机器人 / 具身智能

SayPlan: Grounding Large Language Models using 3D Scene Graphs for Scalable Robot Task Planning

Krishan Rana, Jesse Haviland, Sourav Garg, Jad Abou-Chakra, Ian Reid, Niko Suenderhauf

专题命中 机器人学习 ：robotics(abstract)；分类 cs.RO、cs.AI；robot learning(comments)

Comments Accepted for oral presentation at the Conference on Robot Learning (CoRL), 2023. Project page can be found here: https://sayplan.github.io

A Framework for Learning from Demonstration with Minimal Human Effort

Marc Rigter, Bruno Lacerda, Nick Hawes

专题命中 机器人学习 ：robot learning(abstract)；分类 cs.RO、cs.LG；robotics(comments)

Comments Preprint version of IEEE Robotics and Automation Letters paper

Contrastive Learning from Demonstrations

André Correia, Luís A. Alexandre

专题命中 机器人学习 ：robotic(abstract,journal_ref)；分类 cs.AI、cs.CV

Journal ref IEEE Robotic Computing, Naples, Italy, December 5-7, 2022

Learning Sampling Distributions for Model Predictive Control

Jacob Sacks, Byron Boots

专题命中 机器人学习 ：robotics(abstract)；分类 cs.RO、cs.AI；robot learning(comments)

Comments Accepted at the Conference on Robot Learning (CoRL), 2022. Main paper is 9 pages with 4 figures. Appendix is 12 pages with 11 figures and 1 table

Augmentation for Learning From Demonstration with Environmental Constraints

Xing Li, Manuel Baum, Oliver Brock

专题命中 机器人学习 ：manipulation(abstract)；分类 cs.RO、cs.AI；robotics(comments)

Comments Submitted to 2023 IEEE International Conference on Robotics and Automation (ICRA)

Inferring Smooth Control: Monte Carlo Posterior Policy Iteration with Gaussian Processes

Joe Watson, Jan Peters

专题命中 机器人学习 ：robot learning(abstract,comments)；分类 cs.RO、cs.LG

Comments 43 pages, 37 figures. Conference on Robot Learning 2022

Learning from Human Directional Corrections

Wanxin Jin, Todd D. Murphey, Zehui Lu, Shaoshuai Mou

专题命中 机器人学习 ：robot learning(abstract)；分类 cs.RO、cs.LG；robotics(comments)

Comments This is a preprint. The published version can be accessed at IEEE Transactions on Robotics

Understanding Physical Effects for Effective Tool-use

Zeyu Zhang, Ziyuan Jiao, Weiqi Wang, Yixin Zhu, Song-Chun Zhu, Hangxin Liu

专题命中 机器人学习 ：robot learning(abstract)；分类 cs.RO、cs.AI；robotics(journal_ref)

Journal ref IEEE Robotics and Automation Letters, 2022

APPLD: Adaptive Planner Parameter Learning from Demonstration

Xuesu Xiao, Bo Liu, Garrett Warnell, Jonathan Fink, Peter Stone

专题命中 机器人学习 ：navigation(abstract)；分类 cs.RO、cs.LG；robotics(comments)

Comments Accepted by Robotics and Automation Letters (RAL) and International Conference on Intelligent Robots and Systems (IROS) 2020

Combining Deep Learning and Verification for Precise Object Instance Detection

Siddharth Ancha, Junyu Nan, David Held

专题命中 机器人学习 ：robotic(abstract)；分类 cs.CV、cs.LG；robot learning(journal_ref)

Comments 9 pages main paper, 2 pages references, 10 pages supplementary material

Journal ref Conference on Robot Learning (CoRL), 2019

Stochastic Optimal Control as Approximate Input Inference

Joe Watson, Hany Abdulsamad, Jan Peters

专题命中 机器人学习 ：robot learning(abstract,comments)；分类 cs.RO、cs.LG

Comments Conference on Robot Learning (CoRL 2019)

AC-Teach: A Bayesian Actor-Critic Method for Policy Learning with an Ensemble of Suboptimal Teachers

Andrey Kurenkov, Ajay Mandlekar, Roberto Martin-Martin, Silvio Savarese, Animesh Garg

