ReinDriveGen: Reinforcement Post-Training for Out-of-Distribution Driving Scene Generation
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912995963043840 |
|---|---|
| author | Zhang, Hao Fan, Lue Bian, Weikang Wu, Zehuan Lu, Lewei Zhang, Zhaoxiang Li, Hongsheng |
| author_facet | Zhang, Hao Fan, Lue Bian, Weikang Wu, Zehuan Lu, Lewei Zhang, Zhaoxiang Li, Hongsheng |
| contents | We present ReinDriveGen, a framework that enables full controllability over dynamic driving scenes, allowing users to freely edit actor trajectories to simulate safety-critical corner cases such as front-vehicle collisions, drifting cars, vehicles spinning out of control, pedestrians jaywalking, and cyclists cutting across lanes. Our approach constructs a dynamic 3D point cloud scene from multi-frame LiDAR data, introduces a vehicle completion module to reconstruct full 360° geometry from partial observations, and renders the edited scene into 2D condition images that guide a video diffusion model to synthesize realistic driving videos. Since such edited scenarios inevitably fall outside the training distribution, we further propose an RL-based post-training strategy with a pairwise preference model and a pairwise reward mechanism, enabling robust quality improvement under out-of-distribution conditions without ground-truth supervision. Extensive experiments demonstrate that ReinDriveGen outperforms existing approaches on edited driving scenarios and achieves state-of-the-art results on novel ego viewpoint synthesis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_01129 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ReinDriveGen: Reinforcement Post-Training for Out-of-Distribution Driving Scene Generation Zhang, Hao Fan, Lue Bian, Weikang Wu, Zehuan Lu, Lewei Zhang, Zhaoxiang Li, Hongsheng Computer Vision and Pattern Recognition We present ReinDriveGen, a framework that enables full controllability over dynamic driving scenes, allowing users to freely edit actor trajectories to simulate safety-critical corner cases such as front-vehicle collisions, drifting cars, vehicles spinning out of control, pedestrians jaywalking, and cyclists cutting across lanes. Our approach constructs a dynamic 3D point cloud scene from multi-frame LiDAR data, introduces a vehicle completion module to reconstruct full 360° geometry from partial observations, and renders the edited scene into 2D condition images that guide a video diffusion model to synthesize realistic driving videos. Since such edited scenarios inevitably fall outside the training distribution, we further propose an RL-based post-training strategy with a pairwise preference model and a pairwise reward mechanism, enabling robust quality improvement under out-of-distribution conditions without ground-truth supervision. Extensive experiments demonstrate that ReinDriveGen outperforms existing approaches on edited driving scenarios and achieves state-of-the-art results on novel ego viewpoint synthesis. |
| title | ReinDriveGen: Reinforcement Post-Training for Out-of-Distribution Driving Scene Generation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.01129 |