cs.RO（2026-04-30）

📊 共 23 篇论文 | 🔗 6 篇有代码

🎯 兴趣领域导航

支柱一：机器人控制 (Robot Control) (11 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (10 🔗2) 支柱三：空间感知与语义 (Perception & Semantics) (1 🔗1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱一：机器人控制 (Robot Control) (11 篇)

#题目一句话要点标签🔗⭐
1 LaST-R1: Reinforcing Action via Adaptive Physical Latent Reasoning for VLA Models LaST-R1：通过自适应物理潜在推理强化VLA模型的动作执行 manipulation dual-arm reinforcement learning
2 Simulating Infant First-Person Sensorimotor Experience via Motion Retargeting from Babies to Humanoids 提出一种婴儿第一人称视角传感器运动经验模拟框架，用于机器人和发育科学研究。 humanoid humanoid robot motion retargeting ✅
3 ExoActor: Exocentric Video Generation as Generalizable Interactive Humanoid Control ExoActor：利用第三人称视频生成实现通用人形机器人交互控制 humanoid humanoid control human motion
4 RopeDreamer: A Kinematic Recurrent State Space Model for Dynamics of Flexible Deformable Linear Objects RopeDreamer：利用运动学循环状态空间模型进行柔性线性物体动力学建模 manipulation dreamer latent dynamics
5 Function-based Parametric Co-Design Optimization of Dexterous Hands 提出基于函数的灵巧手参数化协同设计优化框架 manipulation dexterous hand cross-embodiment
6 FlexiTac: A Low-Cost, Open-Source, Scalable Tactile Sensing Solution for Robotic Systems FlexiTac：一种低成本、开源、可扩展的机器人触觉传感解决方案 sim-to-real contact-aware cross-embodiment ✅
7 Robot Learning from Human Videos: A Survey 综述：基于人类视频的机器人学习技术，促进通用机器人系统可扩展学习。 manipulation policy learning embodied AI ✅
8 MotuBrain: An Advanced World Action Model for Robot Control MotuBrain：用于机器人控制的先进世界行动模型 humanoid policy learning world model
9 Task-Conditioned Uncertainty Costmaps for Legged Locomotion 提出任务条件不确定性代价地图，提升腿足机器人复杂地形运动规划的鲁棒性 legged robot legged locomotion locomotion
10 Lucid-XR: An Extended-Reality Data Engine for Robotic Manipulation Lucid-XR：用于机器人操作的扩展现实数据引擎，提升真实环境零样本迁移能力 manipulation dexterous manipulation human-to-robot
11 E$^2$DT: Efficient and Effective Decision Transformer with Experience-Aware Sampling for Robotic Manipulation 提出E$^2$DT，通过经验感知采样提升决策Transformer在机器人操作中的效率和效果 manipulation reinforcement learning policy learning

🔬 支柱二：RL算法与架构 (RL & Architecture) (10 篇)

#题目一句话要点标签🔗⭐
12 GSDrive: Reinforcing Driving Policies by Multi-mode Trajectory Probing with 3D Gaussian Splatting Environment GSDrive：利用3D高斯溅射环境多模态轨迹探测增强自动驾驶策略 reinforcement learning imitation learning flow matching ✅
13 MotuBrain: An Advanced World Action Model for Robot Control MotuBrain：用于机器人控制的先进世界行为模型 policy learning world model world models
14 Can Tabular Foundation Models Guide Exploration in Robot Policy Learning? TFM-S3：利用表格基础模型引导机器人策略学习中的探索，提升样本效率。 policy learning TD3 foundation model
15 Flying by Inference: Active Inference World Models for Adaptive UAV Swarms 提出基于主动推理世界模型的自适应无人机集群轨迹规划框架 world model world models motion adaptation
16 Dreaming Across Towns: Semantic Rollout and Town-Adversarial Regularization for Zero-Shot Held-Out-Town Fixed-Route Driving in CARLA 提出语义Rollout与城镇对抗正则化，提升CARLA中零样本跨城镇固定路线驾驶性能 world model world models dreamer
17 RAY-TOLD: Ray-Based Latent Dynamics for Dense Dynamic Obstacle Avoidance with TDMPC RAY-TOLD：基于光线的潜在动态模型用于高密度动态避障 reinforcement learning latent dynamics
18 GSDrive: Reinforcing Driving Policies by Multi-mode Trajectory Probing with 3D Gaussian Splatting Environment GSDrive：利用3D高斯溅射环境进行多模态轨迹探测，强化端到端自动驾驶策略。 reinforcement learning imitation learning flow matching ✅
19 World Model for Robot Learning: A Comprehensive Survey 机器人学习中的世界模型综述：全面回顾与未来展望 reinforcement learning policy learning world model
20 Being-H0.7: A Latent World-Action Model from Egocentric Videos Being-H0.7：一种基于自中心视频的潜在世界-动作模型，提升机器人控制效率。 world model world models egocentric
21 Dynamic-TD3: A Novel Algorithm for UAV Path Planning with Dynamic Obstacle Trajectory Prediction 提出Dynamic-TD3算法，解决无人机在动态障碍物环境中安全路径规划问题 reinforcement learning deep reinforcement learning DRL

🔬 支柱三：空间感知与语义 (Perception & Semantics) (1 篇)

#题目一句话要点标签🔗⭐
22 FreeOcc: Training-Free Embodied Open-Vocabulary Occupancy Prediction 提出FreeOcc，一种无需训练的具身开放词汇占据预测框架 open-vocabulary open vocabulary ✅

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
23 SASI: Leveraging Sub-Action Semantics for Robust Early Action Recognition in Human-Robot Interaction SASI：利用子动作语义实现人机交互中鲁棒的早期动作识别 spatiotemporal

⬅️ 返回 cs.RO 首页 · 🏠 返回主页