用预测模型在推理时优化生成式机器人策略，不需重训练即可提升表现。
Inference-Time Enhancement of Generative Robot Policies via Predictive World Modeling
- 通过动作条件世界模型预测扩散策略的动作后果，实现轻量在线规划。
- 在仿真与真实场景中，多任务表现优于标准行为克隆和现有推理优化方法。
- 适合需要快速部署、不改原策略的机器人应用，尤其擅长视觉与状态驱动任务。
我们提出生成式预测控制（GPC），一种无需重训练即可在推理阶段改进预训练行为克隆策略的方法。GPC在部署时，将冻结的扩散策略与基于专家示范和随机探索轨迹训练的动作条件世界模型结合。该世界模型预测扩散策略生成的动作提案的后果，支持轻量级在线规划，通过模型驱动的前瞻评估并优化这些提案。通过融合生成先验与预测远见，GPC实现了测试时自适应，同时保持原始策略不变。在多种机器人操作任务中，包括仿真与真实世界的基于状态和视觉的设置，GPC持续优于标准行为克隆，并与其它推理时适应基线方法相当。
原文摘要 · Abstract (English)
We present Generative Predictive Control (GPC), an inference-time method for improving pretrained behavior-cloning policies without retraining. GPC augments a frozen diffusion policy at deployment with an action-conditioned world model trained on expert demonstrations and random exploration rollouts. The world model predicts the consequences of action proposals generated by the diffusion policy and enables lightweight online planning that ranks and refines these proposals through model-based look-ahead. By combining a generative prior with predictive foresight, GPC enables test-time adaptation while keeping the original policy fixed. Across diverse robotic manipulation tasks, including state- and vision-based settings in both simulation and real-world experiments, GPC consistently outperforms standard behavior cloning and compares favorably with other inference-time adaptation baselines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。