MEKAWIRE
Research

Papers

New robotics and physical-AI papers, with what each one means for the industry.

100 papers
TopicAllManipulation · 100VLA · 72Navigation · 52General robotics · 41Robot learning · 32Simulation · 24Perception · 23Humanoid · 22Locomotion · 16Safety · 13Aerial · 4World Action Models · 4World-action model · 3Aerial manipulation · 2Dexterous manipulation · 2Sim-to-Real Manipulation · 2Social Navigation · 2Vision-Language Navigation · 23D Assembly · 13DGS dataset · 15G Simulation · 1AMR Operations · 1Active vision manipulation · 1Adaptive MPC · 1Adaptive action horizons · 1Adaptive intervention · 1Aerial Manipulation · 1Aerial navigation · 1Agricultural Robotics · 1Articulation Estimation · 1Autonomous Driving · 1Autonomous Exploration · 1Battery-aware RL · 1Belief-space planning · 1Benchmark for adding wrist force sensing to VLA manipulation policies · 1Bimanual Manipulation · 1Byzantine localization · 1Canal inspection · 1Co-speech Humanoid · 1Compositional Manipulation · 1Construction navigation · 1Contact-Rich Manipulation · 1Contact-rich Manipulation · 1Continual reinforcement learning on top of robot foundation models · 1Continuum sensing · 1Controlled comparison of how to add world models to VLA robot policies · 1Cross-embodiment transfer of humanoid whole-body controllers · 1Curriculum RL · 1Dexterous Manipulation · 1Distributed Manipulation · 1Distribution Shift · 1Driving Simulation · 1Efficient VIO · 1Egocentric data · 1Egocentric video · 1Embodied Agents · 1Embodied policy · 1Empowerment MPC · 1Endoscopic navigation · 1Endovascular robotics · 1Evaluating robot world models against real executions · 1Event-camera VO · 1Exoskeletons · 1Failure recovery · 1Faster on-device inference for flow-matching vision-language-action models · 1Federated 3D Perception · 1Fisheye detection · 1Floorplan SLAM · 1Force-aware robot foundation model pretrained on 2,000 hours of human tactile-glove data · 1Geometry-aware manipulation · 1Grasping · 1Grounding · 1Humanoid Control · 1Humanoid RL · 1Humanoid Tool Use · 1Humanoid loco-manipulation · 1Humanoid safety · 1Humanoid whole-body control · 1Humanoid whole-body training data from robot-specific video generation · 1Imitation Learning · 1JEPA World Models · 1Lab automation · 1Laboratory Manipulation · 1Latent World Models · 1Legged Manipulation · 1LiDAR Security · 1Lifelong RL · 1MAPF · 1MAV guidance · 1Manipulation LfD · 1Manipulation Safety · 1Marine simulation · 1Memory-centric VLA · 1Miniature Locomotion · 1Mining Perception · 1Mobile manipulation · 1Model-based RL · 1Monocular SSC · 1Motion Planning · 1Multi-Agent RL · 1Multi-Robot Exploration · 1Multi-Robot Manipulation · 1Multi-UAV Exploration · 1Multi-UAV State Estimation · 1Multi-agent RL · 1Multi-agent motion planning · 1Multi-camera odometry · 1Multi-fidelity RL · 1Multi-robot manipulation · 1Multi-skill Manipulation · 1Multi-view perception · 1Multimodal simulation · 1Navigation & Mapping · 1Object Pose Estimation · 1Object Search · 1Panoramic 3D Detection · 1Panoramic segmentation · 1Pedestrian tracking · 1PersonNav · 1Planning · 1Policy Optimization · 1Preference Learning · 1Quadruped locomotion · 1Real-time VLA · 1Real-to-sim · 1Reinforcement learning for industrial assembly · 1Representation alignment · 1Robot-Use Agents · 1Runtime Verification · 1Safe Generative Planning · 1Safe RL-MPC · 1Safe VLA · 1Safety monitoring · 1Safety-critical control · 1Sim-to-real Navigation · 1Sim-to-real manipulation · 1Simulation geometry · 1Skill Discovery · 1Social navigation · 1Soft Pneumatic Actuators · 1Soft manipulation · 1Sonar SLAM · 1Streaming 3D reconstruction · 1Submovement decomposition · 1Surgical · 1Surgical robotics · 1Tactile Exploration · 1Tactile VLA · 1Temporal manipulation · 1Terrain traversability · 1Text-to-3D Policy · 1Tool Co-Design · 1Trajectory Correction · 1Traversability · 1Turning phone-recorded human demonstrations into per-robot post-training data by retargeting object motion · 1Two Unitree G1 humanoids playing table-tennis doubles with learned collision-risk filtering · 1UAV Navigation · 1UAV Swarms · 1UAV traffic monitoring · 1Underwater Robotics · 1Underwater active mapping · 1Underwater perception · 1Unified world-action model · 1VLA Conditioning · 1VLA Defense · 1VLA Robustness · 1VLA evaluation · 1VLA manipulation · 1VLA robustness · 1Video world models · 1Vision-language-action models without wrist cameras (learned foveation) · 1Visual Generalization · 1Visual Grounding · 1Visual Reward Models · 1Visuomotor Fusion · 1Visuotactile manipulation · 1World Models · 1World models · 1World-Action Models · 1compliant humanoid whole-body control · 1humanoid learning from human video · 1humanoid whole-body manipulation · 1real-world reinforcement learning for robot policies · 1robot foundation model training · 1robot learning from human video · 1vision-language-action models · 1world models for dexterous manipulation · 1
LevelAnyBeginnerIntermediateExpert
Filtered by technology vla · clear
arXiv
VLA RobustnessIntermediate

