We used to see automotive manufacturers compete to improve mechanics and quality requirements. However, in the age of digital transformation, electronics turned out to be the most significant differentiating factor between them.
Shortening development cycles seems to be the biggest challenge that software development faces today. Emerging technologies and growing consumer demands require advanced software solutions. The same goes for car manufacturing; the fierce competition pushes them to implement technologies fast and release advanced vehicles to the market.
To deal with that fast-paced environment, manufacturers and suppliers considered a fundamental change in software development. They have switched the hand-coded into a model-based development (MBD).
This approach supports a function based and not code-based engineering of automotive systems. That is why it enables a shift of focus on the development process in the early phases. Furthermore, a great advantage is that it gives the ability to trace errors early on and save time and money in the overall process.
The model-based design is now becoming the foundation of the development process. That is because it supports collaborative engineering and additional testing. As we mentioned, it cuts down development time and cost while improving the adaptability and reliability of the end product.
Despite all the benefits, there are still some challenges which model-based design still faces. If we overcome them it will ensure a substantial time and cost reduction. Here you can find some of the biggest challenges the model-based design faces in 2019.
Lack of Widely Accepted Approach
Automotive electronics and software require a chain of complicated steps in development, which involve different parts of the organization. There are several issues and communication challenges that appear because of the complexity of the process itself. First, OEMs set the requirements and control software. Suppliers develop this software in an isolated manner. Afterward, it is integrated by the OEMs.
That is why a model-based virtual integration is needed — moreover, a lack of quality in formalized modeling causes dissimilarity between models and limitations in tool support.
To deal with this challenge, a formal model-based integration framework was proposed. It is based on the methodology of composition and timing specifications. Legal contracts could be used to standardize a trustworthy timing composition.
The model-based design fundamentally changes the way automotive software and functions are developed. With this new approach, a model is designed using languages such as UML or SysML. This model is put into a model-based environment, and the actual code is automatically generated.
This fundamental change faces resistance from being used by most people. That’s not technical, but a management challenge that requires specific changes. With such differences in management, it has to be demonstrated that the MBD is a more efficient and high-quality code that doesn’t necessarily require hand coding.
Automotive systems are becoming more and more effective. Modern vehicles have an extremely high number of lines of code – 100 million. The traditional approach of software development is now reaching its limits in every aspect.
For this reason, multiple manufacturers have started applying MBD as a solution to the growing complexity. A great example is Honda.
The complexity comes with every phase of the development process, which makes the model-based approach necessary. Managing this through a different perspective is more efficient than reducing it to a single coding phase. Furthermore, the MBD frameworks make the integration of larger systems easier.
Testing at the model and code level is a crucial step in the development process for any software product. The model-based testing brings us closer to the automation of testing activities.
Now, testing is a time-consuming and challenging task. It requires a test engineer’s guidance and efficient techniques. The automating tools can’t replace domain testing experts in any way, as they’re the ones who can develop the most useful tests or fragments, unlike any tool.
There’s a gap between the expertise and the way of communication of function developers and testing experts. As we noted above, there’s no standardized approach to documenting models; all parties can equally well understand that. It leads to more time spent on trying to understand the models before integrating them into the test environment.
To overcome this challenge, an integrated testing approach was developed and applied to a model of the component-based composition of systems based upon the AUTOSAR standard. Its main idea was to find abstract methods for detecting mutants by generating test data to uncover them.
The model-based design is now widely adopted by many leading companies in the automotive industry. The traditional, document-oriented approach has some limitations when it comes to more complex systems.
Despite all of its advantages, there are still many challenges, which the model-based design faced in 2019. They need to be overcome so that a seamless model-based development will be applied. We’ve listed some of the biggest challenges above.