Embodied Agent Atlas

Embodied Agent Atlashttps://hongboduan.github.io/awesome-embodied-agents/具身 Agent 论文发现与版本动态；自动候选不等同于人工精选。zh-CNBiGym 2.0: Benchmarking Learned and Agent-Developed Policies for Humanoid Household Manipulationhttps://arxiv.org/abs/2610.07594v2arXiv 更新至 v2；已有编辑解读保留，待复核新版本。version-2610.07594-v2Mon, 12 Oct 2026 06:06:45 +0000What Should Embodied Agents Remember? Benchmarking and Evaluating Memory Modality for Embodied Agentshttps://arxiv.org/abs/2610.12561v1发现新论文，已进入待审核候选。discovered-2610.12561Mon, 12 Oct 2026 06:06:45 +0000Non-Prehensile Multi-Object Placement with Mobile Robots via Diffusion-Based Trajectory Planninghttps://arxiv.org/abs/2610.12684v1发现新论文，已进入待审核候选。discovered-2610.12684Mon, 12 Oct 2026 06:06:45 +0000REMM: Revisable Embodied Memory for Manipulationhttps://arxiv.org/abs/2610.12710v1发现新论文，已进入待审核候选。discovered-2610.12710Mon, 12 Oct 2026 06:06:45 +0000JoyInside-VLM-Agent: Product-Conditioned Multimodal Policy Learning for Embodied Agentshttps://arxiv.org/abs/2610.12957v1发现新论文，已进入待审核候选。discovered-2610.12957Mon, 12 Oct 2026 06:06:45 +0000EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agentshttps://arxiv.org/abs/2610.13021v1发现新论文，已进入待审核候选。discovered-2610.13021Mon, 12 Oct 2026 06:06:45 +0000Scale and Selection: What Makes Automatic Harness Evolution Work for Visual-Interface Robot Agentshttps://arxiv.org/abs/2609.39304v2arXiv 更新至 v2；已有编辑解读保留，待复核新版本。version-2609.39304-v2Mon, 12 Oct 2026 06:06:45 +0000MAGMA-Bench: Evaluating Long-Horizon Embodied Agents under Dynamic Tasks and Execution Failureshttps://arxiv.org/abs/2610.13534v1发现新论文，已进入待审核候选。discovered-2610.13534Mon, 12 Oct 2026 06:06:45 +0000AutoBio+: Benchmarking Dexterous Manipulation with Dynamic Experimental Phenomena in Realistic Laboratorieshttps://arxiv.org/abs/2610.13643v1发现新论文，已进入待审核候选。discovered-2610.13643Mon, 12 Oct 2026 06:06:45 +0000BT-Agent: Deploying Onboard Robot Agents with Behavior Trees and Multi-Fidelity Evaluationhttps://arxiv.org/abs/2610.13862v1发现新论文，已进入待审核候选。discovered-2610.13862Mon, 12 Oct 2026 06:06:45 +0000Physical RSI 1.0: Recursive Self-Harness for Scaling Embodied Skillshttps://mmlab.hk/research/PhysicalRSI让大模型依据执行证据修改技能、记忆与工具，再用新场景成对评测，选择下一轮继承的控制程序。paper-physicalrsiMon, 12 Oct 2026 00:00:00 +0000LoopROS: A Recursively Self-Evolving Robot Operating System for Agentic Manipulationhttps://loopmaster.ai/loopros/前台用历史观察与动作反馈驱动代码执行，后台把失败和修正整理为跨回合 Skills，让操作经验有机会被后续任务复用。paper-looprosMon, 12 Oct 2026 00:00:00 +0000DSG: Dynamic 3D Scene Graph Construction for Embodied Agents in Changing Indoor Environmentshttps://arxiv.org/abs/2609.00619v1发现新论文，已进入待审核候选。discovered-2609.00619Sun, 11 Oct 2026 08:53:23 +0000Towards Generalizable Visually Grounded Exploration of Household Deviceshttps://arxiv.org/abs/2609.00845v1发现新论文，已进入待审核候选。discovered-2609.00845Sun, 11 Oct 2026 08:53:23 +0000HitMem: Hierarchical Temporal 3D Memory with Multi-Modal Context-Aware Retrieval for Dynamic Environmentshttps://arxiv.org/abs/2609.00950v1发现新论文，已进入待审核候选。discovered-2609.00950Sun, 11 Oct 2026 08:53:23 +0000Retrieved but not ranked: surface-form bias in structural retrieval, from mathematics to agent trajectorieshttps://arxiv.org/abs/2609.01556v1发现新论文，已进入待审核候选。