专题命中 机器人学习 ：robotic(abstract)；分类 cs.AI、cs.LG；robot learning(comments)

Comments Conference on Robot Learning (CoRL) 2019

Disentangled Relational Representations for Explaining and Learning from Demonstration

Yordan Hristov, Daniel Angelov, Michael Burke, Alex Lascarides, Subramanian Ramamoorthy

专题命中 机器人学习 ：manipulation(abstract)；分类 cs.RO、cs.LG；robot learning(comments)

Comments 15 pages, 12 figures, accepted at the Conference on Robot Learning (CoRL) 2019, Osaka, Japan

Learning agile and dynamic motor skills for legged robots

Jemin Hwangbo, Joonho Lee, Alexey Dosovitskiy, Dario Bellicoso, Vassilios Tsounis, Vladlen Koltun, Marco Hutter

专题命中 机器人学习 ：robotics(abstract,journal_ref)；分类 cs.RO、cs.LG

Journal ref Science Robotics 4.26 (2019): eaau5872

Interpretable Latent Spaces for Learning from Demonstration

Yordan Hristov, Alex Lascarides, Subramanian Ramamoorthy

专题命中 机器人学习 ：robot learning(abstract,comments)；分类 cs.RO、cs.CV

Comments 12 pages, 6 figures, accepted at the Conference on Robot Learning (CoRL) 2018, Zurich, Switzerland

Control with Patterns: A D-learning Method

基于模式的控制：一种D学习方法

Quan Quan, Kai-Yuan Cai, Chenyu Wang

机构 * Beihang University（北京航空航天大学） 纠错

专题命中 机器人学习 ：robotics(abstract)；robot learning(comments,journal_ref)；分类 cs.RO

AI总结 本文提出了一种基于模式的控制方法，利用D学习在无需系统动力学知识的情况下解决非线性动力系统稳定性问题，并通过模拟和实际飞行实验验证其有效性。

Comments Accepted for publication at 8th Conference on Robot Learning (CoRL), Munich, Germany. 2024

Journal ref In 8th Annual Conference on Robot Learning, Munich,2024:270

An Alignment-Based Approach to Learning Motions from Demonstrations

Alex Cuellar, Christopher K Fourie, Julie A Shah

机构 * Massachusetts Institute of Technology（麻省理工学院） 纠错

专题命中 机器人学习 ：robot learning(abstract)；robotics(comments,journal_ref)；分类 cs.RO

Comments 8 pages, 8 figures, originally published in the IEEE Robotics and Automation Letters

Journal ref IEEE Robotics and Automation Letters, vol. 10, no. 11, pp. 11912-11919, Nov. 2025

Learning for Deformable Linear Object Insertion Leveraging Flexibility Estimation from Visual Cues

Mingen Li, Changhyun Choi

专题命中 机器人学习 ：manipulation(abstract)；robotics(comments,journal_ref)；分类 cs.RO

Comments 7 pages, 9 figures, 3 tables. 2024 IEEE International Conference on Robotics and Automation (ICRA)

Journal ref 2024 IEEE International Conference on Robotics and Automation (ICRA), Yokohama, Japan, 2024, pp. 5183-5189

DRL-VO: Learning to Navigate Through Crowded Dynamic Scenes Using Velocity Obstacles

Zhanteng Xie, Philip Dames

专题命中 机器人学习 ：robot learning(abstract)；robotics(comments,journal_ref)；分类 cs.RO

Comments Accepted by IEEE Transactions on Robotics (T-RO), 2023

Journal ref IEEE Transactions on Robotics, vol. 39, no. 4, pp. 2700-2719, 2023

A Robust Task-Level Control Architecture for Learned Dynamical Systems

用于学习动力系统的鲁棒任务级控制架构

Eshika Pathak, Ahmed Aboudonia, Sandeep Banik, Naira Hovakimyan

机构 * University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校） 纠错

专题命中 机器人学习 ：robotic(abstract)；分类 cs.RO、cs.LG

AI总结 针对基于动力系统的示教学习存在的任务执行不匹配问题，提出L1增强动力系统架构，结合标称稳定控制器、L1自适应控制器与窗口DTW目标选择器，在LASA和IROS手写数据集上验证了其跟踪性能。

