As vehicles become more software-defined and sensor-driven, the way we test and validate those systems is evolving fast. Traditional road testing isn’t enough—it’s too slow, too costly, and can’t cover every scenario. That’s where cloud-based Software-in-the-Loop (SiL) simulation comes in. At Magna, engineers like Christoph Wellershaus are pioneering ways to run thousands of ADAS test cases virtually—before a single vehicle is built. One recent benchmark? A simulation suite that would’ve taken 8 hours on a workstation ran in just 3 minutes in the cloud. Read the full #MagnaTechTalk on how SiL is reshaping the way we build safer, smarter vehicles. #TechTalk
Thanks for sharing
Shawna Hull
1w
Love this!! Magna always leads the competition! Great place to work! I miss it
Incredible step forward for ADAS development. Cloud-based SiL testing is truly reshaping what's possible in terms of speed and scale.
With vehicles becoming more software driven, traditional road testing just isn't enough. Cloud based Software in the Loop (SiL) simulation is changing the game by enabling faster, more cost-effective testing.
Anna Zwiauer
1w
Thanks for sharing the interesting insights! Well done, Christoph Wellershaus!
Great insights! V1rtualization of test suites is the way to go. Looking at the availability of (private) cloud solutions and the scalability of the machines using virtual machines, containers, and heavy infrastructure hosted in the data centers, the usage of SiL is advancing the speed of development and isolating the potential problems. It is one of the solutions to accelerate the development of ADAS functions, isolated from large HPCs, and enable separate updates of the functions within the SDV context. An additional benefit could be the enabling of the synthetic data for the tests and potential training of AI models (not restricted to image recognition, but going towards multimodal), making the test cases more relevant. The time to market is essential; therefore to consider a solution where the testing time is significantly reduced serves the purpose of quick updates for the SDV. In the competition with Chinese OEMs, the availability of quick updates could be a very important positive factor, especially for ADAS, where the traditional OEMs still have a competitive solution, together with the traditional Tier1s.
I've always assumed that this was part of the process already. Plugging the sensor inputs into the outputs of some kind of massive world multiplayer game type simulator, was to me the only logical way to train traction control, ADAS etc. Seems I was wrong?
Anton Batler
1w
One of the main challenges we faced in developing the simulation was modeling the sim maps. We made good progress in this area, learning how to quickly create maps for any road section across North America to support a wide range of simulation scenarios
Research and Development Manager at Honeywell
1wLove this