Page 56 - ATZ WORLDWIDE

DEVELOPMENT SIMULATION AND TESTING

specific project requirements with little Scene simulation Vehicle
effort, FIGURE 3. – Scene generator – Real prototype
The Command & Control component – Scene database Vehicle data
is used for controlling the processes from Vehicle data
a central point. It receives the relevant
vehicle data and passes them on to the Scene data Command &
scenario simulator. In this simulation, a Control Control
scenario database provides the basis for Scene data – Communication Control
framework
generating and varying the behaviour of
the virtual target objects needed for the
particular driving situation, with the Visualisation Scene data Active safety system
vehicle data received being used to cal- – Mobile display – Emergency braking
culate the position of the objects in rela- – Augmented reality app – Pre-crash
– Emergency steering
tion to the test vehicle. Finally, the object
position is sent to the central control
component in the form of object lists FIGURE 3 Interaction of the five components of the ViL approach (© IAV)
which pass this information on to the
assistance function under test and to the
visualisation facility.
For the test engineers to see the target
objects and their movement in relation
to the test vehicle and, on this basis,
evaluate the vehicle’s response, two
forms of visualisation are predominantly
used: an augmented reality view which
is shown on a mobile device with camera
(for instance a tablet), and a 3-D view,
FIGURE 4.
To generate the augmented reality
view, the virtually generated objects
and scenarios are superimposed on the
image of the real-world test environ-
ment taken by the camera. The 3-D
view, in turn, contains the scenario
generated in the simulation in three-di-
mensional form and provides different
views of the various phases of the test FIGURE 4 Two different views for visualising critical driving situations in the vehicle (© IAV)
scenario. The driver assistance function
can be loaded directly to the relevant
control unit in the real vehicle or, in
emulated form, to the Command & Con- ios to the real-world test environment. real-world vehicle which, with real-world
trol component. Finally, the fifth com- This operating principle makes it possi- vehicle tests, has hitherto either not been
ponent is the real prototype vehicle in ble to use the test method both in early possible or only possible to a limited
which the function being tested will be as well as in late phases of function extent. In particular, the ViL approach
used and in which all actuators are development. In the early phases, when can be used for testing scenarios that are
actually activated, making it possible in the target hardware is not yet available, too dangerous to perform with real-
the vehicle to experience the response driver assistance functions can be expe- world target objects. Examples of this
to the critical driving situation in a real- rienced in the real-world vehicle by include collision or narrow avoidance
istic manner. means of ViL. This permits early valida- scenarios at high relative speeds, critical
tion of the function in conjunction with driving situations, such as oncoming
a real-world vehicle. In this way, it is vehicles while passing and road users
APPLICATIONS
also possible to validate MiL, SiL and swinging in just ahead of the vehicle as
The ViL test method can be seamlessly HiL test cases on a realistic basis. well as multi-object scenarios, FIGURE 6.
integrated into the test processes already In the late phases of the development In view of the direction of the
existing for validating driver assistance process, the ViL can be used to increase approach, the focus is always on vali-
functions. It forms the bridge between the function’s level of maturity before dating driver assistance functions on
simulation-based test methods (MiL, SiL, validation commences in the field. Above the basis of the interaction between
HiL) and real-world test drives, FIGURE 5. all, the focus here is on making it possi- vehicle functions and actuators. As the
Here, elements from both testing types ble to experience and evaluate valida- sensors are simulated, they can be mod-
are combined by adding virtual scenar- tion-relevant driving situations in the elled in any way in terms of their char-
54