PGDG: 为从单个示范中学习鲁棒双臂策略而设计的物理基础数据生成
PGDG: Physically Grounded Data Generation for Robust Bimanual Policy Learning from a Single Demonstration
AI总结：
本文提出PGDG，一种基于物理的数据生成框架，通过零样本校准扩展单个示范为包含物理上合理、成功和多样恢复行为的紧凑数据集，从而提升双臂操作中接触丰富的行为克隆性能。
摘要：
接触丰富的双臂操作中的行为克隆仍然具有挑战性，因为多样化的示范收集成本高，且即使小的扰动也可能将系统推入无恢复监督的流形外状态。我们提出PGDG，一种具有零样本校准的数据生成框架，能够在不额外人工标注的情况下，将单个示范扩展为一个包含物理上合理、成功且多样化的恢复行为的紧凑数据集。PGDG在物理基础采样器和数据集校准器之间迭代，其中校准器选择具有信息量、非冗余性和可恢复性的行为来更新采样分布，朝向未覆盖的恢复模式；而采样器则从更新后的分布中绘制出物理上合理的滚动候选，并保留成功的轨迹。为进一步提高数据质量，PGDG应用短时间域采样基于控制来重新标记所选的高风险状态并应用纠正动作。在四个双臂操作任务中，PGDG在仿真和零样本现实世界迁移中均优于仅空间增强的方法。在RotateBox-Pitch任务中，仿真中的成功率从38%提升到93%，现实世界中的成功率从35%提升到82%。PGDG还能够有效促进如GR00T等基础模型的微调，使成功率从46%提升到77%。更多结果可在我们的网站上查看：https://cunxid.github.io/PGDG/。
英文摘要：
Behavior cloning for contact-rich bimanual manipulation remains challenging because diverse demonstrations are expensive to collect, and even small disturbances can push the system into off-manifold states where no recovery supervision is available. We propose PGDG, a data generation framework with zero-shot curation that expands a single demonstration into a compact dataset of physically plausible, successful, and diverse recovery behaviors without additional human labeling. PGDG iterates between a physics-grounded sampler and a dataset curator, where the curator selects informative, non-redundant, and recoverable behaviors to update the sampling distribution toward under-covered recovery modes, and the sampler draws physically plausible rollout candidates from this updated distribution and retains successful trajectories. To further improve data quality, PGDG applies short-horizon sampling-based control to relabel selected risky states with corrective actions. Across four bimanual manipulation tasks, PGDG consistently outperforms spatial-only augmentation in both simulation and zero-shot real-world transfer. On RotateBox-Pitch, success improves from 38% to 93% in simulation and from 35% to 82% in the real world. PGDG also enables effective foundation models fine-tuning such as GR00T, increasing success from 46% to 77%. Additional results are available in our website: https://cunxid.github.io/PGDG/.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
