学习折叠：LeHome Challenge 2026 获奖解决方案（在线第一名，离线第二名）
Learning to Fold: prizewinning solution at LeHome Challenge 2026 (1st place online, 2nd offline)
AI总结：
提出一种结合强化学习循环的视觉-语言-动作策略，通过自值函数实现优势估计、故障检测和候选选择，在LeHome Challenge 2026中获得在线第一名和离线第二名。
摘要：
我描述了我对LeHome Challenge 2026（ICRA 2026双臂衣物折叠竞赛）的解决方案。该系统在在线（仿真）轮次中位列62支队伍的第一名，在真实世界决赛中获得第二名。它通过强化学习循环改进了视觉-语言-动作（VLA）策略。该策略本身就是其价值函数：预测动作的同一网络也预测成功、进度和一些任务相关的未来量，这些预测驱动优势估计、实时故障检测和候选选择。该工作主要重新组合了现有的强化学习思想，并贡献了工程和优化方面的成果，这些成果可以作为一个整体配方或单独使用：AWR + RECAP结合用于流匹配VLA；通过HuggingFace Hub的异步分布式训练/部署流水线；通过汤普森采样进行推理时超参数优化；以及带有相机对齐工具、大量数据增强和类似DAgger的人机交互数据收集的仿真到现实配方。
英文摘要：
I describe my solution to the LeHome Challenge 2026, an ICRA 2026 competition on bimanual garment folding. The system placed 1st of 62 teams in the online (simulation) round and 2nd in the real-world final. It improves a vision-language-action (VLA) policy with a reinforcement-learning loop. The policy is its own value function: the same network that predicts actions also predicts success, progress, and a few task-relevant future quantities, and those predictions drive advantage estimation, live failure detection, and candidate selection. The work mostly recombines existing RL ideas with engineering and optimization contributions that can be used together as one recipe or individually: AWR + RECAP combined for flow-matching VLA; an asynchronous distributed training / rollout pipeline through HuggingFace Hub; inference-time hyperparameters optimization via Thompson sampling; a sim-to-real recipe with camera-alignment tooling, heavy augmentation and DAgger-like HIL data collection.
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。
