#VLA (100)

Unitree UnifoLM-OminiA-0.3: omni-modal interaction drives whole-body mobile manipulation

HumanoidTwitterUnitreeUnifoLMVLA

Patch Policy: small transformer outperforms large VLAs

Patch PolicyVLAtransformer机器人操作NYU

3D HAMSTER: 3D End-Effector Trajectory from RGB-D

3D HAMSTERVLA操作RGB-D3D轨迹

Astribot Lumo-2 — 20+ household tasks, one foundation model, real robot

TwitterVLAAstribotLumo-2人形

China's robots building other robots — G1 assembling motors on a real line

HumanoidTwitter人形UnitreeG1

TOPReward zero-training robot reward model

VLAReward ModelVLM零训练ZeroTraining

mimic-video Beats pi0 VLAs on LIBERO

VLAmimic-videoLIBERO样本效率sample efficiency

Flex-π: compute-flexible robot foundation model

机器人基础模型VLA世界模型3D感知语义感知

Alibaba DAMO RynnBrain-VLA: Humanoid Cooks Steak

人形机器人阿里达摩院RynnBrainVLA烹饪

TRON 2 VLA autonomous clothes folding

人形机器人VLALimXTRON2叠衣服

MiniCPM-Robot open-sourced — 1.5B general-purpose VLA model

TwitterVLAOpenBMBMiniCPM具身智能

LimX TRON 2 VLA active-vision sorting demo

人形VLA主动视觉抓取分拣

Galaxea G0.5: One Autoregressive VLA for Reasoning and Action

VLAGalaxeaG0.5视觉语言动作模型自回归

NVIDIA Releases DROID+, an Improved Open-Source Control Stack for DROID, with pi0.5 Running a Hello-World Task

NVIDIADROIDVLApi0.5开源控制栈

Sharpa Dual-Hand 63-DoF Robot with VLA Teleoperation

灵巧手DexterousHandVLA遥操作Sharpa

DexRobot open-sources SO101 Pick Cube SFT checkpoint and DM0.5 LoRA SFT workflow

DexRobotSO101DM0.5VLA人形机器人

TRON 2 VLA Autonomous Tracking and Candy Handoff

VLALimX DynamicsTRON 2追踪tracking

Playing Chess with SO-101 Arm: π0.5 + Stockfish

机械臂SO-101下棋chessπ0.5

τ0-VLA: Robot VLA that searches over subtask sequences at test time with a world model

VLA世界模型World Model测试时搜索机器人操作

Robot autonomously traverses a room while building a map

TwitterVLA感知导航建图

Zetta ζ: Closed-Loop Recovery Harness on Frozen VLA Policies

ZettaVLA失败恢复自我进化清华

Agentic Real2Sim: reconstructing a physics simulator from a single video

world modelAgentic Real2Simphysics simulationVLAresearch

TrapVLA: Precise Failure Injection Against VLA Manipulation Policies

VLAmanipulationadversarialrobustnessTwitter

Early access to NVIDIA Isaac GR00T N1.7 is here 🎉 — an open, commerci...

GR00THumanoidIsaacNVIDIATwitter

piR2: real-time flow policies on contact

流策略πR²实时控制接触操作

Sensor downsampling patterns in VLA research

VLA力觉传感降采样传感器融合控制

AgiBot X2 Ultra showcases its behavior foundation model

智元机器人AgiBot行为基础模型VLA自主导航

GEN-1 Embodied Foundation Model: GPT-3 Moment for Robotics

具身AI基础模型VLA操作GEN-1

Xiaomi Xiaomi-Robotics-1 VLA Scaling Laws

VLA缩放定律Xiaomi操作UMI

Google Gemini Robotics 2.0 Coming Soon

人形机器人humanoidGemini RoboticsGoogle双臂

ACT-2 Preview — first robotics model unifying generalization with reliability

TwitterVLAACT-2人形具身智能

Patch Policy: pretrained ViT beats OpenVLA-OFT with 0.7% params

VLAPatch PolicyViTOpenVLAmanipulation

Gemini Robotics 2 Enables Whole-Body Humanoid Control

VLAGemini RoboticsGoogle DeepMindApptronikApollo 2

RL2-VLA: adaptive RL compositional steering framework

VLARL²-VLA测试时缩放强化学习OOD

Google DeepMind Introduces Gemini Robotics 2

VLAGemini RoboticsGoogle DeepMindApptronikApollo 2

VLK: Vision-Language-Kinematics for Humanoid Loco-Manipulation

人形HumanoidVLAKinematics运动学

An incredible diversity of robots running in the real world, demonstrating real applications. Including end-to-end models

