RSS

MEMORA: Embodied Action Memory from Egocentric Videos for Reasoning and Planning

MEMORA：基于自我中心视频的具身动作记忆用于推理与规划

Zihao Yu, Xiu Yuan, Chongjie Zhang

AI总结 研究针对长期机器人规划中对具身经验记忆的需求，提出MEMORA方法，通过形成-巩固-检索生命周期及四种类型存储实现具身动作记忆，经实验评估取得良好结果，能为机器人规划提供记忆上下文。

Comments 50 pages. v1: Oral presentation at the Robotics: Science and Systems 2026 Workshop on Foundation Models for Robot Planning (FM4RoboPlan). v2: Accepted to the Association for Computational Linguistics: EMNLP 2026

TFP: Temporally Conditioned Memory-Fusion Policies for Visuomotor Learning

TFP：用于视觉运动学习的时间条件记忆融合策略

Yushen Liang, Yue Peng, Baosheng Jin, Tianluo Zhang, Xinyu Zhang, Shuyi Zhou, Zhuoran Chen, Xinqi Liu, Shenji Wan

机构 * University of Electronic Science and Technology of China（电子科技大学） ； Beijing Institute of Technology（北京理工大学） 纠错

AI总结 研究针对视觉运动学习中阶段依赖操作问题，提出时间条件记忆融合策略（TFP），通过维护任务进展信念并注入动作解码器来改进VLA策略，在多个任务上提升成功率，证明紧凑且对事件敏感的记忆动态可优化VLA策略。

Comments Accepted to the SemRob 2026 Workshop at Robotics: Science and Systems (RSS 2026)

RLinf-USER: A Unified and Extensible System for Real-World Online Policy Learning in Embodied AI

RLinf-USER: 一个统一且可扩展的系统，用于具身人工智能中的现实世界在线策略学习

Hongzhi Zang, Shu'ang Yu, Hao Lin, Tianxing Zhou, Zefang Huang, Zhen Guo, Xin Xu, Jiakai Zhou, Yuze Sheng, Shizhe Zhang, Feng Gao, Wenhao Tang, Yufeng Yue, Quanlu Zhang, Xinlei Chen, Chao Yu, Yu Wang

AI总结 RLinf-USER是一个统一且可扩展的系统，用于现实世界中的在线策略学习，通过统一硬件抽象层和异步框架实现多机器人协调与长时训练。

Comments Accepted to RSS 2026