discovered-2609.01556Sun, 11 Oct 2026 08:53:23 +0000A computable representation of the physical laboratory enables verifiable workflowshttps://arxiv.org/abs/2609.03621v1发现新论文，已进入待审核候选。discovered-2609.03621Sun, 11 Oct 2026 08:53:23 +0000Open-Set 3D Scene Graphs for Field Robotics: An Outdoor Case Studyhttps://arxiv.org/abs/2609.04607v1发现新论文，已进入待审核候选。discovered-2609.04607Sun, 11 Oct 2026 08:53:23 +0000LookStep: Efficient Vision-Language Navigation with Linguistic Foresight and Event Driven Memoryhttps://arxiv.org/abs/2609.02350v2发现新论文，已进入待审核候选。discovered-2609.02350Sun, 11 Oct 2026 08:53:23 +0000RoboSPA: Can VLA Models Go Beyond Simple Scenes and Short-Horizon Tasks?https://arxiv.org/abs/2609.05324v1发现新论文，已进入待审核候选。discovered-2609.05324Sun, 11 Oct 2026 08:53:23 +0000One MLLM, One Call: Efficient Zero-Shot Vision-and-Language Navigation via Spatial-Aware Waypointshttps://arxiv.org/abs/2609.06476v1发现新论文，已进入待审核候选。discovered-2609.06476Sun, 11 Oct 2026 08:53:23 +0000Physico-Geospatial Grounded Scene Interpretation for Mobile Roboticshttps://arxiv.org/abs/2609.06629v1发现新论文，已进入待审核候选。discovered-2609.06629Sun, 11 Oct 2026 08:53:23 +0000ContextFlow: In-Context Flow Matching for Robot Manipulationhttps://arxiv.org/abs/2609.06852v1发现新论文，已进入待审核候选。discovered-2609.06852Sun, 11 Oct 2026 08:53:23 +0000Human-Aware Target Tracking and Navigation: Fusing Kinematic State Estimation with Structural Map Constraintshttps://arxiv.org/abs/2609.07091v1发现新论文，已进入待审核候选。discovered-2609.07091Sun, 11 Oct 2026 08:53:23 +0000D3ARC: Time-Critical Distributed Disaster Detection for Asynchronous Cooperative Multi-Robot Systemshttps://arxiv.org/abs/2609.07350v1发现新论文，已进入待审核候选。discovered-2609.07350Sun, 11 Oct 2026 08:53:23 +0000PhysReal: Learning Real-World Deformable Object Physics via Hybrid Constitutive Modelinghttps://arxiv.org/abs/2609.07532v1发现新论文，已进入待审核候选。discovered-2609.07532Sun, 11 Oct 2026 08:53:23 +0000Safe Task Planning with Long-Term Graph Memory for Embodied Agentshttps://arxiv.org/abs/2609.08444v1发现新论文，已进入待审核候选。discovered-2609.08444Sun, 11 Oct 2026 08:53:23 +0000Modality-Decoupled Federated Learning for Privacy-Preserving Embodied Intelligence in 6Ghttps://arxiv.org/abs/2609.09591v1发现新论文，已进入待审核候选。discovered-2609.09591Sun, 11 Oct 2026 08:53:23 +0000GTA-2: A Multi-VLM Framework for Synthesizing Robot Manipulation Skills via Grounded Task Axeshttps://arxiv.org/abs/2609.09808v1发现新论文，已进入待审核候选。discovered-2609.09808Sun, 11 Oct 2026 08:53:23 +0000Adaptive Shared Control with Online Bounded-Rational Human Behavior Estimationhttps://arxiv.org/abs/2609.10215v1发现新论文，已进入待审核候选。discovered-2609.10215Sun, 11 Oct 2026 08:53:23 +0000DUET-DINO: Simultaneous Cross-View World Modeling for Latent Planning in Robot Manipulationhttps://arxiv.org/abs/2609.10506v1发现新论文，已进入待审核候选。discovered-2609.10506Sun, 11 Oct 2026 08:53:23 +0000When Information is Worth the Risk: Behavioral Valuation for Hazardous Robotic Explorationhttps://arxiv.org/abs/2609.10726v1发现新论文，已进入待审核候选。