cs.RO（2026-09-30）

📊 共 66 篇论文 | 🔗 8 篇有代码

🎯 兴趣领域导航

支柱一：机器人控制 (Robot Control) (43 🔗5) 支柱二：RL算法与架构 (RL & Architecture) (12 🔗2) 支柱三：空间感知与语义 (Perception & Semantics) (5 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (3) 支柱六：视频提取与匹配 (Video Extraction) (2) 支柱七：动作重定向 (Motion Retargeting) (1)

🔬 支柱一：机器人控制 (Robot Control) (43 篇)

#题目一句话要点标签🔗⭐
1 ECHO-G: Embodied Co-speech Humanoid mOtion Generation 提出ECHO-G框架以生成类人机器人共语动作 humanoid humanoid robot flow matching
2 IronMind: Scaling Humanoid Dexterous Manipulation via Camera-Space Ego-Centric Pretraining 提出IronMind以解决人形机器人灵巧操作的挑战 humanoid humanoid robot manipulation
3 OccluDex: Hierarchical 3D Visuo-Tactile Representation Learning for Egocentric Dexterous Manipulation under Self-Occlusion 提出OccluDex以解决自遮挡下的灵巧操控问题 humanoid manipulation dexterous manipulation
4 Locomotion-Grounded Humanoid Soccer: Task-Gated Reinforcement Learning of a Multi-Directional Kicking Library 提出基于任务门控强化学习的多方向踢球库以解决人形足球运动控制问题 humanoid motion tracking locomotion
5 NEXUS: Perceptive Whole-Body Control for Terrain-Adaptive Teleoperation 提出NEXUS以解决地形适应性远程操作问题 humanoid humanoid robot whole-body control
6 The Planning Limits of Latent World Models 研究潜在世界模型的规划极限以提升机器人行为预测能力 manipulation MPC world model
7 DSDyn-VLA: A Dual-Stream Dynamic Manipulation Framework with Motion Perception, Future Awareness, and Realtime Correction 提出DSDyn-VLA以解决动态环境中的视觉-语言-动作问题 manipulation optical flow vision-language-action
8 Cue the Flow: Steering Flow-Matching Policies for Open-World Delivery Manipulation 提出流匹配策略以解决开放世界物品配送操控问题 manipulation flow matching vision-language-action
9 CEER2: Directional and Tunable End-Effector and Root Compliance for Humanoid Loco-Manipulation 提出方向性和可调节末端执行器及根部柔顺性以解决人形机器人操控问题 humanoid manipulation loco-manipulation
10 Beyond Policy Alignment: Closing the Planning-Learning Loop for Robot Control with Learned World Models 提出PL-MPC以优化机器人控制中的规划与学习闭环 humanoid sim-to-real MPC
11 Online Evolution Strategy for Flow-Matching VLA Policies via Self-Supervised Trajectory Distribution Optimization 提出在线进化策略以优化流匹配VLA策略的轨迹分布 manipulation flow matching vision-language-action
12 From Local Whole-Body VLA Behaviors to Scene-Scale Aerial Manipulation 提出统一框架以解决场景规模无人机操控中的挑战 manipulation vision-language-action VLA
13 Multi-Link Safety Filtering for VLA Policies Around Moving Hazards 提出多链接安全过滤以解决动态危险物体的操控安全问题 manipulation flow matching optical flow ✅
14 UniWAM Technical Report: Unified Mobile Manipulation via Mixed-Stream World-Action Modeling and Manipulation Anchor Pose Supervision 提出UniWAM以解决移动操作中的导航与操控问题 manipulation mobile manipulation policy learning
15 LocoWM: High-Precision Locomotion through World-Model-Guided Residual Adaptation 提出LocoWM框架以解决高精度机器人运动控制问题 locomotion push recovery world model ✅
16 Experience-Driven Continual Learning of Terrain Traversability for Quadruped Robots 提出基于经验驱动的持续学习方法以提升四足机器人地形可通行性预测 quadruped locomotion Unitree
17 Make Code as Policy Great Again: Frontier Agents Write, Call, and Evolve Robot Tools 提出URAI框架以提升机器人工具的编写与执行效率 manipulation dual-arm foundation model
18 Rethinking Legibility in Social Robot Hallway Navigation: Impact of Intent Representation and Human Distraction 提出意图表示以提升社交机器人走廊导航的可读性 model predictive control motion generation human motion ✅
19 Learning from Runtime Feedback through Failure-Bank Self-Evolution for Vision-Language-Action Models 提出FailBank框架以解决VLA模型中的反馈与策略不匹配问题 manipulation vision-language-action VLA
20 Exploiting Vulnerabilities: Universal Adversarial Attacks on Vision-Language-Action Models in Robotics 提出通用对抗物体以解决机器人视觉语言行动模型的安全问题 manipulation vision-language-action VLA
