Honda Research Center Europe on Model-Based software development
Following the first interview with Antonello Ceravola, Principal Scientist at Honda Research Center Europe, Automotive IQ had a follow-up conversation with Mr. Ceravola on the updates for Honda regarding the biggest issues in model-based software development and the necessity of reducing complexity in modelling languages.
1. What is the biggest issue for your model-based practices nowadays?
In our experiece the biggert issues are:
- Having a flexible and always uptodated software infrastructure that supports the different needs we have over time
(e.g. integration of large-scale systems, development of small embedded sys. for different new boards, integration of different laguanges - C, C++, Python, ...)
The issue here is that our needs change over time and for us it is difficult to have them supported in short time by our software infrastructure
- Having people sufficiently trained in modeling. We have a lot of researchers and several engineers, but very few of them have modelling skils
- Debugging at system level it is still chellenging
2. The MBSE is actually meant to reduce the complexity of engineering, but it creates its own complexity in application. What is needed to make the gains from its introduction greater, than the difficulties it brings?
- The time to learn the usage of a MBSE infrastructures is a cost. Here is very important to make infrastructures as simple as possible and as intuitive as possible
- The available levels of abstraction are also a crucial factor. An MBSE infrastructure should provide several abstractions: threading, process, communication channels, synchronization, OS specific services, ...
- User interface also play an important role. A simple and intuitive GUI for developing MBSE is very important
- Monitoring. It should be easy for engineers to monitor the state of a running system. A well grounded log system and event-based notification system is mandatory
3. The questions of modelling languages to use in automotive engineering is still an urgent one. Do you think we are coming to some benchmarking solution, or are we still far away from it?
The problem with a modeling language is still open. Having a specialized modeling language for each domain (engine control, communication system, entertainment, ...) seems to be the best solution, however it require to learn too many languages. A single modeling language (domain-agnostic) can be used, however is less expressive than a specialized one.
We don't have yet a standard modeling language that have proof to be suitable for a large community.
So, we are still far.
4. As I understand there is no perfect software tool to be used to do MBSE for automotive applications. What advise can you give to the solution providers, who are looking forward to cover this gap?
For a solution provider it is crucial to work hand-in-hand with automotive engineers in order to capture the real use-cases that needs to be supported in a MBSE infrastructure.
In order to be successful, a tight cooperation is mandatory.
Until today we see OEM investing in defining their MBSE solution, with the problem that they can't invest in finalizing it to the level of a professiona solution; and we see solution providers realizing EBSE solutions that are too far from the actual requirements and use-case OEM's need.