HumanoidTwitterVLAWAICWAIC 2026

Humanoid Robot Caregiving Is Not Far Off

Twitter人形机器人照护机器人VLA具身智能

Alibaba DAMO open-sources RynnBrain 1.1

人形机器人VLA阿里达摩院RynnBrain开源

pi0: generalist vision-language-action model

VLAπ0Physical Intelligence通用模型具身AI

Tau0-VLA: Robot Foundation Model for Long-Horizon Manipulation

VLA基础模型长时序操作Foundation ModelLong-Horizon Manipulation

World Action Models (WAMs): Unifying VLA and World Models

具身智能Embodied AIVLA世界模型World Models

DeVA: Decoupled Video-Action Model with physical guidance for robot policy learning

VLAPaperVision-Language-Action

Patch Policy: Efficient Embodied Control via Dense Visual Representations

Patch Policydense visual representationDINOv2WebSSLVLA

See like a Robot: Robot-Centric Pointmaps for Vision-Language-Action Models

VLApointmap机器人感知坐标系对齐cross-viewpoint

VLK: Learning Humanoid Loco-Manipulation from Synthetic Interactions in Reconstructed Scenes

humanoid人形机器人loco-manipulation移动操作synthetic data

TemporalFlow-VLA: Learning Physically Grounded Execution History for Long-Horizon Robot Manipulation

VLA视觉-语言-动作时间流Temporal Flow长时程操作

FM-VLA: Force-based Memory for Vision-Language-Action Models in Contact-Rich Manipulation

VLA力觉感知触觉记忆增强接触丰富操作

GigaBrain-0.7: Scaling Embodied Foundation Models to Emergent Capabilities with a Three-System Architecture

VLA具身智能世界模型GigaBrainFlow Matching

Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence

具身智能Embodied AIVLA智能体框架Agentic Harness

Helix: Large-Scale Neural Networks for Whole-Body Humanoid Control

Figure AIHelix人形VLA端到端

Emergent Compositional Skills in Mixture-of-Experts VLAs

VLA混合专家组合泛化涌现能力Paper

AeroAct: Action-Centered World-Action Models for Language-Conditioned Quadrotor Flight

无人机UAV四旋翼quadrotor世界模型

Closing the Loop in Humanoid VLA: Persistent 3D Object Tokens for Verifiable Loco-Manipulation

人形机器人VLA位操作3D物体令牌闭环执行

FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference

VLAflow matching流式解码异步推理低延迟

Addressing the Orchestration Gap in Generalist Robots via Physical Agency

TwitterVLA机器人操作编排具身AI

ABot-AgentOS: A General Robotic Agent OS with Lifelong Multi-modal Memory

机器人智能体AgentOS智能体操作系统具身智能多模态记忆

A Few Words Go a Long Way: Language Guided Robot Policy Synthesis

VLAPaperVision-Language-Action

τ: Learning Touch-Augmented Vision-Language-Action Models from Future Visual Supervision

VLAPaperVision-Language-Action

HyWorldVLA: A Vision-Language-Action Model with Hybrid World Modeling for Autonomous Driving

VLA世界模型自动驾驶混合建模Paper

HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone

操作策略UMI数据采集VLA模仿学习

$π\mathbf{R}^2$: Reactive Real-time Flow Policies

机械臂操作扩散策略实时控制VLA闭环控制

RL$^2$-VLA: Adaptive RL Latent Compositional Steering with Test-Time Scaling for Vision-Language-Action Models

VLA强化学习离线RL推理时引导测试时扩展

X-NavDP: Generalizing Navigation Diffusion Policy to Novel Behavior and Embodiments with Group Q-score Reweighted Matching

导航扩散策略强化学习跨本体NavDP

Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data

VLA机器人操作数据合成第一人称视频动作重定向

PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloud

Physical AI推理引擎VLA世界模型边缘部署

CosFly-VLA: A Spatially Aware Vision-Language-Action Model for UAV Tracking

无人机UAVVLA目标跟踪tracking

StageWAM: Joint-Embedding Stage Prediction for World-Action Models in Robot Manipulation

机器人操作世界模型JEPAWAMVLA

G0.5: One Autoregressive Stream for Robot Reasoning and Action

VLA具身智能自回归思维链人形机器人

HAF: Adapting Generalist VLAs to Humanoid Whole-Body Loco-manipulation via Hierarchical Action Flow and Spectral Latent RL