Is Success All You Need? Investigating the Impact of Input Perturbations on VLA Behaviour in Tabletop Manipulation Tasks

Sophie Higham, Riccardo Andrea Izzo, Matteo Matteucci, Alessandro Suglia

This arXiv preprint introduces a benchmark-agnostic evaluation framework that measures how successful VLA-controlled trajectories change under input perturbations, going beyond task success rate. Testing three state-of-the-art VLA models across four LIBERO suites and seven perturbation conditions, it finds that perturbed trajectories can still succeed yet diverge substantially in motion smoothness, efficiency, and gripper behaviour. This matters because teams relying only on success rate may miss behavioural degradation that affects safety and reliability.

VLA
Read the full paper (PDF)
arXiv
ManipulationExpert

DynaHarness: A Dynamic Physical Harness for Self-Evolving Robot Agents

Haoyuan Deng, Jiebin Liu, Tengxiao Zhang, Langning Yan, Hongye Cao, Ziwei Wang

DynaHarness is an arXiv preprint that adds a dynamic physical harness around a frozen vision-language-action policy: a slow Qwen3-VL-4B planner proposes capabilities, while a fast 2 Hz governor grounds, monitors, substitutes and recovers them. On LIBERO-Pro it reaches 75.2% success on 800 newly sampled states versus 17.5% for the frozen π0.5 policy, and its self-evolution gains transfer to new states. This gives robot-agent builders a retraining-free way to govern and improve existing policies from execution evidence.

VLA
Read the full paper (PDF)
arXiv
Manipulation SafetyIntermediate

Multi-Link Safety Filtering for VLA Policies Around Moving Hazards

Yatharth Agarwal, Vijay Raghunathan

This preprint introduces a training-free runtime safety filter that wraps a pretrained vision-language-action policy, shielding the gripper, wrist, and forearm with five ellipsoids and tracking moving hazards via sparse optical flow. Across six simulated hazard-motion conditions it cuts the collision rate from 65.62% to 27.27% and lifts safe task completion from 29.35% to 50.43%, while a small 16-episode physical arm test shows 3/16 shielded contacts versus 11/16 unshielded. The barrier runs in about 2 ms on CPU, showing onboard shielding is feasible without retraining.

