Archive
Papers
211 robotics papers the field upvoted. The homepage shows this week; this page remembers all of them.
- PDRVLA: A Vision-Language-Action Model with Physical Dynamics Representations
- ThinkWorld: World Imagination Reinforces Chain-of-Dynamics Reasoning
- Zooming without Zooming v2: From Seeing Details to Connecting Evidence for Generalizable Multimodal Perception
- Can LLMs Uncover Hidden Physical Laws through Free Exploration?
- The Trinity:Capability-Driven Recursive Self-Improvement for Multimodal Agents
- Rollback: Experience Backtracking in World Models for Policy Improvement
- InteractiveWorld: A Verifiable Human-Centric Environment for Building Better Interactive MLLM Assistants
- RynnVLA-Latent: Scaling Robotic Manipulation Foundation Models with 3D-Grounded Latent Actions
- HuGe: Scaling Humanoid Generalist Vision-Language-Action Models with Egocentric Experience
- AV-FABLE: Audio-Visual Manipulation Forensics in Long-Form Videos
- RynnUnify: Towards Unified Embodied Brain
- World-in-Context: Interactive Video World Modeling with In-Context Conditioning
- DiscoDemo: Discovering Efficient and Diverse Robot Demonstrations for Imitation Learning
- Sensorimotor Alignment: A Platonic Proxy for Embodied Policy Success
- HiSTEVE 1.0: Harnessing Human-Interface VLA Agents for Open-World Games
- AgentVQA: A Unified Benchmark for Agentic Visual Understanding
- DONIC: Generalist Dexterous Bimanual Tracking Control from Human Demonstrations
- Multimodal Embodied Agents: A Unified Capability-centric Perspective from Computer-Use to Robot-Use
- LaLa-World: Scaling Latent World Models with Natural Language Actions
- Toward Recursive Self-Improvement in Physical AI
- Self-Distilling Future Video Representations for World Action Models
- Learning the 4D World through Point Trajectory Forecasting
- Joints, Camera, Action! Co-Generating 3D Motion and Video for Arbitrary Rigs
- AgenticRoboRL: Bringing Coding-Agent Post-Training to Agentic Robotics
- Scaling in Games: Continued Pre-Training For Embodied Agents In Diverse Virtual Worlds
- ScaleNav: Scalable Internet Video Pretraining for Language-Conditioned Local Navigation
- Learning from Fleet Experience: A Collective Learning Paradigm for Robots
- EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience
- RoboValue: A Fine-Grained Sim-and-Real Benchmark for Unified Evaluation of Robotic Value Models
- Where is Physical RSI capability from? An Observational Scaling Law of LLM-as-policy
- Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience
- On the Fragility of Tasks under Composition: Implicit Budgeting in Language Models
- Omni-Woodpecker: Hallucination Correction via Evidence-Prior Rebalancing for Omni LLMs
- AffordanceWAM: Affordance-Aware Joint World-Action Modeling for Robot Manipulation
- VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse
- Editing Images into Reusable Layers via Compositional Edit-Prior Distillation
- HexaShell: A Thin Harness for Exploring the Limits of VLM Agents as Robot Coding Agents
- VLA-Δ: Learning Physical State Transition for Vision-Language-Action Models
- Velox: Spatial Memory Makes World Models Efficient
- Learning Self-Orchestration in Decentralized Agent Swarms
- Can World and Action Modeling Benefit Each Other under Unified Causal Training?
- CarryOn: A Dual-Loop Harness for Personalized and Reliable Robot Assistance
- CoDA: Co-Evolution of Data Engines and Action Policies
- Has GPT-6-Astra Achieved Embodied Intelligence?
- OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents
- Towards Agentic Robots: Model, Data, Environment, and Harness
- SafeDriveVLA: Navigation-Conditioned World Model Dreaming for Conflict-Aware End-to-End Autonomous Driving
- G^3VLA: Geometric Inductive Bias for Vision-Language-Action Models
- Generative World Simulation with DAO and JING
- From Failures to Learnable Worlds: Policy-Conditioned Environment Generation for Navigation Agents
- BlenderWorld: Benchmarking Visual Agents for 3D Creation from Real-World References
- EgoExplorer: A Vision-Motion-Touch Observatory for Human Loco-Manipulation In Situ
- WAVE-0: Bridging VLA Policies and WAMs with Lightweight Latent Future Prediction
- GreenWAM: A World Foundation Model for Robotics
- Cepheid: Decoupling Scale from Structure for Cross-Family Motion Completion
- ACCRUE: Continual Off-Policy Reinforcement Learning for Robot Foundation Models
- DecisionRobust: Benchmarking Robustness for JEV Style Decision Models Across Text and Vision
- AUGUstus: Autonomous Driving VLA Model with Unified Generation and Understanding
- Data-Centric Recursive Self-Improvement for Navigation via Policy-Synchronized Language-Action Dual Bootstrapping
- BridgeWAM: Bridging Video and Action Experts via Latent Bridge Queries
- EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents
- The value of information and the emergence of communication in evolved e-puck robots
- Learning Transferable World-Action Models from Task-Paired Human Videos
- SuperNav: An Agentic Navigation System for Any Task in Any Scene
- Multi-Agent Egocentric World Model with Fine-Grained Embodied Interaction
- DreamTrue: Action-Faithful Robot World Model with Counterfactual Post-Training
- USDCraft: Geometrically Grounded Programmatic Modeling of Articulated 3D Assets for Simulation
- Embodied Turing Machines: Stateful Code for Robot Recursive Self-Improvement
- Reflex: Visually Grounded Reactive Humanoid Control
- Humanity's Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models
- LeWAM: End-to-End World and Action Modeling with JEPAs
- Long-WAM: Scaling the Context of World-Action Models
- RobotWorld: Benchmarking Multimodal Agents for Robot Use Across Diverse Tasks and Embodiments
- RLHND: Video Foundation Models as Physically Grounded Hand Trackers for Robot Learning
- Q-Learning with Scalar Adjoint Matching
- TouchScale: 500 Hours of Human Vision and Touch for Visual-Tactile Learning
- Systematic Multi-Agent Vision-and-Language Navigation: Formulation, Benchmark, and Method
- EmbodiedSmith: Scaling Embodied Data through Recursive Self-Improvement Flywheel in Simulation
- WorldSonus: Bringing Sound to Worlds
- PhysEvo: Astra Can Act, Let It
- A GPU-Parallel Framework for Heterogeneous Multi-Task Reinforcement Learning
- EMHO: EMbodied Agent Harness Optimization via Experience Traces
- When to Switch: Reliable Action-Chunk Extension for Vision-Language-Action Models
- RealtimeWAM: One-Step Asynchronous World Action Models
- RobotUse: Allocating Computation, Context, and Decisions
- DiVeR: Decision-Critical Verifier Learning for VLA Test-Time Scaling
- PerturBot: Breaking Shortcut Priors in Vision-Language-Action Models with Perturbative Training
- Native Action-Prior Learning from Videos for World Action Models
- Language Models that Play Chess and Explain Their Moves
- PointWAM: 3D World Action Modeling for Dexterous Robotic Manipulation
- World Action Learning via Interaction-Centric Spectral Latent Guidance
- Fewer Tokens, Better Action: GPT-6 Astra Robot Agents with 14% Higher Success Rate but 65% Fewer Tokens
- InterEvolve: Test-Time Evolution of Reward Programs for Humanoid Loco-Manipulation
- World Action Modeling with Progressive Visual Planning
- Ego2Act: Evaluating Goal-Directed Manipulation in Egocentric Video Generation
- Spatial Memory Intelligence: Endowing World Models with Understanding-Driven Long-Term Memory
- UniWAM: Unified World-Action Model
- Kinematic MeanFlow: One-Step Action Generation Policy for Robotic Foundation Models
- HumanoidToolBench: Benchmarking Humanoid Tool Use from Selection to Mobile Execution
- WorldAuditBench: Interactive 3D World Auditing with Multimodal Agents