Comments Accepted to the 8th Annual Learning for Dynamics & Control Conference (L4DC 2026)

Journal ref Proceedings of Machine Learning Research, PMLR 331:1243-1259, 2026

OrthoSkillVLA: Continual Skill Learning via Gradient-Informed Skill Subspace Adaptation

OrthoSkillVLA：通过梯度感知技能子空间自适应实现持续技能学习

Jiaqi Wang, Zhou Fang, Qiongfeng Shi, Yi Zhou

机构 * School of Computer Science and Engineering, Southeast University（东南大学计算机科学与工程学院） ； School of Electronic Science and Engineering, Southeast University（东南大学电子科学与工程学院） 纠错

专题命中 机器人学习 ：robot learning(abstract)；分类 cs.RO、cs.CV

AI总结 OrthoSkillVLA是一种参数高效的持续技能学习框架，通过对VLM和ActionHead施加独立子空间约束、引入特征感知MoE解码器，实现预训练VLA模型的技能学习，可缓解灾难性遗忘。

Comments Accepted by PRCV 2026

Deliberate Practice: Learning Robot Skills under a Budget

刻意练习：有限预算下的机器人技能学习

Shivam Vats, Sudarshan Harithas, Mete Tuluhan Akbulut, Arvind Raghunathan, George Konidaris

专题命中 机器人学习 ：manipulation(abstract)；分类 cs.RO、cs.AI

AI总结 研究有限预算下机器人序列任务技能学习问题，提出刻意练习算法，通过双线性规划计算预算最优分配，经仿真与真实实验验证可提升机器人长程规划能力。

Comments 16 pages including appendices

OLIVE: Online Low-Rank Incremental Learning for Efficient Adaptive Exoskeletons

OLIVE: 面向高效自适应外骨骼的在线低秩增量学习

Dong Liu, Yanxuan Yu, Ben Lengerich, Tong Geng, Ying Nian Wu

机构 * University of California, Los Angeles（加州大学洛杉矶分校） ； Columbia University（哥伦比亚大学） ； University of Wisconsin-Madison（威斯康星大学麦迪逊分校） ； Rice University（里奇大学） 纠错

专题命中 机器人学习 ：navigation(abstract)；分类 cs.RO、cs.LG

AI总结 提出OLIVE框架，通过低秩残差分解和奖励驱动策略梯度实现外骨骼控制的在线个性化自适应，在多种地形上提升步态平滑度、降低努力并增强稳定性。

Comments Accepted to UbiComp / ISWC 2026

Weights or Skills? A Survey of Robot-Learning Techniques: from Action-Predicting Weights to Robots that Write their Own Skills

权重还是技能？机器人学习技术综述：从动作预测权重到自主编写技能的机器人

Gaytri Jena, Kapil Wanaskar, Vinija Jain, Aman Chadha, Vasu Sharma, Amitava Das

专题命中 机器人学习 ：robot learning(abstract)；分类 cs.RO、cs.AI

AI总结 本综述围绕机器人学习的“权重vs技能”轴，梳理两类技术的分类、自我提升机制与开放问题，考察77个代表性系统，为机器人学习领域提供分析框架。

Comments 40 pages, 11 figures, 11 tables

AnyBody: Free-Form Whole-Body Humanoid Control from Arbitrary Keypoint Guidance

AnyBody: 基于任意关键点引导的自由形态全身类人控制

Shuning Li, Sikai Li, Jiachen Li, Mingyu Ding

专题命中 机器人学习 ：manipulation(abstract)；分类 cs.RO、cs.AI

AI总结 提出AnyBody，一种统一全身类人控制器，通过任意身体关键点子集实现运动跟踪与控制，利用特权教师蒸馏和掩码自注意力关键点编码器，支持灵活的下游任务学习。

Learning New Tasks via Reusable Skills: Skill-Compositional Experts for Embodied Continual Learning

通过可复用技能学习新任务：面向具身持续学习的技能组合专家

Shuaike Zhang, Shaokun Wang, Haoyu Tang, Jianlong Wu, Liqiang Nie

机构 * Harbin Institute of Technology, Shenzhen（哈尔滨工业大学（深圳）） ； Shandong University（山东大学） ； Shenzhen Loop Area Institute（深圳河套学院） 纠错

