Pushing that boundary involving advanced technologies from different fields, i.e. data processing, AI, perception algorithms, 3D modelling, physics-based sensor modelling,… which are also the main focuses of different ADAS teams in Siemens Digital Industries Software.
This demo shows an example how to reconstruct lidar point clouds data in simulation from the original nuScenes traffic dataset, implemented on Simcenter Prescan. Not just qualitative evaluations, the performance of the Real2Sim pipeline and quality of the simulation point cloud data (real-sim gap) are also analyzed and benchmarked quantitatively.
This work is part of the Delft University of Technology Master thesis student Joost Zaalberg, in a joint collaboration/supervision between Prof. Holger Caesar and our R&D ADAS team. Hope you enjoy the demo, and feel free to reach out to Holger or me if you have comment or interest.