cs.RO（2026-03-09）

📊 共 37 篇论文 | 🔗 6 篇有代码

🎯 兴趣领域导航

支柱一：机器人控制 (Robot Control) (26 🔗5) 支柱九：具身大模型 (Embodied Foundation Models) (4 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (4) 支柱三：空间感知与语义 (Perception & Semantics) (3)

🔬 支柱一：机器人控制 (Robot Control) (26 篇)

#题目一句话要点标签🔗⭐
1 MetaWorld-X: Hierarchical World Modeling via VLM-Orchestrated Experts for Humanoid Loco-Manipulation MetaWorld-X：通过VLM协调专家实现人型机器人运动操作的分层世界建模 humanoid humanoid robot humanoid control
2 Towards Human-Like Manipulation through RL-Augmented Teleoperation and Mixture-of-Dexterous-Experts VLA 提出基于强化学习辅助遥操作和专家混合VLA的类人灵巧操作方法 manipulation dexterous manipulation in-hand manipulation
3 Omnidirectional Humanoid Locomotion on Stairs via Unsafe Stepping Penalty and Sparse LiDAR Elevation Mapping 提出基于稀疏LiDAR和不安全步进惩罚的人形机器人楼梯全向行走方法 humanoid humanoid robot humanoid locomotion
4 EquiBim: Learning Symmetry-Equivariant Policy for Bimanual Manipulation EquiBim：学习对称等变策略的双臂操作模仿学习框架 manipulation bi-manual dual-arm
5 AtomVLA: Scalable Post-Training for Robotic Manipulation via Predictive Latent World Models AtomVLA：通过预测潜在世界模型实现机器人操作的可扩展后训练 manipulation world model vision-language-action
6 StructBiHOI: Structured Articulation Modeling for Long--Horizon Bimanual Hand--Object Interaction Generation StructBiHOI：用于长时程双手-物体交互生成的结构化关节建模框架 manipulation bi-manual bimanual manipulation
7 3PoinTr: 3D Point Tracks for Robot Manipulation Pretraining from Casual Videos 提出3PoinTr，利用3D点轨迹从日常视频中预训练机器人操作策略。 manipulation teleoperation behavior cloning ✅
8 Embedding Classical Balance Control Principles in Reinforcement Learning for Humanoid Recovery 提出嵌入经典平衡控制原则的强化学习方法，提升人形机器人跌倒后的恢复能力。 humanoid humanoid robot Unitree
9 PhaForce: Phase-Scheduled Visual-Force Policy Learning with Slow Planning and Fast Correction for Contact-Rich Manipulation PhaForce：面向接触式操作，基于相位调度的视觉-力觉策略学习框架 manipulation policy learning contact-aware
10 STRIDE: Structured Lagrangian and Stochastic Residual Dynamics via Flow Matching 提出STRIDE框架以解决不确定环境下的机器人动态建模问题 quadruped humanoid Unitree
11 MoMaStage: Skill-State Graph Guided Planning and Closed-Loop Execution for Long-Horizon Indoor Mobile Manipulation MoMaStage：技能状态图引导的长时程室内移动操作规划与闭环执行 manipulation mobile manipulation ✅
12 Multifingered force-aware control for humanoid robots 提出一种力感知的多指控制方法，用于人形机器人稳定操作 humanoid humanoid robot ✅
13 RoboRouter: Training-Free Policy Routing for Robotic Manipulation RoboRouter：一种用于机器人操作的免训练策略路由方法 manipulation vision-language-action VLA
14 Vector Field Augmented Differentiable Policy Learning for Vision-Based Drone Racing DiffRacing：提出向量场增强的可微策略学习框架，用于视觉无人机竞速。 quadruped locomotion sim-to-real
15 CONTACT: CONtact-aware TACTile Learning for Robotic Disassembly 提出基于触觉感知的机器人拆卸学习框架，提升接触密集型任务性能 manipulation contact-aware
16 LAR-MoE: Latent-Aligned Routing for Mixture of Experts in Robotic Imitation Learning 提出LAR-MoE，通过潜在空间对齐路由解决机器人模仿学习中专家网络技能分解难题 manipulation policy learning imitation learning
17 Dual-Horizon Hybrid Internal Model for Low-Gravity Quadrupedal Jumping with Hardware-in-the-Loop Validation 提出双视野混合内部模型，解决低重力四足机器人连续跳跃难题 quadruped locomotion
18 FlowTouch: View-Invariant Visuo-Tactile Prediction FlowTouch：提出一种视角不变的视觉-触觉预测模型，用于提升机器人操作任务的感知能力。 manipulation sim-to-real flow matching
19 SAIL: Test-Time Scaling for In-Context Imitation Learning with VLM SAIL：利用VLM进行上下文模仿学习的测试时扩展，提升机器人泛化能力 manipulation imitation learning
20 TRIAGE: Type-Routed Interventions via Aleatoric-Epistemic Gated Estimation in Robotic Manipulation and Adaptive Perception -- Don't Treat All Uncertainty the Same TRIAGE：通过概率-认知门控估计实现机器人操作和自适应感知中的类型路由干预 manipulation ✅
21 NaviDriveVLM: Decoupling High-Level Reasoning and Motion Planning for Autonomous Driving NaviDriveVLM：解耦高层推理与运动规划，提升自动驾驶性能 motion planning
22 Viewpoint-Agnostic Grasp Pipeline using VLM and Partial Observations 提出一种基于VLM和部分观测的视角无关抓取流程，提升腿式机器人复杂环境下的抓取成功率。 quadruped open-vocabulary open vocabulary
23 See and Switch: Vision-Based Branching for Interactive Robot-Skill Programming 提出基于视觉的交互式机器人技能编程框架，实现条件分支和异常检测。 manipulation dexterous manipulation
24 Exp-Force: Experience-Conditioned Pre-Grasp Force Selection with Vision-Language Models Exp-Force：利用视觉-语言模型和经验学习进行预抓取力选择 manipulation MAE ✅
25 Bilevel Planning with Learned Symbolic Abstractions from Interaction Data 提出一种双层神经符号规划框架，利用交互数据学习符号抽象，提升复杂环境下的规划性能。 manipulation
26 Tactile Recognition of Both Shapes and Materials with Automatic Feature Optimization-Enabled Meta Learning 提出AFOP-ML框架，实现触觉传感器对形状和材质的少样本元学习识别。 manipulation

