论文HF Daily Papers · arXiv 2611.0912416 天前
A2-MEAS框架发布:代码驱动纠错,BEHAVIOR基准Q分超冠军2.2倍
本文由 AI 根据公开内容整理,可能有误,请以原文为准:Toward the Era of Agentic Embodied Intelligence(HF Daily Papers · arXiv 2611.09124)
核心内容
针对具身智能长程自主性与故障恢复难题,研究团队推出A2-MEAS框架及A2-WLA动作模型。该方案在BEHAVIOR 2026公开子集取得0.5755 Q-score,是2025年冠军成绩的2.21倍;在全量100项任务中Q-score达0.6411、成功率45.40%。RoboTwin 2.0平均成功率95.05%,12项真实家庭任务成功率57.92%,较pi0.5的26.67%提升逾一倍。
技术核心是将纠正语言、空间证据与执行历史编码为可学习修正码,通过MetaBrain提供监督;训练时由编码器注入策略权重,部署时冻结权重仅靠新代码适配行为。A2-WLA集成运动感知历史与潜在世界模型,在手冲咖啡等长程任务中将失败后完成率从14%提升至94%。训练基础设施采样率达427.1 samples/s,为pi0.5同硬件下的4.03倍,并将开源模型与脚本。
关键看点
- 基准性能:BEHAVIOR 2026全量任务Q-score 0.6411、成功率45.40%,子集成绩达2025冠军2.21倍
- 真机验证:5类家庭场景12项任务平均成功率57.92%,较pi0.5基线提升31.25个百分点
- 纠错机制:历史条件适配器使手冲咖啡关键步骤失败后完成率从14%跃升至94%
- 训练效率:采样吞吐427.1 samples/s,同等硬件下为pi0.5参考实现的4.03倍
工程与产业观察
该工作验证了“代码即接口”在具身智能中的工程价值:通过结构化修正码解耦策略学习与运行时适配,在不增加推理算力的前提下大幅提升长程任务鲁棒性。其开源承诺与4倍训练加速为社区提供了可复现的高效VLA训练范式,对解决真机数据稀缺与故障恢复具有直接参考意义。
英文摘要 (Abstract)
Foundation models increasingly guide robot actions, yet sustained autonomy requires efficient control and failure recovery. We introduce A2-MEAS, an embodied multi-agent framework built around code as a scaling interface. Learned correction codes combine corrective language, spatial evidence, execution history, and action proposals. MetaBrain provides training supervision and runtime guidance. A correction encoder and adapter absorb this knowledge into policy weights during training, while fresh codes adapt behavior with frozen weights at deployment. The action model, A2-WLA, adds motion-aware history and a latent world model to a vision-language-action policy. Surprisingly, this simple design with action pretraining and task specialization reaches 0.5755 Q-score on the 50-task BEHAVIOR 2026 public subset before online correction, 2.21x the 2025 winner's reported 0.2605 across different benchmark editions and training configurations. On all 100 BEHAVIOR 2026 public tasks, A2-MEAS achieves 0.6411 Q-score and 45.40% task success. On RoboTwin 2.0, A2-WLA achieves 93.98% average success, rising to 95.05% with the complete system. Across 12 real-robot tasks in five household settings, A2-WLA achieves 57.92% average success versus 26.67% for pi0.5. In long-horizon pour-over coffee preparation, a history-conditioned adapter raises post-failure completion at two filter-handling steps from 14% to 94% and from 8% to 82%, compared with A2-WLA retries. Our training infrastructure delivers 427.1 samples per second, 4.03x the throughput of a pi0.5 reference implementation on matched hardware. We will release the system, models, training scripts, and infrastructure to support efficient embodied learning.
Hugging Face Daily Papers 热度:45 upvotes 作者:Chao Liang, Ke Chen, Fengxiaoxiao Li, Shibo Huang, Yufeng Sun, Zhaokai Yin, Jianhuan Lai, Haozhe Feng, Chenghuan Jia, Zhipeng Zhang
外部来源
本页事实对应的公开页面,以原文为准。(点击将离开本站)