discovered-2609.10726Sun, 11 Oct 2026 08:53:23 +0000When Validation Stops Learning: Auditing Update Admission for Continual Embodied Agentshttps://arxiv.org/abs/2609.10873v1发现新论文，已进入待审核候选。discovered-2609.10873Sun, 11 Oct 2026 08:53:23 +0000ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMshttps://arxiv.org/abs/2609.10895v1发现新论文，已进入待审核候选。discovered-2609.10895Sun, 11 Oct 2026 08:53:23 +0000LTLDiff: Finite Linear Temporal Logic-Guided Data Generation and Diffusion Policies for Multi-agent Robotic Manipulationhttps://arxiv.org/abs/2609.11043v1发现新论文，已进入待审核候选。discovered-2609.11043Sun, 11 Oct 2026 08:53:23 +0000Harness Robotic OS: A Unified Embodied-Agent Runtime for Closed-Loop Quadruped Inspectionhttps://arxiv.org/abs/2609.11225v1发现新论文，已进入待审核候选。discovered-2609.11225Sun, 11 Oct 2026 08:53:23 +0000Autonomy, Social Norms, and Alignment: Towards a Developmental Framework for Autonomous Artificial Agentshttps://arxiv.org/abs/2609.11660v1发现新论文，已进入待审核候选。discovered-2609.11660Sun, 11 Oct 2026 08:53:23 +0000Bridging Thought and Action: Taming Long-Horizon Instability in Open-Source LLM Agents with a MetaTool-Enhanced ROS Frameworkhttps://arxiv.org/abs/2609.13335v1发现新论文，已进入待审核候选。discovered-2609.13335Sun, 11 Oct 2026 08:53:23 +0000Embodied-BenchForge: A Closed-Loop Agentic Workflow for Embodied Benchmark Constructionhttps://arxiv.org/abs/2609.13082v1发现新论文，已进入待审核候选。discovered-2609.13082Sun, 11 Oct 2026 08:53:23 +0000STAGE: Diagnosing Semantic Transfer at Grounded Execution in Embodied Agentshttps://arxiv.org/abs/2609.13458v1发现新论文，已进入待审核候选。discovered-2609.13458Sun, 11 Oct 2026 08:53:23 +0000Task-Specified Active Metrological Inspection with Measurement-Steered VLA Manipulation and Deterministic Evidence Gatinghttps://arxiv.org/abs/2609.14219v1发现新论文，已进入待审核候选。discovered-2609.14219Sun, 11 Oct 2026 08:53:23 +0000Managing Action Preconditions in Neuro-Symbolic RL: Three Placement Strategies for Embodied Agentshttps://arxiv.org/abs/2609.16056v1发现新论文，已进入待审核候选。discovered-2609.16056Sun, 11 Oct 2026 08:53:23 +0000GLAM: Training a latent world model over global spatiotemporal memory for active exploration and navigationhttps://arxiv.org/abs/2609.14561v1发现新论文，已进入待审核候选。discovered-2609.14561Sun, 11 Oct 2026 08:53:23 +0000One Word, Different Action: A Real-Robot Benchmark for Language-Conditioned Embodied Reasoninghttps://arxiv.org/abs/2609.05260v2发现新论文，已进入待审核候选。discovered-2609.05260Sun, 11 Oct 2026 08:53:23 +0000EvoNav-Bench: Benchmarking Lifelong Navigation in Evolving Environmentshttps://arxiv.org/abs/2609.08292v2发现新论文，已进入待审核候选。discovered-2609.08292Sun, 11 Oct 2026 08:53:23 +0000Bench2Dex: Benchmarking Visuo-Tactile Bimanual Dexterous Manipulation Across Dexterous Handshttps://arxiv.org/abs/2609.15726v1发现新论文，已进入待审核候选。discovered-2609.15726Sun, 11 Oct 2026 08:53:23 +0000World Models for Embodied Intelligence: From Plausible to Controllable to Actionablehttps://arxiv.org/abs/2609.16697v1发现新论文，已进入待审核候选。discovered-2609.16697Sun, 11 Oct 2026 08:53:23 +0000The Robot Data Factoryhttps://arxiv.org/abs/2609.16705v1发现新论文，已进入待审核候选。discovered-2609.16705Sun, 11 Oct 2026 08:53:23 +0000Motion planning in high dimensional spaces hybridizing RRT and HAR via position-direction decouplinghttps://arxiv.org/abs/2609.16810v1发现新论文，已进入待审核候选。