Rollback-Free Stable Brick Structures Generation

TL;DR

This paper introduces a reinforcement learning approach for generating physically stable brick structures without rollback, significantly improving efficiency and quality over previous methods that relied on external simulators.

Contribution

It presents the first rollback-free method for stable brick structure generation by training a policy with assembly-level rewards, internalizing physical priors.

Findings

Achieves state-of-the-art quality in brick structure generation.

Speeds up inference by orders of magnitude.

Enables stable structure generation without external physical simulation.

Abstract

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.