
S P E C I A L : S O F T WA R E machine learning 5 9
“We really had to break new ground
to develop real remote solutions,”
says Werner Fuchs, Head of Software
Product Management and Operations
at AVL. In parallel, we were
facing strong pressure to adapt since
testing plays a major role within the
development process.
As the complexity of systems increases,
such as ADAS/AD functions,
the more testing effort is
required.
But not all conceivable
combinations of all potential parameters
– from weather conditions to
vehicle distance, from lane changes
to speed rules – can be tested on a
testbed or even on the road. This
makes simulation indispensable.
After all, OEMs who assume liability
for accidents from Level 3 onwards
should be capable of ensuring
the safety and security of their
ADAS/AD systems. Contingencies
should be ruled out through simulated
testing. While 30 percent of
“
LIKE THE MODERN WORKPLACE, TESTING
IS BECOMING INCREASINGLY HYBRID.”
Werner Fuchs
Head of Software Product Management and Operations at AVL
tests in this field are simulated today,
the ratio to real tests will decrease in
the future, Andrea Conti, Vice President
ADAS/AD at AVL believes,
“since up to 80 percent can be virtualized.”
As a leader in testing solutions and
technologies, we operate around
7,000 testbeds worldwide. Originally
designed for the world of
combustion engines, the expertise
has also been systematically built
up for electrification in the areas of
batteries, e-motors, inverters, and
fuel cells. It is no longer just temperature,
pressure, fuel consumption
or emissions that are measured,
but also, for example, the durability
of a battery and its charging cycles.
To achieve testing efficiency in
these labs, it is critical to employ
automated processes. This is where
Werner Fuchs’s division comes in.
His team is responsible for the leading
testbed automation software
AVL PUMA 2™, the only system
suitable for validating conventional
and electrified powertrains.
“State-of-the-art test environments
must be able to handle the increasing
complexity arising from the different
units under test. Therefore, virtual
development and test execution
on the testbed are necessary. Realworld
testing will not be dispensable,
but the number of physical
test scenarios will decrease,” states
Fuchs. “Like the modern workplace,
testing is becoming increasingly
hybrid. Virtual and real testing
must be better interlinked. Artificial
intelligence can provide assistance
with this.”
Crisis makes you inventive. This
was one of the few positive side effects
of the pandemic since travel
and movement possibilities were
generally restricted. As a result,
many systems had to be operated,
maintained, and updated remotely
– including vehicle testbeds.
N O . 1 2 0 2 3