High-fidelity reconstruction of driving scenes is crucial for autonomous driving. While recent feedforward 3D Gaussian Splatting (3DGS) methods enable fast reconstruction, their per-pixel Gaussian prediction paradigm often suffers from multi-view inconsistency and layering artifacts. Moreover, existing methods often model dynamic instances via dense flow prediction, which lacks explicit cross-view correspondence and instance-level consistency. In this paper, we propose PointForward, a feedforward driving reconstruction framework through point-aligned representations. Unlike pixel-aligned methods, we initialize sparse 3D queries in world space and aggregate multi-view image information via spatial-temporal fusion onto these queries, enforcing explicit cross-view consistency in a single feedforward pass. To handle scene dynamics, we introduce scene graphs that explicitly organize moving instances during reconstruction. By leveraging 3D bounding boxes, our method enables instance-level motion propagation and temporally consistent dynamic representations. Extensive experiments demonstrate that PointForward achieves state-of-the-art performance on large-scale driving benchmarks. The code will be available upon the publication of the paper.
@article{arxiv.2605.11594,
title = {PointForward: Feedforward Driving Reconstruction through Point-Aligned Representations},
author = {Cheng Chi and Xianqi Wang and Hongcheng Luo and Mingfei Tu and Gangwei Xu and Zehan Zhang and Bing Wang and Guang Chen and Hangjun Ye and Sida Peng and Xin Yang and Haiyang Sun},
journal= {arXiv preprint arXiv:2605.11594},
year = {2026}
}