VIRTUAL TESTING FOR NCAP 2026 CRASH AVOIDANCE - ASCS Brochure #2025 - Magazine - Page 42
VIRTUAL TESTING FOR NCAP 2026 CRASH AVOIDANCE
LEVERAGING VIRTUAL TESTING
FOR NCAP 2026 "CRASH
AVOIDANCE" AND BEYOND
Euro NCAP 2026 introduces tougher requirements for crash
avoidance systems, creating major challenges for automakers and suppliers. Traditional testing on proving grounds can
no longer meet these demands – it is expensive, time-consuming, and doesn’t fully capture rare or complex driving
scenarios. This is why virtual testing has become essential for
validating advanced driver assistance systems (ADAS) and
automated driving (AD) features and is now included in the
NCAP 2026 as part of the validation toolchain.
The new NCAP 2026 structure, effective January 1, 2026,
demands far more testing across a wider range of situations
and creates exponential testing increases for suppliers and
OEMs. Real test drives struggle to provide the consistency,
coverage, and scalability needed. Extended range testing
doubles the costs of proving ground tests, which struggle
with edge cases, variable environmental conditions, and
reproducibility issues, especially for perception-based
robustness
layers.
Scenario-based
virtual
testing
(Software-in-the-Loop, SiL) with tools like Ansys AVxcelerate
handles millions of test variations per week, enabling statistical exploration and edge-case discovery. These simulations
support safety analysis of a particular ADAS function, discover edge cases, and validate performance efficiently.
Perception systems, especially camera-based ones, are sensitive to environmental changes and require realistic, high-fidelity testing.
To go beyond just ADAS function validation and ensure
overall system robustness especially in challenging environmental conditions it is also necessary to challenge the
FUTURE-ORIENTED STRATEGY
perception system. This is where the concept of Perception-in-the-Loop arises. Two parts are essential to enable it:
the camera-based perception itself (the device-under-test)
and the accurate, physics-based environment and sensor
simulation. When camera software isn’t available for simulation, real camera hardware can be tested using synthetic
AUTHENTICITY
image injection. Hardware-in-the-Loop (HiL) testing
processes 100,000+ variations weekly by engaging real
camera units in virtual environments. The traditional
Over-the-Air (OTA) HiL setup is a widely used in the industry
but has several limitations that prevent its usage to test
perception under challenging conditions (low standing sun
or glare, nighttime, etc.). Direct Synthetic Image Injection
approach developed by Ansys, NI and Valeo can effectively
solve these limitations and reduce the reliance on the costly
real-world test drives. Physically accurate multi-spectral
simulation in Ansys AVxcelerate creates environments with
validated sky models,CORE
spectral
light sources, and material
COMPETENCIES
properties, replicates how camera sensor captures light
(optical path) and generates images (photosensitive imager,
electronic circuitry and signal processing) – all crucial for
testing not only the ADAS function, but the perception
system as well.
Cost reduction comes from ensuring first-time proving ground success, focusing physical testing on decision
and control validation. Enhanced coverage enables testing of dangerous or impractical scenarios, including
edge cases and nighttime conditions difficult or impossible to recreate safely. Virtual testing addresses evolving
NHTSA and NCAP regulations and paves the way for digital homologation of Level 2+ to Level 3 automated
driving systems.
Successful implementation of virtual testing requires three validation steps: in-house simulation tool qualification by OEMs, physical verification testing through random selection, and approval of virtual test procedures.
Key challenges remain in ensuring simulation accuracy reflects real-world performance, protecting sensor
supplier IP during perception testing, and adapting to evolving regulatory requirements. Overall, NCAP 2026
marks the beginning of the journey toward virtual driving and digital homologation. The ultimate goal is replacing physical testing kilometers with virtual alternatives while maintaining and increasing safety quality. Virtual
testing will shift from validation-only to full development integration, including machine learning training and
algorithm development throughout the entire product lifecycle.
42