VLA人形机器人humanoid全身控制whole-body

EATR-Stereo: Embodiment-Aware Token Routing of Paired Stereo Evidence for Humanoid Vision-Language-Action Control

VLAstereo visiontoken routinghumanoid robotembodiment-aware

DECOWAM: Decoupled Whole-Body World-Action Model for Legged Mobile Manipulation

世界模型VLA移动操作四足机器人全身控制

Beyond Imitation: Self-Improving Robot Policies via Off-Policy Q-Planning

机器人操控强化学习行为克隆Q-Learning自改进

RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control

VLAGoogle具身智能零样本语言指令

GR00T N1: NVIDIA Humanoid Robot Foundation Model

NVIDIA人形基础模型IsaacVLA

AtlasVLA: Persistent World-Ego State Modeling for Vision-Language-Action Models

VLA世界-自我状态持久记忆扩散Transformer腕部相机

ReferTrack: Referring Then Tracking for Embodied Visual Tracking

VLA目标跟踪指代具身智能Paper

Closing the Lab-to-Store Gap: A Data-Efficient Post-Training and Experience-Driven Learning VLA Framework for Retail Humanoids

VLA人形机器人零售数据高效后训练

MulRobBench: A Decision-Level Benchmark for Safe and Security-Policy-Compliant Multimodal UAV Agents

VLAPaperVision-Language-Action

FutureRTC: Real-Time Robot Execution with Anticipatory-Conditioned Action Chunking

VLAPaperVision-Language-Action

Reasoning as a Double-Edged Sword: Architecture and Cross-Stage Robustness in Vision-Language-Action Models

VLAPaperVision-Language-Action

JoyNexus: Service-Oriented Multi-Tenant Post-Training for VLA Models

VLA后训练post-training多租户multi-tenant

The Developer Inflection Point: Leju's Open-Source LeTools Post-Training Toolchain Compresses Embodied-AI Development from Weeks to Days

具身智能LeToolssim-to-real行为树VLA

Perceptron Open-Sources Isaac 0.5: First Open Model at the Frontier of Video Understanding, Embodied Reasoning and Robot Control — 210x Less Teleop via Video Scaling

具身基础模型embodied foundation model扩展定律scaling lawVLA

LingBot-VLA 2.0 Reproduction Radar: Weights Open-Sourced, but 60K-Hour Pre-Training Cannot Be Equivalently Reproduced

LingBot-VLAVLA复现雷达RoboTwin具身智能

Introducing S1: In-Context Learning for Robotics — One Video Demonstration Is All It Takes

Skild AIS1上下文学习in-context learning机器人基础模型

Helix 02: Figure's First Unified Whole-Body Loco-Manipulation VLA

Figure AIHelix人形机器人VLA行走操作

Helix Learns to Fold Laundry: First Humanoid with Multi-Fingered Autonomous Folding

Figure AIHelix灵巧操作叠衣服VLA

Two Hands Are Harder Than One Brain: Interpreting the VLA Bimanual Manipulation Survey

双臂操作VLAbimanualflow matching动作分块

Zetta ζ: Closed-Loop Self-Evolution for a Frozen VLA Policy

具身智能embodied AIVLA自我进化self-evolution

moz1

轮式Wheeled千寻智能Spirit AIMoz1

ollobot-olloni

陪伴机器人赛博宠物VLA端侧智能情感交互

#VLA (100)

Robot-centric pointmaps align VLA perception with the robot frame

Symbiosis Robotics puts a Unitree G1 behind a go-kart wheel

Flex-π: Multi-Stream World-Action Model

X-Humanoid unveils TG-VLA for human-like intelligence

T800 robot fights in China — robots being blind is the big issue

Ego2Robot: egocentric videos as robot training data

CoorDex: Humanoids Walk and Manipulate Simultaneously

Humanoid robots enter garment factories

Unitree UnifoLM-OminiA-0.3: omni-modal interaction drives whole-body mobile manipulation

Patch Policy: small transformer outperforms large VLAs

3D HAMSTER: 3D End-Effector Trajectory from RGB-D

Astribot Lumo-2 — 20+ household tasks, one foundation model, real robot

China's robots building other robots — G1 assembling motors on a real line

TOPReward zero-training robot reward model

mimic-video Beats pi0 VLAs on LIBERO

Flex-π: compute-flexible robot foundation model

Alibaba DAMO RynnBrain-VLA: Humanoid Cooks Steak

TRON 2 VLA autonomous clothes folding

MiniCPM-Robot open-sourced — 1.5B general-purpose VLA model

LimX TRON 2 VLA active-vision sorting demo

Galaxea G0.5: One Autoregressive VLA for Reasoning and Action

NVIDIA Releases DROID+, an Improved Open-Source Control Stack for DROID, with pi0.5 Running a Hello-World Task