- RoboCoach: World Models as Active Coaches for Compositional Robot Skills
- OSWorld-Science: A Benchmark of Computer Use Agents for Learning and Using Scientific Software
- Beyond the Current Scene: Event-Referential Grasping with Active View Selection
- EgoTools: Towards Tool-Centric Reasoning in Real-World Egocentric Videos
- AutoDataBench: A Data-centric Testbed for Accelerating Auto Research
- Learning from Runtime Feedback through Failure-Bank Self-Evolution for Vision-Language-Action Models
- GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives
- LongLive-Plug: Once-for-All Distillation for Video Generation
- WorldLine: Action-Driven Visual Simulation for Robotic Manipulation
- Follow the Entities: A Corpus Map for Agentic Search
- EVO-WAM: Evolving World Action Models through Video-Action Verification
- Systematically Exploring the Capabilities of GPT-6 Astra as Embodied Policies
- TERRA: Terrain-Aware Reconstruction, Retargeting and Control for Musculoskeletal Locomotion
- MotorMind: Scaffolding General Vision Language Models for Zero-Shot Robot Manipulation
- Video2Skill: From Streaming Experience to Reusable Embodied Skills
- Taming VLAs under Robot Execution Errors: Self-Compensation and Stress Testing
- ProgressCompass: Embodied Progress Reward Models Are Lost Without the Right Context
- In-context Robot Learning Made Simple: A Democratized Recipe for Manipulation Tasks
- LabInstruct: Benchmarking Situated Instructional Video Generation for Lab Procedures
- Self-Evolving Coding Agents: From Digital Programs to Physical-World Intelligence
- Precise Editing and Flexible Referencing for Interactable Worlds
- WorldAttention: An Efficient Attention Architecture for Interactive Video World Models
- PanoVLN: Towards Effective Panoramic Vision-and-Language Navigation
- In-Context Learning for Robots: Methods and Applications
- LLMs are General Asynchronous Agents
- The Low-Rank Structure of VLA Reinforcement Learning
- Agent Priors-guided Policy Learning
- REALM: A Coarse-to-Fine Generative Framework for Embodied Reactive Listening
- Recursive Harness Distillation across Agents for Robot Manipulation
- RoboFoundry: System-as-Policy Evolution for Self-Learning Embodied Agents
- ExpVoyager: Direct Experience Navigation for Dynamic Agent Skill Synthesis
- InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data
- MemBodied: Recurrent Associative Memory for Vision-Language-Action Models
- EmbodiedMemory-Bench: Benchmarking Embodied Memory for Long-Horizon Embodied Tasks
- Beyond Future Prediction: Denoising as Generative Adaptation for Robot Control
- RoboFollow: Unveiling the Instruction Following Mirage in Embodied Agents
- CARE: Experience-Guided Atomic Corrective Execution for Vision-Language-Action Policies
- HappyWorld-Bench
- Tactile-JEPA: Topology-Aware Self-Supervised Representation Learning for Distributed Tactile Sensors
- Grounded Action Model: 3D Grounding as a Foundation for Robotics
- OmniEcho: Spatial Audio Understanding for Embodied Agents
- PackLab: A Comprehensive Framework for Developing, Training, and Evaluating MLLMs in Robotic Bin Packing
- Transferring the Intelligence of VLMs to Robotic Control
- HuRo: Robotizing Human Videos for Scalable VLA Pretraining
- MintAct: A Unified Visual Agent for Digital Environments
- From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention
- VABench: Measuring Embodied Spatial Intelligence through Visual Demonstrations, Active Perception, and Metric Control
- DeformSmith: Physics Harness-Guided Hierarchical Generation of Deformable Assets for Robot Manipulation
- ActionPiece: Rethinking Action Tokenization for Autoregressive Vision-Language-Action Models
- In-Context Robot Learning with VLM Agents
- Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control
- EventEgoHands++: Event-based Egocentric 3D Hand Mesh Reconstruction with Real Dataset
- Embedding Physics Priors in Robot Learning: A Survey
- PhysBrain 1.5: From Vision-Language Models to Physical Foundation Models
- CADWorld: Computer-Use Benchmark for Long-Horizon Computer-Aided Design
- Reasoning with Image Generation
- Breaking the Vision-Action Shortcut: Latent Interface Training for Generalizable Robotics Foundation Models