discovered-2609.16810Sun, 11 Oct 2026 08:53:23 +0000Exploring 2D backbone effects for indoor semantic occupancy predictionhttps://arxiv.org/abs/2609.17257v1发现新论文，已进入待审核候选。discovered-2609.17257Sun, 11 Oct 2026 08:53:23 +0000Finder: Agentic Closed-Loop Object Finding for Embodied Groundinghttps://arxiv.org/abs/2609.18058v1发现新论文，已进入待审核候选。discovered-2609.18058Sun, 11 Oct 2026 08:53:23 +0000AeroWeaver: An Embodied-Agent Harness for Weaving Aerial Skills into Distributed, Adaptive Swarm Executionhttps://arxiv.org/abs/2609.18520v1发现新论文，已进入待审核候选。discovered-2609.18520Sun, 11 Oct 2026 08:53:23 +0000TRACER: Adaptive Multi-Robot Social Navigation via Joint Human-Response Prediction and Interaction-Aware Replanninghttps://arxiv.org/abs/2609.18776v1发现新论文，已进入待审核候选。discovered-2609.18776Sun, 11 Oct 2026 08:53:23 +0000In-Context Robot Learning with VLM Agentshttps://arxiv.org/abs/2609.19138v1发现新论文，已进入待审核候选。discovered-2609.19138Sun, 11 Oct 2026 08:53:23 +0000GAVEL: Graph World Models for Verified and Efficient Long-Horizon LLM Task Planninghttps://arxiv.org/abs/2609.19315v1发现新论文，已进入待审核候选。discovered-2609.19315Sun, 11 Oct 2026 08:53:23 +0000Kinematics-Grounded Agentic AI for Robotic Additive Manufacturing Process Planninghttps://arxiv.org/abs/2609.19347v1发现新论文，已进入待审核候选。discovered-2609.19347Sun, 11 Oct 2026 08:53:23 +0000Navigate or Relocate? Planning Among Movable Obstacles in Unknown Environmentshttps://arxiv.org/abs/2609.19541v1发现新论文，已进入待审核候选。discovered-2609.19541Sun, 11 Oct 2026 08:53:23 +0000Towards High-DoF Dexterous Manipulation through VLA Post-Traininghttps://arxiv.org/abs/2609.19666v1发现新论文，已进入待审核候选。discovered-2609.19666Sun, 11 Oct 2026 08:53:23 +0000SPARROW: Survival-POMCP for Adaptive Robot Routing, Observation, and Waitinghttps://arxiv.org/abs/2609.21008v1发现新论文，已进入待审核候选。discovered-2609.21008Sun, 11 Oct 2026 08:53:23 +0000Navi-Agent: Unlocalized Monocular Navigation Agenthttps://arxiv.org/abs/2609.20388v2发现新论文，已进入待审核候选。discovered-2609.20388Sun, 11 Oct 2026 08:53:23 +0000RoboFind: Multi-Agent Personalized Object Search for People Who Are Blind or Have Low Visionhttps://arxiv.org/abs/2609.20330v2发现新论文，已进入待审核候选。discovered-2609.20330Sun, 11 Oct 2026 08:53:23 +0000A Robot Among People:From Social Imitation to the Social Becoming of Human Groupshttps://arxiv.org/abs/2609.12937v2发现新论文，已进入待审核候选。discovered-2609.12937Sun, 11 Oct 2026 08:53:23 +0000Transferring the Intelligence of VLMs to Robotic Controlhttps://arxiv.org/abs/2609.22966v1发现新论文，已进入待审核候选。discovered-2609.22966Sun, 11 Oct 2026 08:53:23 +0000Toward Physically Grounded JEPA World Models for Goal-Conditioned Robotic Planninghttps://arxiv.org/abs/2609.03565v2发现新论文，已进入待审核候选。discovered-2609.03565Sun, 11 Oct 2026 08:53:23 +0000AquaCap: A Training-Free Underwater Embodied Agent with Code-as-Policyhttps://arxiv.org/abs/2609.23133v1发现新论文，已进入待审核候选。discovered-2609.23133Sun, 11 Oct 2026 08:53:23 +0000When Successful Memories Mislead Embodied Agents:Memory Adaption For Task-Conditioned Executionhttps://arxiv.org/abs/2609.35808v1发现新论文，已进入待审核候选。