cs.RO（2026-01-29）

📊 共 25 篇论文

🎯 兴趣领域导航

支柱一：机器人控制 (Robot Control) (21) 支柱三：空间感知与语义 (Perception & Semantics) (2) 支柱九：具身大模型 (Embodied Foundation Models) (1) 支柱二：RL算法与架构 (RL & Architecture) (1)

🔬 支柱一：机器人控制 (Robot Control) (21 篇)

#题目一句话要点标签🔗⭐
1 CoFreeVLA: Collision-Free Dual-Arm Manipulation via Vision-Language-Action Model and Risk Estimation CoFreeVLA：基于视觉-语言-动作模型和风险估计的双臂无碰撞操作 manipulation bi-manual dual-arm
2 AIR-VLA: Vision-Language-Action Systems for Aerial Manipulation 提出AIR-VLA：面向空中操作的视觉-语言-动作系统基准 manipulation vision-language-action VLA
3 DynamicVLA: A Vision-Language-Action Model for Dynamic Object Manipulation DynamicVLA：用于动态物体操作的视觉-语言-动作模型 manipulation teleoperation cross-embodiment
4 Towards Bridging the Gap between Large-Scale Pretraining and Efficient Finetuning for Humanoid Control 提出基于SAC预训练和模型预测控制微调的人形机器人控制方法 humanoid humanoid control humanoid locomotion
5 DexTac: Learning Contact-aware Visuotactile Policies via Hand-by-hand Teaching DexTac：通过手把手示教学习接触感知型灵巧操作策略 manipulation dexterous hand dexterous manipulation
6 MoE-ACT: Improving Surgical Imitation Learning Policies through Supervised Mixture-of-Experts 提出MoE-ACT，通过监督式混合专家模型提升手术模仿学习策略。 manipulation imitation learning vision-language-action
7 Abstracting Robot Manipulation Skills via Mixture-of-Experts Diffusion Policies 提出基于混合专家扩散策略的机器人操作技能抽象方法，提升多任务学习效率。 manipulation dual-arm diffusion policy
8 From Instruction to Event: Sound-Triggered Mobile Manipulation 提出基于声音触发的移动操作方法，提升机器人自主性和环境适应性 manipulation mobile manipulation
9 GAZELOAD A Multimodal Eye-Tracking Dataset for Mental Workload in Industrial Human-Robot Collaboration GAZELOAD：用于工业人机协作中精神负荷评估的多模态眼动追踪数据集 manipulation multimodal
10 HPTune: Hierarchical Proactive Tuning for Collision-Free Model Predictive Control 提出HPTune：一种用于无碰撞模型预测控制的分层主动调参框架 MPC model predictive control motion planning
11 Disentangling perception and reasoning for improving data efficiency in learning cloth manipulation without demonstrations 解耦感知与推理，提升无示教cloth操作强化学习的数据效率 manipulation sim-to-real reinforcement learning
12 LLM-Driven Scenario-Aware Planning for Autonomous Driving 提出基于LLM的场景感知规划方法LAP，提升自动驾驶在复杂交通环境下的效率与安全性 model predictive control motion planning scene understanding
13 Information Filtering via Variational Regularization for Robot Manipulation 提出变分正则化以解决机器人操作中的信息过滤问题 manipulation
14 Multi-Modular MANTA-RAY: A Modular Soft Surface Platform for Distributed Multi-Object Manipulation 提出多模块MANTA-RAY软表面平台，用于分布式多物体操作，提升可扩展性。 manipulation
15 Spotlighting Task-Relevant Features: Object-Centric Representations for Better Generalization in Robotic Manipulation 提出基于Slot的对象中心表示SBOCR，提升机器人操作策略在视觉变化下的泛化性。 manipulation
16 PocketDP3: Efficient Pocket-Scale 3D Visuomotor Policy PocketDP3：高效的口袋级3D视觉运动策略，显著降低模型参数量 manipulation diffusion policy distillation
17 Nimbus: A Unified Embodied Synthetic Data Generation Framework Nimbus：统一具身智能合成数据生成框架，提升数据吞吐量。 manipulation foundation model
18 Towards Space-Based Environmentally-Adaptive Grasping 提出基于潜在空间的强化学习方法，解决空间环境下的自适应抓取问题 manipulation reinforcement learning SAC
19 Disturbance-Aware Flight Control of Robotic Gliding Blimp via Moving Mass Actuation 提出基于移动质量驱动的扰动感知滑翔飞艇飞行控制方法，提升抗风能力。 MPC model predictive control
20 mjlab: A Lightweight Framework for GPU-Accelerated Robot Learning mjlab：一个轻量级的GPU加速机器人学习框架 manipulation
21 Macro-Scale Electrostatic Origami Motor 提出一种宏观尺度静电折纸电机，实现可折叠的连续旋转运动 locomotion

🔬 支柱三：空间感知与语义 (Perception & Semantics) (2 篇)

#题目一句话要点标签🔗⭐
22 InspecSafe-V1: A Multimodal Benchmark for Safety Assessment in Industrial Inspection Scenarios InspecSafe-V1：用于工业巡检场景安全评估的多模态基准数据集 scene understanding foundation model multimodal
23 DSCD-Nav: Dual-Stance Cooperative Debate for Object Navigation 提出DSCD-Nav，通过双立场协同辩论提升零样本物体导航的可靠性。 scene understanding

🔬 支柱九：具身大模型 (Embodied Foundation Models) (1 篇)

#题目一句话要点标签🔗⭐
24 IROS: A Dual-Process Architecture for Real-Time VLM-Based Indoor Navigation 提出IROS双过程架构，用于基于VLM的实时室内导航 vision-language-action VLA

🔬 支柱二：RL算法与架构 (RL & Architecture) (1 篇)

#题目一句话要点标签🔗⭐
25 Training slow silicon neurons to control extremely fast robots with spiking reinforcement learning 利用脉冲强化学习训练硅神经元，控制极高速机器人进行气垫球博弈 reinforcement learning

⬅️ 返回 cs.RO 首页 · 🏠 返回主页