- Memory as Plans: World-Action Modeling with Memory-Grounded Planning
- The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement
- Show-Harness: Just a VLM Agent Can Play Robots
- ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMs
- SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem
- Occamy-1.0: Open Pareto-frontier 35B Intelligence for Co-work
- Exploring Collaboration between a language and a non-language agent
- RoboTok: An Internet-Scale Data Engine for Human Demonstration Retrieval and Dexterous Manipulation Learning
- ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action Models
- Evaluating Multimodal LLMs as Generalist Vision-Language-Action Agents for Drone Control: Commanding, Approaching, Tracking and Searching
- Lucida: Parse, Generate, and Place for Composable Real-to-Sim Scene Modeling
- LightNav-0: Eliciting VLM Spatial Intelligence for Generalist Embodied Navigation
- ZimaBlue: Evolving Generalizable World Action Models through Scalable Video Pre-training
- Lies We Can See: Joint Verbal and Non-Verbal Deception by VLM Agents in Embodied Social Interactions
- Matrix-Game 3.5: Enhancing Real-Time Streaming Interactive World Models with Patch Memory
- Scaling Automatic Research Agents via World Models
- UrbanGround: From Local Perception to Spatial Agency in a Real-Scale City
- Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models
- Zero-WAM: In-Context World-Action Modeling from Human Videos for Open-Ended Task Generalization
- Act with Intent: Distilling Behavior Intent for Vision-Language-Action Models
- DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation
- Spatial Memory Agent: Experience-Grounded Procedure Memory for Spatial Intelligence
- H2R-Bench: Benchmarking Human-to-Robot Manipulation Video Generation in World Models
- ComBodied Agents: a New Paradigm of Human-Centric Agentic AI
- Self-Evolving Embodied Agents via Skill-Harness Evolution
- SmartMage: Dynamic Modality Orchestration for 3D Scene Understanding
- RynnValue: Scaling Robotic Value Foundation Models with Temporal Distance
- CEAA: A Cognitive Embodied Agents Architecture for Interactive Computing Systems
- 360CityArena: A Realistic Virtual Urban Navigation Benchmark for Embodied Agents
- GST-Bench: Can VLMs Develop Global Spatial Awareness from Video?
- DyPES-VLA: Learning Shared Dynamics Priors and Embodiment-Specific Control for Cross-Embodiment Manipulation
- From Economic Agents to Agentic Economies: A Systems Blueprint for Economic World Models
- Uncertainty-Aware World Model for Aerial Image-Goal Navigation
- SkillZip: Contract-Preserving Graph Compression for Scalable Agent Skill Libraries
- When Memory Lies: An Empirical Study of Spatial Memory Staleness in VLM Agents
- World-to-Wrist: Task-Conditioned Future Wrist Modeling for Fine-Grained Robot Manipulation
- SkillJack: Persistent Skill Backdoors in Self-Evolving Agents
- UniWorld-Design: From Pixel Generation to Layer-Native Design
- Deferred Exposure of Future Trajectories for Verifiable Reasoning in Autonomous Driving VLMs
- Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data
- Weights or Skills? A Survey of Robot-Learning Techniques: from Action-Predicting Weights to Robots that Write their Own Skills
- MiniWorld: Democratizing the Training of Video World Models from Scratch
- GaussianSelector: Lightweight Human-Guided Object Selection in 3D Gaussian Splatting with Graph Optimization
- DreamTraj: Generating 6-DoF Object Trajectories by Reading Unrendered Video Diffusion Latents
- WCM: A World Critic Model for Vision-Language-Action Reinforcement Learning
- ACE-Data-0: Human-Centric Ambient Capture as Embodied Data Engine
- SpatialCLI: Learning to Reason With Spatial Tools, Then Without Them
- ODEWorld: A Continuous Predictive Architecture via Physical-Time Flow
- EMBL AI Librarian: Life-Sciences Knowledge Layer for AI Agents
- TurboVLA: Real-Time Vision-Language-Action Model at 32 Hz on an RTX 4090 with <1 GB VRAM
- HumanCLAW: Can Vision-Language Models Act Through a Body?
- HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone
- CodeNib: A Multi-View Data System for Serving Repository Context to Coding Agents
- N_0-TWAM: Scaling Tactile-Native World-Action Model for Contact-Rich Manipulation