discovered-2609.35808Sun, 11 Oct 2026 08:53:23 +0000CE$^4$L: Continual Ego, Exo, and Ego-Exo Learninghttps://arxiv.org/abs/2609.23492v1发现新论文，已进入待审核候选。discovered-2609.23492Sun, 11 Oct 2026 08:53:23 +0000Structured World-State Reasoning for Agentic Robotic Searchhttps://arxiv.org/abs/2609.23841v1发现新论文，已进入待审核候选。discovered-2609.23841Sun, 11 Oct 2026 08:53:23 +0000An Unexpected Robot Policy: Early Evaluations of GPT-6 Astra on RoboDojo and Beyondhttps://arxiv.org/abs/2609.24170v1发现新论文，已进入待审核候选。discovered-2609.24170Sun, 11 Oct 2026 08:53:23 +0000A Topological Representation with Object-Path Graphs for Open-Vocabulary Instance Navigationhttps://arxiv.org/abs/2609.24189v1发现新论文，已进入待审核候选。discovered-2609.24189Sun, 11 Oct 2026 08:53:23 +0000RoboFollow: Unveiling the Instruction Following Mirage in Embodied Agentshttps://arxiv.org/abs/2609.25636v1发现新论文，已进入待审核候选。discovered-2609.25636Sun, 11 Oct 2026 08:53:23 +0000Deploying Foundation Models for Embodied Navigationhttps://arxiv.org/abs/2609.25666v1发现新论文，已进入待审核候选。discovered-2609.25666Sun, 11 Oct 2026 08:53:23 +0000Listening and Mirroring: The Effects of Verbal Attunement and Behavioral Mimicry on Social and Empathic Perceptions of Embodied AI Agents in VRhttps://arxiv.org/abs/2609.27246v1发现新论文，已进入待审核候选。discovered-2609.27246Sun, 11 Oct 2026 08:53:23 +0000OmniEcho: Audio-Visual Spatial Understanding for Omni-Modal Embodied Agentshttps://arxiv.org/abs/2609.23407v2发现新论文，已进入待审核候选。discovered-2609.23407Sun, 11 Oct 2026 08:53:23 +0000RegenHarness: A Robot Agent Harness with Evidence-Gated Recursive Self-Improvementhttps://arxiv.org/abs/2609.27612v1发现新论文，已进入待审核候选。discovered-2609.27612Sun, 11 Oct 2026 08:53:23 +0000EmbodiedMemory-Bench: Benchmarking Embodied Memory for Long-Horizon Embodied Taskshttps://arxiv.org/abs/2609.28236v1发现新论文，已进入待审核候选。discovered-2609.28236Sun, 11 Oct 2026 08:53:23 +0000Controlling Collectives of AI Agents in Reasoning Space with Spatial Transformershttps://arxiv.org/abs/2609.28247v1发现新论文，已进入待审核候选。discovered-2609.28247Sun, 11 Oct 2026 08:53:23 +0000HarnessPAI: An Evolving Harness for Physical AIhttps://arxiv.org/abs/2609.29166v1发现新论文，已进入待审核候选。discovered-2609.29166Sun, 11 Oct 2026 08:53:23 +0000Robo-Harness K1: Harnessing Robot-Use Agents via Perception Augmentationhttps://arxiv.org/abs/2609.29389v1发现新论文，已进入待审核候选。discovered-2609.29389Sun, 11 Oct 2026 08:53:23 +0000RACaP: Agentic Reasoning, Acting, and Coding as Policies for Evolvable Robot Learninghttps://arxiv.org/abs/2609.29394v1发现新论文，已进入待审核候选。discovered-2609.29394Sun, 11 Oct 2026 08:53:23 +0000Robot Manipulation with GPT-6-Astra: Body Knowledge, Experience Reuse, Emergent Skills, and Sim2Real Transferhttps://arxiv.org/abs/2609.31770v1发现新论文，已进入待审核候选。discovered-2609.31770Sun, 11 Oct 2026 08:53:23 +0000World Action Agent: Harnessing VLMs for Robot Manipulation via World Action Rehearsalhttps://arxiv.org/abs/2609.29964v1发现新论文，已进入待审核候选。discovered-2609.29964Sun, 11 Oct 2026 08:53:23 +0000RAPID: Robot Agentic Programming from Demonstrationshttps://arxiv.org/abs/2609.30249v1发现新论文，已进入待审核候选。discovered-2609.30249Sun, 11 Oct 2026 08:53:23 +0000SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Modelshttps://arxiv.org/abs/2609.05533v2发现新论文，已进入待审核候选。