Rollback-Free Stable Brick Structures Generation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911661001015296 |
|---|---|
| author | Xu, Chenhui Bai, Ziyue Yu, Fuxun Huang, Heng Xiong, Jinjun |
| author_facet | Xu, Chenhui Bai, Ziyue Yu, Fuxun Huang, Heng Xiong, Jinjun |
| contents | While autoregressive models have advanced 3D generation, creating physically stable brick structures remains a challenge due to the strict requirements of gravity and interconnectivity. Existing approaches rely on external physical simulators during inference to perform rejection sampling and brick-by-brick rollbacks, which severely bottlenecks efficiency. To address this, we propose a reinforcement learning paradigm that shifts physical validity enforcement from test-time correction to training-time policy optimization. By utilizing assembly-level rewards, the model optimizes for collision avoidance, global connectivity, structural interlocking, and shape conformity. This paradigm allows the model to internalize physical priors, enabling the first rollback-free generation of stable brick structures. Experimental results demonstrate that our approach achieves state-of-the-art generation quality while accelerating inference speed by orders of magnitude. Our code and dataset are available at https://github.com/miniHuiHui/STABLE. Our models are available at https://huggingface.co/miniHui/STABLE. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_06947 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Rollback-Free Stable Brick Structures Generation Xu, Chenhui Bai, Ziyue Yu, Fuxun Huang, Heng Xiong, Jinjun Machine Learning While autoregressive models have advanced 3D generation, creating physically stable brick structures remains a challenge due to the strict requirements of gravity and interconnectivity. Existing approaches rely on external physical simulators during inference to perform rejection sampling and brick-by-brick rollbacks, which severely bottlenecks efficiency. To address this, we propose a reinforcement learning paradigm that shifts physical validity enforcement from test-time correction to training-time policy optimization. By utilizing assembly-level rewards, the model optimizes for collision avoidance, global connectivity, structural interlocking, and shape conformity. This paradigm allows the model to internalize physical priors, enabling the first rollback-free generation of stable brick structures. Experimental results demonstrate that our approach achieves state-of-the-art generation quality while accelerating inference speed by orders of magnitude. Our code and dataset are available at https://github.com/miniHuiHui/STABLE. Our models are available at https://huggingface.co/miniHui/STABLE. |
| title | Rollback-Free Stable Brick Structures Generation |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2605.06947 |