专题命中 机器人学习 ：manipulation(abstract)；分类 cs.RO、cs.CV

AI总结 提出技能组合专家（SCE）框架，通过组合技能基础（CSG）分解演示为可复用技能，并利用双执行-转换专家（DETE）实现新任务学习，有效缓解具身持续学习中的灾难性遗忘。

Comments 13 pages, 5 figures

Sample from What You See: Visuomotor Policy Learning via Diffusion Bridge with Observation-Embedded Stochastic Differential Equation

从所见中采样：基于观测嵌入随机微分方程的扩散桥视觉运动策略学习

Zhaoyang Liu, Mokai Pan, Zhongyi Wang, Kaizhen Zhu, Haotao Lu, Haipeng Zhang, Jingya Wang, Ye Shi

机构 * University of Science and Technology of China（中国科学技术大学） 纠错

专题命中 机器人学习 ：robotic(abstract)；分类 cs.AI、cs.LG

AI总结 提出BridgePolicy，通过扩散桥公式将观测直接集成到随机动力学中，利用语义对齐器处理异构观测，在模拟和真实任务中超越现有生成式策略。

Comments Accepted by ICML 2026

Double Preconditioning (DoPr): Optimization for Test-Time Performance, not Validation Loss

双重预处理 (DoPr)：针对测试时性能而非验证损失的优化

Thomas T. Zhang, Alok Shah, Yifei Zhang, Vincent Zhang, Nikolai Matni, Max Simchowitz

机构 * University of California, Berkeley（加州大学伯克利分校） ； Stanford University（斯坦福大学） ； University of Cambridge（剑桥大学） ； DeepMind（深度Mind） ； Google Research（谷歌研究） 纠错

专题命中 机器人学习 ：robot policy(abstract)；分类 cs.AI、cs.LG

AI总结 提出双重预处理优化范式，通过结合梯度级和激活级预处理，缓解自回归语言建模等场景中训练/验证损失与下游指标不匹配的测试时反馈问题，提升测试时性能而不一定改善验证损失。

PerceptTwin: Semantic Scene Reconstruction for Iterative LLM Planning and Verification

PerceptTwin：面向迭代LLM规划与验证的语义场景重建

Charlie Gauthier, Sacha Morin, Liam Paull

机构 * Department of Computer Science and Operations Research, Université de Montréal（蒙特利尔大学计算机科学与运筹学系） ； Mila - Quebec AI Institute（魁北克人工智能研究所） ； CIFAR AI Chair（CIFAR人工智能主席） 纠错

专题命中 机器人学习 ：robot policy(abstract)；分类 cs.RO、cs.AI

AI总结 提出PerceptTwin自动管道，从机器人感知的语义场景表示构建交互式仿真，结合LLM法官验证规划正确性与人类偏好，提升规划成功率约39%。

Comments Accepted at ICRA 2026 (Vienna); published on arxiv for archival purposes. See also https://percept-twin.github.io/

Virtual-to-Real: Learning to Control in Visual Semantic Segmentation

虚拟到现实：学习在视觉语义分割中的控制

Zhang-Wei Hong, Chen Yu-Ming, Shih-Yang Su, Tzu-Yun Shann, Yi-Hsiang Chang, Hsuan-Kung Yang, Brian Hsi-Lin Ho, Chih-Chieh Tu, Yueh-Chuan Chang, Tsu-Ching Hsiao, Hsin-Wei Hsiao, Sih-Pin Lai, Chun-Yi Lee

机构 * Elsa Lab（Elsa实验室） ； Department of Computer Science（计算机科学系） ； National Tsing Hua University（国立清华大学） 纠错

专题命中 机器人学习 ：robot learning(abstract)；分类 cs.RO、cs.CV

AI总结 本文提出了一种模块化架构，通过将感知模块和控制策略模块结合，利用语义图像分割作为元表示，解决虚拟到现实的迁移问题，并在障碍避让和目标跟随任务中展示了优越的性能。

Comments 7 pages, accepted by IJCAI-18