21 SplineWAM: Adaptive Action Horizons for World Action Models via B-Spline Representations 提出SplineWAM以解决世界动作模型的时间分配问题 manipulation bi-manual world action model
22 Magic-W0: A Structured World-Action Foundation Model for Physical Intelligence 提出Magic-W0以解决物理智能中的结构化世界-动作建模问题 manipulation egocentric foundation model
23 RoboAssist: Interactive Human-Humanoid Planning for Long-Horizon Surgical Assistance 提出RoboAssist以解决长时间手术辅助中的人机协作问题 humanoid humanoid robot Unitree
24 Dream4ACT: A Shared Visual Action Interface for Multi-Embodiment Video-Action Modeling 提出Dream4ACT以解决多体视频动作建模中的动作表示统一问题 manipulation flow matching world model
25 RoboCoach: World Models as Active Coaches for Compositional Robot Skills 提出RoboCoach以解决长时间机器人操作技能组合问题 manipulation world model world models
26 Tactile Curiosity Drives Robot Interaction 提出TacEx框架以解决机器人操控技能样本效率低下问题 manipulation reinforcement learning vision-language-action
27 Discrete Forcing: Infusing Discrete Guidance into Continuous Denoising for Few-Step Action Experts 提出离散强制方法以解决视觉-语言-动作模型的高效动作生成问题 manipulation flow matching vision-language-action
28 Video2SwimFish: An Automated Pipeline for Reconstructing Controllable Fish Models and Biological Locomotion from Real Fish Videos 提出Video2SwimFish以自动重建可控鱼类模型和生物运动 locomotion embodied AI ✅
29 SimEX: Simulation-Integrated Robotics AutoResearch 提出SimEX框架以解决机器人物理能力获取难题 manipulation large language model Code as Policies
30 Local-Minimum Escaper: Programmatic Subgoal Generation for Robust Navigation in Unknown Environments 提出LME框架以解决移动机器人在未知环境中的局部最小值问题 quadruped reinforcement learning deep reinforcement learning
31 A Biophysically Detailed C. elegans Circuit as a Task-Agnostic Dynamical Core for Visually Robust Robot Manipulation 提出生物物理详细的线虫电路以解决机器人视觉鲁棒性问题 manipulation diffusion policy
32 DODGER: Safety-Guided Reinforcement Learning for Robot Navigation Among Dynamic Obstacles 提出DODGER以解决动态障碍物下的机器人安全导航问题 humanoid reinforcement learning
33 Passive Stiffness Shaping in Cable-Suspended Aerial Manipulation via Movable Compliant Anchors 提出可移动柔性锚点以优化缆索悬挂空中操作的被动刚度 manipulation
34 LIBERO-Agent: Evaluating General-Purpose Agents for Direct Embodied Manipulation 提出LIBERO-Agent以评估通用代理在机器人操作中的表现 manipulation
35 Benchmarking and Enhancing Skill-Level Memory for Partially Observable Robotic Manipulation 提出HIDE基准与SEEK框架以增强部分可观测机器人操作的技能记忆 manipulation
36 DynaHarness: A Dynamic Physical Harness for Self-Evolving Robot Agents 提出DynaHarness以解决长时间操作中的语义推理与物理执行协调问题 manipulation VLA ✅
37 PhasePlan: Ordered Future-Phase Planning for Robot Brain Models 提出有序未来阶段规划以解决机器人动作生成问题 manipulation multimodal
38 Text-to-3D Policy: Fine-Grained Language-Behavior Alignment for Unseen Specification Generalization 提出T3DP框架以解决文本到3D策略中的细粒度行为对齐问题 manipulation diffusion policy
39 Occlusion-Aware, Quasi-Static, Stability-Oriented Trajectory Planning on Uneven Terrain 提出基于模型的框架以解决不均匀地形下的自主导航问题 trajectory optimization flow matching
40 Magnetic based In-situ Self 3D Pose Estimation for a Modular Soft Tendon-Driven Continuum Robot via IMU-Fusion 提出基于磁场的自适应3D姿态估计方法以解决连续机器人定位问题 manipulation
41 Tool-Policy Co-Design for Powder Weighing in Laboratory Automation 提出工具与策略协同设计以解决实验室自动化中的粉末称量问题 manipulation
42 Making Waves: A Membrane-Coupled Delta Array for Manipulating Objects Below the Actuator Spacing 提出膜耦合的Delta阵列以解决操控小物体的问题 manipulation
43 SteerQuant: Steering Quantization Error with Action-Guided Scaling in World-Action Models 提出SteerQuant以解决世界-动作模型中的量化误差问题 dual-arm

🔬 支柱二：RL算法与架构 (RL & Architecture) (12 篇)

