Upcoming FMVSS 127 requirements introduce stringent performance criteria
for Pedestrian Automatic Emergency Braking (PAEB) systems under diverse
environmental conditions. Accurate validation of these systems demands sensor
models that replicate real-world imaging behavior at a physical level rather than relying
on idealized or simplified representations.
This session, jointly presented by OmniVision and dSPACE, will detail the integration of
Omnivision CMOS image sensor models into dSPACE’s sensor simulation platform. We
will discuss the modeling approach for key sensor characteristics and how these
influence perception algorithms. An example will be demonstrated based on dusk and
nighttime pedestrian detection, where realistic imager performance directly impacts
system reaction times and compliance with FMVSS 127.
Key technical topics include:
o Physics-based modeling of image sensor behavior for virtual testing.
o Scenario-based testing workflows for regulatory compliance.
o Quantitative analysis of a completed integration, showing perception metrics across
varied illumination and environmental conditions using both an ideal and realistic
imager noise model.
By leveraging high-fidelity sensor models in simulation, developers can evaluate system
robustness against FMVSS 127 scenarios earlier in the development cycle, reducing
reliance on costly physical testing while maintaining regulatory confidence.
