用强化学习让机器人双臂自动叠衣服，真实世界表现第二
Learning to Fold: prizewinning solution at LeHome Challenge 2026 (1st place online, 2nd offline)
- 用同一网络同时预测动作和成功概率，驱动策略优化
- 仿真中排名第一，真实场景中排名第二，性能领先
- 适合做具身智能、机器人操作和强化学习落地的参考
本文介绍我在LeHome Challenge 2026中的解决方案，该竞赛是ICRA 2026举办的双臂衣物折叠挑战。系统在62支队伍中，线上（仿真）环节排名第1，线下（真实世界）决赛排名第2。方法基于视觉-语言-动作（VLA）策略，通过强化学习循环进行改进。策略自身作为价值函数：同一个神经网络同时预测动作、任务成功度、进展及若干任务相关未来量，这些预测用于优势估计、实时失败检测和候选动作选择。工作主要整合现有强化学习思想，辅以工程与优化贡献，可整体使用或单独拆解：采用AWR + RECAP实现流匹配型VLA；通过HuggingFace Hub构建异步分布式训练/推理流水线；利用Thompson采样在推理时优化超参数；采用相机对齐工具、强数据增强和类似DAgger的人工干预数据收集实现有效的仿真到现实迁移。
原文摘要 · Abstract (English)
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.
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。