VLA
Read the full paper (PDF)
arXiv
Safe VLAExpert

Learning from Runtime Feedback through Failure-Bank Self-Evolution for Vision-Language-Action Models

Mingyue Cui, Zheyuan Liu, Yihan Zhu, Zheyuan Zhang, Meng Jiang

This preprint presents FailBank, a framework that turns an observe-only safety teacher's runtime corrections into training records for vision-language-action policies. On VLA-Arena static-obstacle tasks, it raises task success by 8.5 and 6.9 percentage points over base policies for two VLA backbones while cutting policy-induced cumulative cost by 35.6% and 23.8%. The method matters because it lets robot policies learn a persistent balance between completing a task and avoiding unintended contact, instead of relying on temporary runtime shields.

VLA
Read the full paper (PDF)
arXiv
Memory-centric VLAIntermediate

Inline Memory Meets Reusable Skills: Memory-centric Framework for Vision-Language-Action Model

Zaijing Li, Rui Shao, Bing Hu, Haoyu Zhang, Dongmei Jiang, Liqiang Nie

This preprint introduces Optimus-R, a memory-centric VLA framework that turns robotic manipulation adaptation into explicit query-skill memory tuning instead of repeated parameter updates. With only 30% of training data it raises LIBERO average success to 88.6% (vs 72.1% for π0.5) and reaches 33.3% real-world success from 20 demos per task; in lifelong learning it reduces forgetting to 10.0% vs 15.0%. The approach matters for low-data robot adaptation and retaining prior skills.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

When Instructions Retrieve Trajectories: Diagnosing and Mitigating Generalization Failures in VLA Models

Hung-Jen Chen, Yu-Hsun Hou, Yan-Hong Chen, Yan-Fu Chen, Binghua Cai, Min Sun, Chun-Yi Lee

This preprint identifies a failure mode in vision-language-action (VLA) policies where instructions retrieve familiar trajectory families but do not bind language to scene-dependent action choices, so robots can execute a clean but wrong task. It introduces Equivariant Counterfactual Training (ECT), a paired data and loss scheme that supplies same-instruction demonstrations with different scenes and trains counterparts jointly. In controlled LIBERO-PRO tests, mean position-swap success rises from 36% to 59%, and on a real UR5e, unseen-position success rises from 3/40 to 35/40, showing the approach reduces counterfactual task errors.

VLA
Read the full paper (PDF)
arXiv
World-action modelIntermediate

Magic-W0: A Structured World-Action Foundation Model for Physical Intelligence

Xuhua Chen, Zhenhan Yin, Yuan Zhang, Lingfeng Zhang, He Zheng, Tong Mu, Shun Zuo, Dian Zhou, Di Wu, Xuan Zhou, Shaojie Wan, Rongtian Shen, Qiulong Xu, Yiduo Li, Yinglong Wang, Yanqian Wang, Kun Wang, Tao ZhangMagiclab Robotics Inc. (Magic-Lab Team)

Robot maker MagicLab has posted a preprint on Magic-W0, a 3.4-billion-parameter robot model that works out how the 3D scene will move and what it should look like afterwards while it is choosing an action. In the paper's October 3 revision it ranks first on the RoboDojo-Sim leaderboard with an average score of 36.75, only 0.48 points clear of the runner-up, and beats Physical Intelligence's π0.5 on five real-robot tasks by 94.6% to 91.8%.

