Capella & SysON: The MBSE force awakens

This talk presents two products from the Eclipse Foundation: Eclipse Capella and Sissone, along with their integration with Eclipse SDV. The speaker emphasizes the importance of model-based system engineering (MBSE) in the context of software-defined vehicles (SDV), highlighting the need for collaboration across various engineering disciplines and the necessity of establishing a shared architecture. Eclipse Capella is introduced as a mature solution that guides users through the design process, allowing for function-oriented system architecture and alternative exploration. Sissone is discussed as an open-source SysML V2 editor that offers flexibility and is geared towards democratizing system modeling. The talk also showcases practical applications and interoperability advantages of combining these tools, reinforcing their potential in modern engineering practices.

Coming over to the room after lunch he comes with its own challenges. I would like to welcome Etienne Julio. He's the founder and vice president of Obio and he's going to present a Capella and Sissone the MBSE force of Acans. Please round of applause for him. Thank you. >> Hello everybody. So we I'm here to present you so two products from the Eclipse Foundation so Eclipse Capella

and Sissone and how it is possible to couple them with Eclipse SDV. So we already work with the blueprint initiative inside SDV to use this seat adjuster sample to Capella and I will use it to explain you how it work to make it usable aside the other project and is the other initiative give us SDV. But first why model based system engineering is important in the context

of SDV. A key point is that you need to work with other disciplines. By other disciplines could be mechanicals, electrical engineers, 3D designers, hardware and others and so to it's very important to have exactly the same inputs when you do you do your design your detailed design. And this architecture which is common to everybody has to be made in a way that it is simple enough that

everybody can understand them and at your architecture level to not be lost into much detail. And by doing this there is the idea is to use model so you can define your component so component, subcomponent, subcomponent dependence between your component and function. Function are very key. And that when I say function is that C function. It's not about software. It's what the system is supposed to do.

So, you have a functional analysis with a pure functional view. So, somebody with specialized of sub part of the cars. And then the mapping the location of function to the component. And then you have a clean architectures to share with with others. This work is made inside models. Uh so, you can compu- do some computation la- ver- verification, simulation. And you can so drive this with a

code first approach because it's structured data. It's easier to connect um with other other tools. Um classical V-cycle to explain that we don't do detailed design at all. It's really the idea is really to stop at a level which is really cross-discipline, function-oriented, and architectures. And so, the idea is to couple with So, first, Capella. Capella is a very mature solutions uh designed to do um to

do this kind of principle but with a methodological guidance. Which mean that at each step uh the tool guide you to across activities, how you design a system as a black box, and then you are focused on the integration point with the external system you need to connect with. But as early as possible and it's impossible to design the internal. So, you keep the big picture with

your customer or the integration team of what is the integration point you have to do. Then you can design the internals. And then it force you to allocate functions in the right way. One function could be only on one component. It's impossible to to be on second one. So, it's provide plenty of good practice to do this um and to explore several alternative of architecture. SDV, you

know, is very flexible. So, you could do several alternative or solution for the same needs. With this, it helps you to create this kind of alternative as early as possible. Very do some verification and select the the good one for your for your detailed design layer. It's very mature again. It's used by more than 1,000 of enterprise all over the world. It's used for the open space

agency at Deutsche Bahn and in the automotive it's widely used as well. Um now, how it is on the example of the companion of the blueprint of SDV, uh how it it is translated. we start with the existing documentation which is available on GitHub um about the example. It was something like this. Uh so, it's great. It's explained by uh with some example of code and some

schema which are basically some drawers. There isn't any real semantic between beyond this schema. It just try to explain for the documentation. Only we use it to do a reverse engineering to first extract the operational analysis. So, who are the actors? Who are supposed to do any action on the seat adjuster? So, you can see here different actors, the vehicle user, the vehicle control, the remote user,

and different activities uh they are supposed to do and some scenarios of usage that they highlight in blue and red. Then you have the black box system. So, the internal of my black box is this one, the seat adjuster application. And I so I have other system I need to What is in in green is are the functions. So, I here look at four functions that the

four functions that the documentation of the blueprint explain that it's uh supposed to do in this way. And to perform this four functions, I need other function outside of my responsibility. And you have again the two scenarios which are the one from the operational point of view. So, from the end user point of view, how it is translate in this architectures. Then you design the internal. Um

in the the sample, there are three main component. And so, you find these three component and the previous allocation has been refined. So, there were only three. Now, there are three, six, seven, eight. Because it's more in detail, it speak about your solution. So, they kind of they are breakdown of the function to be more precise. And so, when you are developer, you know exactly the responsibility

of this component. You know that you have three three function to perform. And this three functions will have to interact with other components. It doesn't do at all your software design. Not at all. It's really ex- uh give you the input of what you are supposed to do, why, and with who as an integration point. Last, you have the physical architectures. It's just to say um from

an architecture point of view, I force that these fee- this system have to be deployed on a Raspberry Pi and only one. I can select perhaps several um several hardware if I want. Or I can just keep it free for the implementer to do as you want. But, in this example, it was only one. So, in green in yellow, sorry, you see that all the component has

been deployed on only one system. Second uh view, it's Season. So, Season is a project that we launched um at Thales. I think it was uh 1 year and a half to implement a pure SysML V2 uh open source editor. Um it's like this. We did it because sometimes people don't want to have a mythology guidance. They want to have all the language and organize their mythology

in the way they want, which is completely different from the one or the one from Capella. Uh, it also provides the foundation to support SysML V2. That's why the first point here is flexible by design. The idea is to create on the top of SysML V2 a kind of wrapper to use the, for example, the SDV like vocabulary. Uh, so you keep something which is oriented to,

