[Paper] Engineering Workflow, Reimagined with Agentic AI.

1. Physical fidelity: accurately segment and model the materials and textures of traffic elements

2. Realism: capturing details not only of vehicles and lane markings, but also of broader 3D environment

3. Automation: rapidly generating integrated and editable simulation models directly from camera data, covering entire pipeline from data processing, computer vision, to 3D reconstruction and simulation API integration.

In work that will be presented at the IEEE Intelligent Vehicles (IV) Symposium this week, we have taken an important step toward this ambitious vision. Please take a look at the video demo for more details, and visit our presentation.

See here the full paper: https://lnkd.in/eAnb8i2F

This exciting work is led by Andy Huynh and João Malheiro Silva, and our great autonomous driving collaborator, Prof. Holger Caesar (TU Delft).

The demo with Simcenter Prescan is possible with our Siemens colleagues supports: Rene Molenaar, Pieter Sibma, Serkan Ergun, Paul Baumgartner.

This work is also part of the EU project RobustifAI project on the Generative AI technology for Autonomous Driving use case that Siemens is working with partners.

Hope you enjoy it.