Page 28 - ATZ WORLDWIDE

COVER STORY AUTOMATED DRIVING

FIGURE 1 Six aspects of ADAS and autonomous vehicle simulation – linking via Ansys simulation platform (© Ansys)

ing scenarios can be evaluated rapidly to SOFTWARE AND ALGORITHM ware, with model-based software develop-
check whether the vehicle sensors, con- MODELLING AND DEVELOPMENT ment techniques makes the software
trol algorithms and driving systems robust, less error-prone and safe. Moreo-
behave situationally as expected. Simulations play a key role in software ver, the model-based development of
Ansys’s simulation platform software development. Developing and testing sig- embedded software along with an ISO
connects all tools in the control loop, nal processing routines, object recognition 26262 qualified code generator, signifi-
FIGURE 2, to perform high-fidelity driving functions, control algorithms and cantly accelerates the development pro-
simulations of autonomous vehicles. Human-machine Interface (HMI) soft- cess. The Safety Critical Application
Development Environment (Scade) soft-
ware development package in the Ansys
simulation platform can incorporate third-
Drive scenario model Sensor models party machine learning software and neu-
Creates a model of the virtual world and animates “Observe” the surroundings in the virtual world of ral network for the AI functions of ADAS
motions of the test car and other objects in a the drive scenario model and output processed
test drive sensor signals and autonomous driving, FIGURE 3.
– 3-D road and landscape model – Sensing simulation
– 3-D models of stationary and moving objects – Signal processing PMD Cameras
– Object sensory attributes (e. g. radar reflectivity) FUNCTIONAL SAFETY ANALYSIS
– Object motion definition
– Motion simulation in time domain Radar Lidar V2X
ADAS and autonomous driving technolo-
System model GPS Ultrasonic sensors gies multiply the complexity of vehicle
Vehicle dynamics model Physics-based, high-fidelity systems. They not only increase the num-
Computes position, velocity and
orientation of test vehicle – Offline simulation ber of possible causes of failure but also
– Real-time simulation Signal proc. and sensor fusion the number of failure cascade paths.
– Vehicle mechanical model – Test automation
– Sub-models for vehicle attributes Identifies objects and driving conditions from ADAS and autonomous driving systems
sensor data
inherently have consequences for safety –
any failure can be catastrophic. Perform-
Vehicle component models ing functional safety analysis of such
Uses actuator inputs and computes Control algorithms and HMI
response of vehicle sub-systems such as brakes Makes main control decisions; displays critical complex systems is tedious and error
and steering information to the driver prone. Therefore, automated functional
– 3-D models of vehicle components – Software lifecycle, model-based development, safety analysis tools are indispensable to
– Detailed multi-physics simulation software testing, code generation
– ISO 26262, functional safety ensure safety of ADAS and autonomous
FIGURE 2 Simulation of the control loop for autonomous driving (© Ansys) driving systems. The Ansys simulation
26