A web-based platform that allows users to test and validate ADAS/AV automotive sensors in a digital twin virtual environment that replicates real-world conditions is being developed by Ansys in partnership with Cognata and Microsoft.
The Automated Driving Perception Hub (ADPH) platform, which is run on Microsoft Azure and managed by Cognata, will benefit from the high-fidelity radar and electromagnetic (EM) wave propagation simulation capabilities provided by the Ansys AVxcelerate Sensors simulation software.
The digital twin platform has a library of manufacturer-certified virtual sensor models and is powered by AMD EPYC central processing units (CPUs) and Radeon PRO graphics processing units (GPUs) for workloads including machine learning inference and visualization.
In order to expedite safety, certification, and virtual homologation, it has also created an integrated toolchain in partnership with Kontrol, Microsoft, and TÜV SÜD.
94% of the top 100 automotive suppliers utilize Ansys’ simulation tools, and the company is expanding its agreements with Sony Semiconductor and Marelli Electronic Systems. For the past year, Synopsys has been pursuing a $35 billion deal to buy Ansys. Next week at CES 2025, the tools for creating autonomous driving software and ADAS systems will be on display.
The business has created ConceptEV, a design and simulation platform that speeds up the development of EV powertrain systems and helps cross-functional teams to satisfy market and customer demands. Cognata’s Automated Driving Perception Hub, which uses a high-fidelity simulation platform with virtual twin technology, provides access to the AVxcelerate Sensors for the Cognata agreement.
OEMs and sensor manufacturers can test and validate certified sensors using the ADPH against a variety of industry standards, such as those established by the New Car Assessment Program (NCAP) and the National Highway Traffic Safety Administration (NHTSA). Cognata sensor versions for thermal cameras, LiDAR, and RGB cameras with different lens aberrations are currently available on the platform.
In order to improve signal strength and accuracy, physics-based radar models replicate the propagation of electromagnetic waves, taking into consideration material properties at high frequencies.
The algorithms that handle radar signal interference, such as minute frequency variations brought on by moving objects (doppler effect), can be tested and improved using the raw data from the radar simulation. AVxcelerate Sensors creates a virtual twin of the sensor when it is connected to a virtual model from a radar supplier. This allows OEMs to assess the sensor’s performance with increased predicted accuracy.
The AMD Radeon PRO V710 GPUs power Cognata’s Generative AI transfer technology, which improves the RGB camera simulation platform by faithfully capturing and simulating the behavior of sensors in the actual environment.
For scenario-based perception testing of the Sony high dynamic range (HDR) Image Sensor Model, real-time multispectral camera simulation is offered by the Ansys AVxcelerate Sensors autonomous vehicle (AV) sensor simulation program.
By taking into consideration sensor behavior in a variety of driving situations, including as low light, dusk, rain, snow, and fog, OEMs are able to evaluate advanced driver assistance systems (ADAS) and AV features.
To replicate how light moves through the surroundings, the camera lens, and finally reaches the imager, the AVxcelerate Sensors platform creates a virtual world with a variety of lighting, weather, and material conditions.
When combined with Sony’s sensor model, this simulation can accurately anticipate the pixel properties, signal processing capabilities, and system operations of Sony’s HDR imager.
The accuracy, reliability, and safety of Sony’s HDR imager-based perception systems can be enhanced in ADAS and AV applications by using this simulation model to do thorough, scenario-based testing with either pre-defined inputs or real-time feedback.
Combined simulations can introduce visuals into sophisticated Software-on-Chip perception systems to reduce on-road testing.
In order to verify its performance, an electronic control unit—which is used to regulate things like engine management and transmission—is simultaneously included into this simulated environment. This method guarantees the accuracy and dependability of the entire simulation process, from sensors to processor chips.
While the Discovery 3D simulation software uses Nvidia’s Omniverse Blueprint digital twin platform to visualize and speed up large-scale computational fluid dynamics workflows, the Ansys SimAI cloud-enabled generative AI platform predicts physics behavior and can be trained with existing simulation results for applications including fluid dynamics, thermal and electromagnetic performance, and structural deformation.
Key Comments
“We are excited to integrate Ansys’ radar simulation technology into the ADPH platform, bringing OEMs and tier-one suppliers an unmatched level of accuracy in sensor validation,” said Danny Atsmon, founder and CEO at Cognata. “Ansys’ ability to simulate complex EM wave interactions enhances our platform’s ability to deliver precise, real-world insights for radar-based ADAS and AV systems. This collaboration significantly advances the industry’s ability to test and refine sensor performance under diverse conditions.”
“Ansys’ AVxcelerate Sensors platform includes real-time radar capabilities for accurate modeling of radar interactions in complex environments,” said Shane Emswiler, vice president of products at Ansys. “As the industry works toward fully autonomous driving, safety validation is paramount, and the joint effort between Ansys and Cognata streamlines this typically long and complicated process.”
“Achieving full autonomy involves OEMs working with leading technology providers like Ansys to enhance the accuracy of the integrated tools used to validate AV systems,” said Tomoki Seita, general manager,automotive business division, Sony Semiconductor Solutions Corporation. “Through this collaboration, customers can confidently verify their systems using highly reproducible, predictively accurate simulations. This is especially useful for OEMs and Tier 1 suppliers that run actual camera simulations to verify recognition algorithms and vehicle control software.”
“Using Ansys tools, we have realized a 25% reduction in product development cycle, 15% to 20% savings on engineering development costs, and the 15% to 20% improvement in product performance — to the satisfaction of our customers,” said Luciano Saracino, head of the Mechanics and Optics Centre of Expertise, Marelli Electronic Systems.