Imitation LearningVLAWorld ModelsRobot Foundation Models
Physical IntelligenceMagicLabAgiBotNVIDIA
Read the full paper (PDF)
arXiv
GroundingExpert

GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives

Qize Yu, Lianrui Fan, Boyu Chen, Jiaqi Liang, Xini Ding, Yue Chen, Zetian Song, Yuran Wang, Yi Zou, Kaixuan Wang, Tianxing Chen, Wenxuan Song, Bohan Zhou, Mingleyang Li, Siqiao Huang, Yuqi Ye, Caigao Jiang, Wei Wei, Ruihai Wu, Hang Zhang

This preprint introduces GroundingPI, a 4B grounding foundation model that predicts object points and boxes as quantized token coordinates rather than using a general-purpose vision-language backbone. Across 34 grounding benchmarks it averages 73.68%, ahead of a larger GPT-6 Astra baseline at 71.54%, and as a visual backbone it improves downstream manipulation (RoboTwin 2.0, RoboCasa-GR1) and autonomous driving (nuScenes L2 0.296 m). The result matters because it argues grounding is a distinct perceptual layer that can make embodied foundation models more precise.

VLARL
Physical Intelligence
Read the full paper (PDF)
arXiv
Aerial ManipulationExpert

From Local Whole-Body VLA Behaviors to Scene-Scale Aerial Manipulation

Weixiang Guo, Rui Jin, Haotian Jin, Xinhang Xu, Ruiyang Liu, Haoran Zhao, Yi Wang, Weiqi Gai, Kun Cao, Lihua Xie

This preprint introduces a framework for scene-scale aerial manipulation on an articulated uncrewed aerial manipulator, combining synthetic whole-body VLA training, measured-progress-aligned trajectory realization, and scene-graph-guided mission composition. In simulation, local skills achieve 39/60 successes under oracle handoff, and full multi-site missions achieve 21/50 (42.0%), while a monolithic whole-task VLA baseline achieves 0/50. Physical trials validate representative tasks without platform demonstrations, reducing risk and data cost for aerial manipulation.

VLA
Read the full paper (PDF)
arXiv
Preference LearningExpert

PrefPI: Preference-Guided Steering into Out-of-Distribution Behaviors

Seungeun Rho, Wontaek Kim, Danfei Xu, Sehoon Ha

This preprint introduces PrefPI, an iterative framework that steers pretrained generative robot policies toward user-preferred behaviors using only relative preference labels on the robot's own rollouts. On a real-robot cube pick-and-place task, it lifted object transport height from 10.7 cm to 19.8 cm after 150 labeled trajectories and cut completion time from 19.8 s to 10.8 s. It matters because it shows preference feedback can push policies outside their original behavior distribution without new demonstrations or reward functions.

VLA
Read the full paper (PDF)
arXiv
VLAIntermediate

ChunkTrust: Adapting Execution Horizons for Robot Policies with Action-Expert Evidence

Fanding Huang, Jingyan Jiang, Shifeng Bao, Mingkang Pu, Shiwei Li, Jing Xu, Shijia Xu, Guanbo Huang, Chenghao Gu, Yuzhi Huang, Chenxin Li, Faisal Nadeem Khan, Huan Yang, Yan Wang, Cheng Chi, Zhi WangTsinghua University, Beijing Academy of Artificial Intelligence (BAAI), Renmin University of China, Shenzhen Technology University, Hefei University of Technology, Jiangnan University, Chongqing University, The Chinese University of Hong Kong

Robot AI models plan a short burst of movements at a time, and how many of those moves the robot carries out before it looks again is usually fixed by hand. This preprint from Tsinghua University, BAAI and partners adds a plug-in that picks that number while the robot works, raising success rates of existing Physical Intelligence and NVIDIA models in simulation and on a real two-arm robot without retraining them.

Imitation LearningVLARobot Foundation Models
Physical IntelligenceNVIDIA
Read the full paper (PDF)
arXiv
Real-time VLAExpert

Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation

Di Wu, Rongtian Shen, Ping Liu, Yan Shen, Zhenhan Yin, Shun Zuo, Xuhua Chen, He Zheng, Lingfeng Zhang, Jianglin Zhang, Tao Zhang

This preprint reports a two-stage non-uniform Flow Matching denoiser that cuts the π0.5 VLA's action-generation steps from 10 to 2 and model-inference time from 61.6 ms to 22.0 ms, a 2.8× speedup. The authors also built a distributed real-time VLA execution framework and benchmarked six action-scheduling methods on a 180-trial bimanual physical garment-folding task; Legato achieved 96.7% success as the best training-based method, while Temporal Smoothing led training-free methods at 76.7%. Combining the fast sampler with those methods gave 1.68–2.80× inference speedups at a 6.7–10.0 percentage-point success loss, highlighting a practical speed–quality tradeoff for real-time robot policies.

