Generating Driving Scenes with Diffusion
This addresses the need for scalable and region-specific simulation data in autonomous driving, though it appears incremental as it builds on latent diffusion methods.
The paper tackles the problem of generating realistic traffic scenes for self-driving car perception simulation by developing a 'Scene Diffusion' system that combines diffusion models with object detection to create physically plausible arrangements of bounding boxes for agents, achieving adaptation to different US regions.
In this paper we describe a learned method of traffic scene generation designed to simulate the output of the perception system of a self-driving car. In our "Scene Diffusion" system, inspired by latent diffusion, we use a novel combination of diffusion and object detection to directly create realistic and physically plausible arrangements of discrete bounding boxes for agents. We show that our scene generation model is able to adapt to different regions in the US, producing scenarios that capture the intricacies of each region.