Open House 2020 keynote: Virtual integration at Continental Engineering Services

The validation of assistance
and automated driving functions is one of the top issues among OEMs
in today’s series development. In my view,
three aspects are important: A robust process
for handling test cases, with requirements and result data, the option of very early testing
within a full virtual vehicle and a powerful integration environment that makes all this possible. At Continental Engineering Services, we heavily rely on virtual methods
for these purposes. We know that many new players
have entered the game from other technological fields. In some cases,
they bring entirely new use cases, new processes and tools. This is a massive change
we’re observing. This also impacts
the development methods – speaking of agile methodology. The challenge is
the massive number of requirements because systems are becoming
more complex and partly distributed
across different ECUs. Finding a test strategy,
and tools that support this, that is the key challenge
in my opinion. Simulation is not an end in itself, simulation is a means to an end. Illustrated along the V model, two primary goals become obvious: For one, that is
increasing the test depth. You can run a breadth of tests
much more flexibly and take a close look
at various issues. That is definitely beneficial
to the quality. But the aspect of cost savings
is important as well. Each kilometer not driven
on the road, each kilometer not driven because the software
wouldn’t work anyway, is a kilometer saved. Illustrated along the V model, simulation has two applications
in our view: On the left,
the requirements stage or specifications stage, we can use simulation to make
the presumably best decision early on, and track whether requirements
are met and effective. And with early virtual integration,
we can find whether we will develop a valid product
the customer will accept. On the right side of the V, we can carry out earlier validations
based fully on simulation. And for later validations, we can perform only those parts
in the vehicle that would be too elaborate
in the simulation or for which the discussions
about the validity of results would be too tiring to have. In my department, CarMaker is the standard
integration environment for the validation of ADAS
and brake control systems. We use CarMaker
starting at model-in-the-loop to software-in-the-loop
and hardware-in-the-loop and up to vehicle-in-the-loop
to build test systems for our internal
and external customers. But we also use CarMaker to perform simulation studies
for concept analyses, or for sensitivity analyses. It also happens
that customers need support in creating RFQs
for their component suppliers. There we also use CarMaker, to find through sensitivity analyses which KPIs are critical
that also become part of the RFQs and the component specification. There simply is no alternative. The systems have by now
become so complex, so fraught with requirements, that we can’t do without simulation
any more. The key is rather
to find a process or a tool that can be accepted by everyone. So that a strong test management
can decide early on which parts will be tested
in the simulation and which to test in the vehicle. A good solution will then be welcomed
with open arms.

Leave a Reply

Your email address will not be published. Required fields are marked *