#题目一句话要点标签🔗⭐
44 DiFF: Doppler-informed Flow Matching for Human Motion Flow 提出DiFF以解决4D毫米波雷达人类运动流估计问题 flow matching scene flow human motion
45 Social-WM: Safety-Aware Latent World Models for Robot Social Navigation 提出Social-WM以解决机器人安全社交导航问题 reinforcement learning world model world models
46 DiffWAM: A Fast and Efficient Navigation World Action Model 提出DiffWAM以解决无人机导航中的视频运动转换问题 world action model world action models dreamer ✅
47 MotionWeave: Learning Motion-Centered Future Dynamics for Vision-Language-Action Policies 提出MotionWeave以解决视觉-语言-动作模型的动态预测问题 world model world models vision-language-action ✅
48 Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation 提出阶段感知的两步流去噪以解决VLA模型实时性问题 flow matching vision-language-action VLA
49 ReWAM: Reciprocal World Action Models for Interactive Autonomous Driving 提出ReWAM以解决互动场景下自主驾驶的决策问题 flow matching world model world models
50 Linear Recurrent Memory Suffices to Distil a World-Model Policy for Robot Air Hockey 提出线性递归记忆以解决机器人空气曲棍球中的控制问题 world model world models dreamer
51 Sparse-WAM: Accelerating World Action Models via Action-Guided Sparse Imagination 提出Sparse-WAM以加速世界动作模型推理 world action model world action models
52 PrefPI: Preference-Guided Steering into Out-of-Distribution Behaviors 提出PrefPI以引导机器人政策超越分布外行为 diffusion policy preference learning classifier-free guidance
53 Looking Back to Move Forward: Temporal Verification for Generative Robot Policies 提出时序验证框架以解决生成机器人策略中的动作验证问题 imitation learning flow matching VLA
54 PRICE the Action Chunks: Physical Relational Credit Assignment for Embodied Reinforcement Learning 提出PRICE方法以解决嵌入式强化学习中的信用分配问题 reinforcement learning
55 Identifiable Decomposition of Submovements in Human Hand Trajectories 提出Sub-ID方法以解决人手轨迹分解问题 imitation learning spatiotemporal

🔬 支柱三：空间感知与语义 (Perception & Semantics) (5 篇)

#题目一句话要点标签🔗⭐
56 STARS: From Spatiotemporal Dynamics to Social Representations in Human-Robot Interaction 提出SocialNav-SUB基准以解决社交机器人导航场景理解问题 scene understanding spatiotemporal foundation model ✅
57 FORTE: Forecasting Occupancy for Spatiotemporal Risk-Aware Planning in Dynamic Environments 提出FORTE以解决动态环境中的安全导航问题 occupancy grid spatiotemporal
58 Centralized Multi-UAV Exploration and 3D Reconstruction Using Single-UAV Planners 提出集中式多无人机探索框架以解决安全导航与映射一致性问题 3D reconstruction
59 LBDU-VIO: Learned Bias Dynamics and Uncertainty for Visual-Inertial Odometry with Unreliable Vision 提出LBDU-VIO以解决视觉信息不可靠问题 VIO
60 MVP-SLAM: Multi-Camera Visual-Inertial Floorplan-Prior SLAM 提出MVP-SLAM以解决室内建筑工地的视觉SLAM漂移问题 visual SLAM

🔬 支柱九：具身大模型 (Embodied Foundation Models) (3 篇)

#题目一句话要点标签🔗⭐
61 When Instructions Retrieve Trajectories: Diagnosing and Mitigating Generalization Failures in VLA Models 提出等变反事实训练以解决VLA模型的指令-动作绑定问题 vision-language-action VLA
62 Blackout vs. Freeze: Analyzing Physical Failure Modes of VLAs under Camera Faults 分析视觉-语言-动作模型在摄像头故障下的物理失效模式 vision-language-action VLA
63 TACTIC: Temporal and Context-Aware LLM Tactical Planning for Roadside LiDAR Attacks 提出TACTIC框架以解决路边LiDAR攻击的场景感知问题 large language model multimodal

🔬 支柱六：视频提取与匹配 (Video Extraction) (2 篇)

#题目一句话要点标签🔗⭐
64 EWAM: Emergent Depth-Wise Specialization in a Unified Embodied Model -- From Semantic Understanding through Visual Foresight to Action 提出EWAM模型以解决多模态理解与动作生成问题 egocentric cross-embodiment embodiment transfer
65 Ego4WAM: What Matters When Scaling Egocentric Human Data for Robot Learning? 提出Ego4WAM以优化机器人学习中的自我中心人类数据使用 egocentric

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
66 Function beyond Form: Functional Correspondence for Cross-Embodiment Dexterous Grasp Generation 提出FunCo-Grasp以解决跨实体灵巧抓取生成问题 cross-embodiment

⬅️ 返回 cs.RO 首页 · 🏠 返回主页