Autotech Blog

20200214 Vehicle Development

MEETING RECAP: Vehicle Development

The Autotech Council meeting on Vehicle Development took a close look at the newest tools for improving the design and manufacturing of vehicles, specifically Virtualization and Simulation.


Recent advances in the fields of AI, Machine Learning, 3D Virtual Design tools, and Simulation engines have made a huge impact at various stages of vehicle development. Right from the start, with visual design, then all the way through to 3D world simulations for developing and testing autonomous driving.


Our host, NVIDIA, explained how they are now mainly an AI company, and how their HQ, the very building we were in, was fully designed in a 3D virtual environment prior to construction. This enabled key stakeholders to virtually "enter the building", test out spaces, lighting, and sun beams at various seasons or times of day -- all before the first bulldozer had moved dirt.  This construction example perfectly illustrated how virtualization can improve safety and performance, enhance comfort, find bugs & errors, and greatly reduce costs for the overall project.  Applying this to the automotive arena, this entails curating the daily petabytes of data to extract the 10-14% of useful information to simulate the millions of miles travelled to find the most critical scenarios to test.  Nvidia’s Drive Constellation Virtual AV Simulator, enables corner case scenarios to be tested, that on-road testing might not encounter in sufficient quantities to ensure safety, along with supplementing and reinforcing the data collected from on-road testing.


From our speakers, we learned of the importance of data, of labelled data, and of using correction tools to bridge the gaps between the virtual worlds and the real one. We learned how models can combine virtual elements with real ones to test varous scenarios, and we were explained the importance of establishing a virtuous development loop whereby results and real-world data is fed back into the virtual development process, EVEN after the vehicle is sold to customers.


Following the break, our ten rapid fire presenters covered technology from the more established Stimulus product of Autotech Council Member Dassault, and NextChip’s Automotive Image Edge Processor, to new to the market camera clusters from Cluster Imaging, offered our audience a broad range of innovation to add to their vehicle development arsenal.


With thanks to our host Nvidia, our analyst Jim Hines from The Analyst Syndicate, Keynote Matt Cragun, Product Manager for Autonomous Vehicle Simulation at Nvidia, and panelists Vijay Nagasamy, Research Scientist at Ford and Heidi Hysell, CEO & Co-Founder at Alpha Drive for their expertise and insights.  In addition to Cluster Imaging, we thank our startup presenters from Samasource, Metamote, Viaduct, Aceinna, CalmCar Vision Systems, A2Mac and Alpha Drive for introducing their innovative products and services.  As always, members can view today’s presentations from our Member Library.

comments powered by Disqus