I don't know, simulation software, while under the hood it's 100% a standard, the SysML V2 standard. Um, why that's why it is has been designed from scratch on a low-code uh, technology called Sirius Web, which is also an Eclipse Foundation project. Uh, it's web-based, uh, so very modern in terms of look and feel, and it's open source, of course. Um, our vision is to to make it

as simple as possible to democratize this uh, everywhere uh, at scale. So, easy easy to to use, but also thanks to open source, it is already the tool today in plenty of engineering school in the world and for most of the trainers about SysML V2 all over the world. um, and the idea also is to prepare to to scale at a very large level um, the number

of data and the performance of the systems. Um, compared to what exists before in in the SysML, perhaps some of you already heard about this, is really really a game-changer. It's not at all a an evolution V1 to V2 as usual, it's something completely new, but they keep the same trademark. For example, uh, you have now the equivalent uh, between um, a graphical uh, notation and the

textual notation. So, here at SDV, I know that code first is something very important. And so, it is possible now to write with something like a code your your models, and you have you have a transformation with the graphical part. And in the other way that works as well. And in season, for example, we are also able to to inject fragment of textual element directly in the

diagrams. It's fast and it explodes as a graphical element. You have also standardized API. So, your model is not just format that you can exchange between tools, but yeah, there are REST API like this one. So, other tools can interact in live with your model. Um there are some as you can see, there are some get endpoint. So, you can drive your SDV tool with the system

architecture models. But you can also come back to say, for example, you do a simulation and you consider that the initial constraint was not good or not satisfied. So, you use a put REST endpoint to send back to the architect the result of your simulation. And it work with any tool in the world which are conformed to SysML V2. So, that's very cool in terms of interoperability.

You have this kind of concept. So, now you have function-based systems. You have also variability directly in the language. You have verification case to do some verification directly inside the element, requirement of of course. And something interesting also is that there is an expression language to do some mathematics um directly at at this level. Uh there are also plenty of ready-to-use library for units, for example, a

kilograms, speed, temperatures. All these ISO standard of of unit are already built in the the So, we don't yet contribute on the GitHub of SDV this example because we finish it last week. it's a conference driven development. Um but we will uh very soon. Um so, the idea here is that it is exactly the equivalent of what you saw before with the Capella example, but this time

with um System LV2. So, you can still have the three different uh subcomponent of your system. And you have um the uh inside them, you have actions which um describe functions. So, that's not exactly the same terminology. and you can see that the the ergonomic is quite flexible. Here, for example, I decide that this port um have the different functions like a list under the hood. Here,

it is explode as as specific nodes graphically. And here, that's just two times the same just for fun. So, just to explain that you have a the freedom of how you show the the element in action. The result uh simplified version of what you just see before about the C C gesture application is something like this. So, you can exploit it directly if you want to have

a parser. In System, we we are working on a specific parser. Um a Java base on Taylor base parser for this which will be um autonomous piece of code that you can reuse to not reinvent the wheel. But you can also use a API directly if you don't want to pass by yourself. System and Capella, it's so you you have two solution. What we are working also

at um Obeo is to try to couple the two. So, System will will continue his road as a methodological neutral um tool. But we we we like to refactor to quite a new version of Capella or at least a declination of the mythology behind Capella based on SysML V2 and a web-based version of the tools. So, that's something which going on and one of the result we

also did with this with exactly the same content that you see before is this one. So, you can have um the same color map thing that the previous one with a component functions while under the hood that's really a SysML V2 element. So, if I show you in action oh, this one was my videos. So, I am here in my environment. If I want to create something

that just here and oh, the the network is quite slow, but okay. You can see that I cannot create plenty of elements. It's really guide me inside the component I can create a sub component or functions. Very small, very guide, easy to remember And um that everything I do here is exactly the same than this one. So, here if I come just at the logical element I

show I create before under the hood that's a part that I can show in a SysML vocabulary aside. So, that's very cool and I think that's something we could you could benefit Everything I showed to you is available in the open. So, here is the entry point. So, it's on the GitHub of Eclipse SDV Blueprints. You have the component application. Um the first link here, so after

you have the how you can develop this seat adjuster application with SDV SDV framework. And the first one here explain what is Capella operational analysis example. So So if you want to to have more detail it will be here. And the same CML version we have to connect with the SDV community to see how we can contribute it as an alternative for to to having the two

solution.