🔬 支柱九：具身大模型 (Embodied Foundation Models) (4 篇)

#题目一句话要点标签🔗⭐
27 Seed2Scale: A Self-Evolving Data Engine for Embodied AI via Small to Large Model Synergy and Multimodal Evaluation Seed2Scale：通过小模型收集、大模型评估实现具身智能的自进化数据引擎 embodied AI vision-language-action multimodal ✅
28 SaiVLA-0: Cerebrum--Pons--Cerebellum Tripartite Architecture for Compute-Aware Vision-Language-Action 提出SaiVLA-0：一种用于计算感知视觉-语言-动作的三脑架构 vision-language-action multimodal
29 RAPID: Redundancy-Aware and Compatibility-Optimal Edge-Cloud Partitioned Inference for Diverse VLA models RAPID：面向多样VLA模型的冗余感知与兼容性最优边缘云协同推理 vision-language-action VLA
30 Less is More: Robust Zero-Communication 3D Pursuit-Evasion via Representational Parsimony 通过表征简约性实现鲁棒的零通信3D追逐-逃逸 zero-shot transfer

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#题目一句话要点标签🔗⭐
31 Choose What to Observe: Task-Aware Semantic-Geometric Representations for Visuomotor Policy 提出任务感知语义-几何表示，提升视觉运动策略对外观变化的鲁棒性 flow matching monocular depth Depth Anything
32 CRED: Counterfactual Reasoning and Environment Design for Active Preference Learning CRED：基于环境设计和反事实推理的主动偏好学习方法 preference learning
33 DeReCo: Decoupling Representation and Coordination Learning for Object-Adaptive Decentralized Multi-Robot Cooperative Transport DeReCo：解耦表征与协同学习，实现对象自适应的多机器人协同搬运 reinforcement learning representation learning privileged information
34 Diff-Muscle: Efficient Learning for Musculoskeletal Robotic Table Tennis Diff-Muscle：利用微分平坦性高效学习肌肉骨骼机器人打乒乓球 reinforcement learning policy learning

🔬 支柱三：空间感知与语义 (Perception & Semantics) (3 篇)

#题目一句话要点标签🔗⭐
35 UniGround: Universal 3D Visual Grounding via Training-Free Scene Parsing UniGround：通过无训练场景解析实现通用3D视觉定位 open-vocabulary open vocabulary spatial relationship
36 AffordGrasp: Cross-Modal Diffusion for Affordance-Aware Grasp Synthesis AffordGrasp：提出基于跨模态扩散的抓取姿态合成方法，提升语义感知能力。 affordance affordance-aware embodied AI
37 R2F: Repurposing Ray Frontiers for LLM-free Object Navigation R2F：利用射线前沿重构，实现无LLM的物体导航 open-vocabulary open vocabulary VLN

⬅️ 返回 cs.RO 首页 · 🏠 返回主页