发表机构
NUS-CLEAR; RCL-Lab（新加坡国立大学CLEAR实验室; RCL实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文总结ICRA 2026 REAL-I挑战赛经验，提出固定演示预算下机器人策略学习需整合数据、适应、评估与部署，并强调适应环境、保留能力及离线指标局限。
AI 中文摘要
机器人策略如何从固定的演示预算中更有效地学习？在ICRA 2026上举办的首届真实世界具身智能学习（REAL-I）挑战赛通过仿真、真实机器人评估以及在共享双臂人形平台上的现场决赛来考察这一问题。我们描述了挑战任务、数据和部署接口以及竞赛结果，然后比较了NUS-CLEAR、RCL-Lab和此http URL所贡献的方法。他们的系统结合了预训练的视觉-语言-动作模型和特定任务的模仿策略，并采用了不同的数据整理、分阶段适应、检查点选择和动作空间设计策略。团队报告强调了适应部署环境同时保留先前能力的重要性，以适当的时间尺度处理演示质量，并抑制非活动机器人组件中的错误。他们还揭示了离线动作预测指标在预测闭环成功方面的局限性。这些观察促使我们形成一种整合数据、适应、评估和部署的固定数据机器人学习观点。
英文摘要
How can robot policies learn more effectively from a fixed demonstration budget? The first Real-world Embodied AI Learning (REAL-I) Challenge at ICRA 2026 examined this question through simulation, real-robot evaluation, and an on-site final on a shared dual-arm humanoid platform. We describe the challenge tasks, data and deployment interfaces, and competition results, then compare the approaches contributed by NUS-CLEAR, RCL-Lab, and DeepTouch AI. Their systems combined pretrained vision-language-action models and task-specific imitation policies with different strategies for data curation, staged adaptation, checkpoint selection, and action-space design. The team reports highlight the importance of adapting to the deployment environment while retaining prior capabilities, treating demonstration quality at an appropriate temporal scale, and suppressing errors in inactive robot components. They also expose the limitations of offline action-prediction metrics for forecasting closed-loop success. These observations motivate a view of fixed-data robot learning that integrates data, adaptation, evaluation, and deployment.
Comments10 pages, 4 figures, 3 tables