Sharpa Dual-Hand 63-DoF Robot with VLA Teleoperation

DexRobot open-sources SO101 Pick Cube SFT checkpoint and DM0.5 LoRA SFT workflow

TRON 2 VLA Autonomous Tracking and Candy Handoff

Playing Chess with SO-101 Arm: π0.5 + Stockfish

τ0-VLA: Robot VLA that searches over subtask sequences at test time with a world model

Robot autonomously traverses a room while building a map

Zetta ζ: Closed-Loop Recovery Harness on Frozen VLA Policies

Agentic Real2Sim: reconstructing a physics simulator from a single video

TrapVLA: Precise Failure Injection Against VLA Manipulation Policies

Early access to NVIDIA Isaac GR00T N1.7 is here 🎉 — an open, commerci...

piR2: real-time flow policies on contact

Sensor downsampling patterns in VLA research

AgiBot X2 Ultra showcases its behavior foundation model

GEN-1 Embodied Foundation Model: GPT-3 Moment for Robotics

Xiaomi Xiaomi-Robotics-1 VLA Scaling Laws

Google Gemini Robotics 2.0 Coming Soon

ACT-2 Preview — first robotics model unifying generalization with reliability

Patch Policy: pretrained ViT beats OpenVLA-OFT with 0.7% params

Gemini Robotics 2 Enables Whole-Body Humanoid Control

RL2-VLA: adaptive RL compositional steering framework

Google DeepMind Introduces Gemini Robotics 2

VLK: Vision-Language-Kinematics for Humanoid Loco-Manipulation

An incredible diversity of robots running in the real world, demonstrating real applications. Including end-to-end models

Humanoid Robot Caregiving Is Not Far Off

Alibaba DAMO open-sources RynnBrain 1.1

pi0: generalist vision-language-action model

Tau0-VLA: Robot Foundation Model for Long-Horizon Manipulation

World Action Models (WAMs): Unifying VLA and World Models

DeVA: Decoupled Video-Action Model with physical guidance for robot policy learning

Patch Policy: Efficient Embodied Control via Dense Visual Representations

See like a Robot: Robot-Centric Pointmaps for Vision-Language-Action Models

VLK: Learning Humanoid Loco-Manipulation from Synthetic Interactions in Reconstructed Scenes

TemporalFlow-VLA: Learning Physically Grounded Execution History for Long-Horizon Robot Manipulation

FM-VLA: Force-based Memory for Vision-Language-Action Models in Contact-Rich Manipulation

GigaBrain-0.7: Scaling Embodied Foundation Models to Emergent Capabilities with a Three-System Architecture

Zetta ζ: An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence

Helix: Large-Scale Neural Networks for Whole-Body Humanoid Control

Emergent Compositional Skills in Mixture-of-Experts VLAs

AeroAct: Action-Centered World-Action Models for Language-Conditioned Quadrotor Flight

Closing the Loop in Humanoid VLA: Persistent 3D Object Tokens for Verifiable Loco-Manipulation

FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference

Addressing the Orchestration Gap in Generalist Robots via Physical Agency

ABot-AgentOS: A General Robotic Agent OS with Lifelong Multi-modal Memory

A Few Words Go a Long Way: Language Guided Robot Policy Synthesis

τ: Learning Touch-Augmented Vision-Language-Action Models from Future Visual Supervision

HyWorldVLA: A Vision-Language-Action Model with Hybrid World Modeling for Autonomous Driving

HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone

$π\mathbf{R}^2$: Reactive Real-time Flow Policies

RL$^2$-VLA: Adaptive RL Latent Compositional Steering with Test-Time Scaling for Vision-Language-Action Models

X-NavDP: Generalizing Navigation Diffusion Policy to Novel Behavior and Embodiments with Group Q-score Reweighted Matching

Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data

PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloud

CosFly-VLA: A Spatially Aware Vision-Language-Action Model for UAV Tracking

StageWAM: Joint-Embedding Stage Prediction for World-Action Models in Robot Manipulation

G0.5: One Autoregressive Stream for Robot Reasoning and Action

HAF: Adapting Generalist VLAs to Humanoid Whole-Body Loco-manipulation via Hierarchical Action Flow and Spectral Latent RL

EATR-Stereo: Embodiment-Aware Token Routing of Paired Stereo Evidence for Humanoid Vision-Language-Action Control