LidaRF: Delving into Lidar for Neural Radiance Field on Street Scenes
This work addresses the challenge of photorealistic simulation for autonomous driving applications, offering incremental improvements in NeRF quality for street scenes.
The paper tackles the problem of poor neural radiance field (NeRF) reconstruction quality in street scenes due to collinear camera motions and sparse sampling, by better utilizing LiDAR data to improve novel view synthesis, resulting in largely improved performance under real driving conditions.
Photorealistic simulation plays a crucial role in applications such as autonomous driving, where advances in neural radiance fields (NeRFs) may allow better scalability through the automatic creation of digital 3D assets. However, reconstruction quality suffers on street scenes due to largely collinear camera motions and sparser samplings at higher speeds. On the other hand, the application often demands rendering from camera views that deviate from the inputs to accurately simulate behaviors like lane changes. In this paper, we propose several insights that allow a better utilization of Lidar data to improve NeRF quality on street scenes. First, our framework learns a geometric scene representation from Lidar, which is fused with the implicit grid-based representation for radiance decoding, thereby supplying stronger geometric information offered by explicit point cloud. Second, we put forth a robust occlusion-aware depth supervision scheme, which allows utilizing densified Lidar points by accumulation. Third, we generate augmented training views from Lidar points for further improvement. Our insights translate to largely improved novel view synthesis under real driving scenes.