discovered-2609.05533Sun, 11 Oct 2026 08:53:23 +0000Knowing When to Yield: Grounded Arbitration of User Corrections in Text-Based Embodied Agentshttps://arxiv.org/abs/2610.00282v1发现新论文，已进入待审核候选。discovered-2610.00282Sun, 11 Oct 2026 08:53:23 +0000Skip the Talk, Re-Focus on Vision: Latent Reasoning for Reasoning Segmentation in Multimodal Large Language Modelshttps://arxiv.org/abs/2609.30783v1发现新论文，已进入待审核候选。discovered-2609.30783Sun, 11 Oct 2026 08:53:23 +0000SciHorizon-eLab: An Agentic Protocol-to-Task Compiler for Scalable Benchmarking of Scientific Embodied Agentshttps://arxiv.org/abs/2609.30971v1发现新论文，已进入待审核候选。discovered-2609.30971Sun, 11 Oct 2026 08:53:23 +0000PUBG Ally: A Conversational Embodied Agent as an AI Teammatehttps://arxiv.org/abs/2609.29837v2发现新论文，已进入待审核候选。discovered-2609.29837Sun, 11 Oct 2026 08:53:23 +0000CognitiveReality: Robot-Agnostic Semantic Gaussian Mapping with an LLM Agent for Immersive Collaborative VR Teleoperationhttps://arxiv.org/abs/2609.31418v1发现新论文，已进入待审核候选。discovered-2609.31418Sun, 11 Oct 2026 08:53:23 +0000RAO-Nav: Probing Omni-Language Models for Zero-shot Semantic Audio-Visual Navigationhttps://arxiv.org/abs/2609.32224v1发现新论文，已进入待审核候选。discovered-2609.32224Sun, 11 Oct 2026 08:53:23 +0000MemTransfer: Benchmarking Memory Beyond Matched Experience in Embodied Decision-Makinghttps://arxiv.org/abs/2609.32313v1发现新论文，已进入待审核候选。discovered-2609.32313Sun, 11 Oct 2026 08:53:23 +0000RE-0: Verified Recursive Improvement of Embodied Code-as-Policy Agents through Local On-Policy Distillationhttps://arxiv.org/abs/2609.32416v1发现新论文，已进入待审核候选。discovered-2609.32416Sun, 11 Oct 2026 08:53:23 +0000In-Context Guidance: Learning Inter-Task Synergies via Numerical Foundational Models for Few-Shot Multitask Optimizationhttps://arxiv.org/abs/2609.25836v2发现新论文，已进入待审核候选。discovered-2609.25836Sun, 11 Oct 2026 08:53:23 +0000PlanGuard: A Guardrail for Multi-Step Plan Safety in Embodied Agentshttps://arxiv.org/abs/2609.32801v1发现新论文，已进入待审核候选。discovered-2609.32801Sun, 11 Oct 2026 08:53:23 +0000Nutri-ATLAS: Embodied Agent for Tabulated Lookup and Assistance for Smarter nutritionhttps://arxiv.org/abs/2609.32803v1发现新论文，已进入待审核候选。discovered-2609.32803Sun, 11 Oct 2026 08:53:23 +0000RoboFoundry: System-as-Policy Evolution for Self-Learning Embodied Agentshttps://arxiv.org/abs/2609.32862v1发现新论文，已进入待审核候选。discovered-2609.32862Sun, 11 Oct 2026 08:53:23 +0000What Must a World Model Distinguish for Planning?https://arxiv.org/abs/2609.33030v1发现新论文，已进入待审核候选。discovered-2609.33030Sun, 11 Oct 2026 08:53:23 +0000Beyond Tasks: A Vision for Reproducing an Animal-like Behavioral Substrate Using Modern Robot Learning Techniqueshttps://arxiv.org/abs/2609.33165v1发现新论文，已进入待审核候选。discovered-2609.33165Sun, 11 Oct 2026 08:53:23 +0000Scope-WM: Scoped Computation for Efficient Visual World Modelshttps://arxiv.org/abs/2609.33218v1发现新论文，已进入待审核候选。discovered-2609.33218Sun, 11 Oct 2026 08:53:23 +0000CodeActionBench: Evaluating Agentic Code-as-Policy for Embodied Manipulationhttps://arxiv.org/abs/2609.33807v1发现新论文，已进入待审核候选。discovered-2609.33807Sun, 11 Oct 2026 08:53:23 +0000