- At the Nvidia GTC conference held in Berlin on October 21, 2026, Dspace unveiled workflows that combine neural reconstruction with automotive simulation for the development and validation of Advanced Driver Assistance Systems (ADAS) and automated driving functions.
- The approach involves converting sensor data recorded on the road into virtual environments that can be used in simulation testing—ranging from software validation to hardware-in-the-loop testing.
This integration combines the NVIDIA HALOS Safe safety evaluation framework with NVIDIA Omniverse NuRec, a technology for the neural reconstruction of 3D environments based on data from cameras and LiDAR sensors. These reconstructed environments can then be used to create traffic scenarios, which can be modified or supplemented with new situations to expand test coverage.
A validation chain spanning from software to hardware
Dspace demonstrated this approach through two configurations. At the Nvidia booth, a software-in-the-loop (SIL) architecture allowed for the testing of software functions within a simulated environment. At the Dspace booth, a hardware-in-the-loop (HIL) setup incorporated physical hardware components into the simulation loop.
These two methods are applied at different stages of development. SIL testing is primarily used to evaluate the behavior of driving software in reproducible scenarios, while HIL testing verifies the operation of electronic control units (ECUs) and their software under simulated conditions. By using environments reconstructed from real-world data, these virtual tests are brought closer to the actual situations encountered on the road.
The Dspace Automotive Simulation Models (ASM) tool suite provides the necessary simulation models, including those for vehicle dynamics and traffic. Integration with Omniverse NuRec makes it possible to combine the reconstruction of environments—based on sensor data—with the simulation of vehicle behavior and traffic conditions.
Test Automation
SIL (Software-in-the-Loop) and HIL (Hardware-in-the-Loop) workflows, powered by Nvidia Omniverse, also support the automated execution of Euro NCAP tests. The goal is to contribute to the virtual validation of ADAS (Advanced Driver Assistance Systems) functions and to complement testing conducted under physical conditions. This same approach can be applied to higher-level automated driving systems, such as those designed for robotaxis.
By combining real-world data with synthetic data, it is possible to replicate traffic scenarios and then vary specific parameters or introduce events that were not necessarily present in the original recordings. This capability facilitates the repetition of tests under controlled conditions and allows for the evaluation of system behavior across a variety of situations.
For Dspace , this integration aims to expand the scope of virtual validation by leveraging its existing vehicle dynamics and traffic simulation models. It is part of a simulation-driven development approach designed to overcome the limitations of test campaigns that rely exclusively on physical vehicles and real-world driving scenarios.
