulation information is not «block,scope»
1..* Scope
explicitly provided; thus a method for establishing the
Ports
simulation information is needed to set up the
IN:In
configuration in the Simulink environment. According to
Cao et al., the simulation process in Simulink has four Figure 10. Simulation model in SysML
steps [14]: (1) System design modelling, in which every
element is defined as standard Simulink and/or Simscape
blocks and then the components are composed in the For this study, steps (1)‐(3) are explicitly modelled in
Simulink diagram by connecting them. (2) Auxiliary SysML in order to prepare the system‐level simulation
elements for simulation modelling, for instance, the model. It is worth noting that at least one local solver
Simulink scope component that can be used to monitor needs to be configured for each simulation in Simscape
the variable changes. (3) Simulation parameters and it is manually added by the designers. In Fig.10, we
configuration. This task must be done to set up the make up some new blocks such as the ‘S_PS’ and ‘PS_S’
simulation parameters or simply a global solver. (4) converters, the ‘Solver’ and the ‘Scope’ in order to
Lastly, the execution of the simulatable model. In step (2), illustrate the auxiliary elements of the simulation model.
we model the simulation‐related components such as the The block ‘ClosedLoopSystem’ shows the global
scope, local solver and Simulink‐Simscape signal simulation settings. Asides from the start/stop time, the
converters. solver name and its type also need to be defined. In
dataport_uL «energyBind»
ul:real_T
* :Kinematics
dataport_uL
dataport_y 1 :Body energyport_T
y:real_T
energyport_T signalport_x 1..* scope
in 1 :PS_S out
energyport_R energyport_R IN
signalport_y
dataport_A
theta:real_T
1 :Solver
«energyBind»
energyport
energyport_T energyport_R
1..* :Sensor dataport_S
signalport_A
dataport_theta
signalport_S
Figure 11. IBD for the simulation context
RPY:RhpString=.rpy java_lib:RhpString=rhapsody.jar SSC:RhpString=.mdl
Operations
Operations
In SysML, multiple Simulink and Simscape models are
Operations
export():void
export():void export():void
encapsulated and be referenced as a Simulink diagram in_out in_out in_out
Attributes
create_model mdl
simulation component package, and (2) its relative Configurations:RhpString
Operations
configuration files are then created to prepare for the create():void
:Simulink Env
model execution. In Rhapsody, the simulation component
setup is a level of organization which defines the Figure 12. SysML‐Simulink Model Integration
www.intechopen.com Mohd Azizi Abdul Rahman and Makoto Mizukawa: Model-Based Development and Simulation 9
for Robotic Systems with SysML, Simulink and Simscape Profiles
5.2 Implementation and Discussion output responses for a variety of controller coefficients
tuning exercises. As for comparison, the controller can
To validate our work, these models have to be executed in compensate the orientation error efficiently and settle to
order to see if the system performs as expected. To this end, an acceptable setpoint in a short time. It should be borne
the system design model (see Fig.8) and the simulation in mind that the results are shown just to prove the
model (see Fig.11) are evaluated. Once the model integration practical use of our system‐level design framework and
process is completed, the Simulink‐ and Simscape‐
methodology. For more simulation and analysis,
subsystem models can be generated by applying the
engineers have the option of manipulating other variables
‘generate and simulate’ function available in the
or parameters within the system.
Rhapsody context menu. The execution process is shown in
www.intechopen.com Mohd Azizi Abdul Rahman and Makoto Mizukawa: Model-Based Development and Simulation 11
for Robotic Systems with SysML, Simulink and Simscape Profiles
[13] K. Ohara, T. Takubo, Y. Mae and T. Arai, “SysML‐ [22] S. Bensalem, F. Ingrand and J. Sifakis, “Autonomous
Based Robot System Design for Manipulation Tasks,” Robot Software Design Challenge,” Proceedings of
Proceedings of Intl. Conf. on Advanced Mechatronics, Sixth IARP‐IEEE/RAS‐EURON Joint Workshop on
Osaka, 2010, pp. 522‐527. Technical Challenge for Dependable Robots in Human
[14] Y. Cao, Y. Liu and C.J.J. Paredis, “System‐Level Environments, Pasadena, 2008.
Model Integration of Design and Simulation for [23] R. Siegwart, I.R. Nourbakhsh and D. Scaramuzza,
Mechatronics Systems Based on SysML,” Introduction to Autonomous Mobile Robots : Intelligent
Mechatronics, vol. 21(6), pp. 1063‐1075, 2011. Robotics and Autonomous Agents series, 2nd Edition.
[15] T.A. Johnson, C.J.J. Paredis, R. Burkhart and J.M. Jobe, Massachusetts: The MIT Press, 2009.
“Modeling Continuous System Dynamics in SysML,” [24] J. Holt and S. Perry, SysML for Systems Engineering:
Proceedings of ASME Intl. Mechanical Engineering Professional Applications of Computing series, 1st
Congress and Exposition, Seattle, 2007, pp. 1‐9. Edition. London: The IET, 2007.
[16] T.A. Johnson, C.J.J. Paredis and R. Burkhart, [25] J.V. Amerongen and P. Breedveld, “Modeling of
“Integrating Models and Simulations of Continuous Physical Systems for The Design and Control of
Dynamics into SysML,” Proceedings of Modelica Conf., Mechatronic Systems : IFAC Professional Brief,”
Bielefeld, 2008. Annual Reviews in Control, vol. 27(1), pp. 87‐117, 2003.
[17] Modelica Association. (2011). Modelica Language. [26] IBM Corporation. (2010). Rational® Rhapsody® User
Available: http://www.modelica.org/ Guide. Available:
[18] Y. Venderperren and W. Dehaene, “From http://publib.boulder.ibm.com/infocenter/rsdp/v1r0m
UML/SysML to Matlab/Simulink : Current State and 0/topic/com.ibm.help.download.rhapsody.doc/pdf75/
Future Perspectives,” Proceedings of Design, Automation UserGuide.pdf/
and Test in Europe (DATE) Conf., Munich, 2006. [27] MathWorks Inc. (2011). Simscape™ User’s Guide.
[19] R. Kawahara, et al., “Verification of Embedded Available:
System’s Specification Using Collaborative http://www.mathworks.com/help/pdf_doc/physmod
Simulation of SysML and Simulink Models,” /simscape/simscape_ug.pdf/
Proceedings of Second Intl. Conf. On Model Based [28] X.H. Liu and Y.F. Cao, “Design of UAV Flight
Systems Engineering, Haifa, 2009, pp. 21‐28. Control System Virtual Prototype Using Rhapsody
[20] S. Turki and T. Soriano, “A SysML Extension for and Simulink,” Proceedings of Intl. Conf. On Computer
Bond Graphs Support,” Proceedings of Fifth Intl. Conf. Design and Applications, Qinhuangdao, 2010.
On Technology and Automation, Thessaloniki, 2005.
[21] K. Berkenköutter, S. Bisanz, U. Hannemann and J.
Paleska, “HybridUML Profile for UML2.0,” Int J
Softw Tools Technol Transfer, vol. 8(2), pp. 167‐176,
2006.