SEM-ROVER: Semantic Voxel-Guided Diffusion for Large-Scale Driving Scene Generation
This addresses the need for scalable 3D scene generation in autonomous driving and simulation, offering a novel method that improves over existing limited or incoherent approaches.
The paper tackles the problem of generating large-scale outdoor driving scenes with 3D consistency, proposing a framework that uses a semantic-conditioned diffusion model on a voxel grid to enable scalable, multiview-consistent scene generation without per-scene optimization, achieving photorealistic rendering with moderate computation cost.
Scalable generation of outdoor driving scenes requires 3D representations that remain consistent across multiple viewpoints and scale to large areas. Existing solutions either rely on image or video generative models distilled to 3D space, harming the geometric coherence and restricting the rendering to training views, or are limited to small-scale 3D scene or object-centric generation. In this work, we propose a 3D generative framework based on $Σ$-Voxfield grid, a discrete representation where each occupied voxel stores a fixed number of colorized surface samples. To generate this representation, we train a semantic-conditioned diffusion model that operates on local voxel neighborhoods and uses 3D positional encodings to capture spatial structure. We scale to large scenes via progressive spatial outpainting over overlapping regions. Finally, we render the generated $Σ$-Voxfield grid with a deferred rendering module to obtain photorealistic images, enabling large-scale multiview-consistent 3D scene generation without per-scene optimization. Extensive experiments show that our approach can generate diverse large-scale urban outdoor scenes, renderable into photorealistic images with various sensor configurations and camera trajectories while maintaining moderate computation cost compared to existing approaches.