VLA
Read the full paper (PDF)
arXiv
Embodied policyExpert

EWAM: Emergent Depth-Wise Specialization in a Unified Embodied Model -- From Semantic Understanding through Visual Foresight to Action

Hao Wang, Jiajun Wen, Jingzhi Liu, Shuoshuo Xue, Zhiliang Chen, Min Lin, Yicheng Chang, Xiaoyu Guo, Yukang Zhuo, Zheng Chong, Yunshuang Nie, Jian Zhang, Weijia Liufu, Qingman Wu, Heming Xu, Bingchang Song, Dantong Wu, Zhiyuan Wang, Hang Xu, Jianhua Han

This preprint introduces EWAM, a unified embodied model whose action tokens asymmetrically attend to vision-language, video, and action experts. It reports 72.1% clean-to-random success on RoboTwin 2.0, a 92.9% in-domain average, 98.8% on LIBERO, and top real-robot results across three embodiments. The work matters because it shows a learned depth-wise handoff from semantic understanding to visual foresight to action can improve robustness and transfer, including from low-cost human egocentric data.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

MotorMind: Scaffolding General Vision Language Models for Zero-Shot Robot Manipulation

Bingxuan Li, Siqi Song, Yizhuo Wu, Jiarui Yao, Tong Zhang, Huan Zhang

Vision-language-action (VLA) models have advanced robotic manipulation, but their zero-shot generalization in new tasks and environments remains limited, and their reliance on specialized training keeps them from benefiting directly from rapidly advancing general-purpose vision-language models (VLMs). In parallel, recent agentic robotic systems leverage VLMs for high-level reasoning or coding agents for robot control, but often depend on extensive external models and tools, introducing additional complexity and cost.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

WayFinder: Hierarchical Visual-Language-Action for Zero-Shot Waypoint Generation and Low-Level Kinematic Control

Timothy K Johnsen, Marco Levorato

Visual Language Action (VLA) models offer unprecedented generalization for autonomous robots; however, their real-world deployment is frequently bottlenecked by unreliable execution and the prohibitive computational cost of fine-tuning for specific robot embodiments and tasks. To bridge this gap, we propose WayFinder, an end-to-end, closed-loop hierarchical VLA framework that circumvents the need for fine-tuning by decoupling high-level task reasoning from low-level kinematic control.

VLA
Read the full paper (PDF)
arXiv
VLA robustnessExpert

Taming VLAs under Robot Execution Errors: Self-Compensation and Stress Testing

Sohyun Lee, Yoonjae Baek, Jaesang Won, Jinnyeong Kim, Kang Hyunwoo, Seung-Hwan Baek, Ivan Laptev, Suha Kwak

This arXiv preprint introduces self-compensating VLA, a deployment-time adaptation method that updates a robot policy online using the residual between commanded and executed motion, plus RoboStress, a simulation benchmark for joint-level execution errors. On two physical Piper arms, it improves average task success by more than 30 percentage points per arm, and on RoboStress it beats base policies and training-time robustness methods. The work matters because mechanical wear, payload, and thermal drift cause real execution errors that standard VLA training does not address.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

Urgent Actions Go First: Urgency-Aware Denoising for Real-Time VLA Control

Zibo Wang, Haochen Han, Pengzhen Ren, Mingtong Dai, Fangming Liu

Diffusion and flow-matching Vision-Language-Action (VLA) policies generate action chunks through iterative denoising, incurring substantial inference latency that severely limits real-time robotic control. Existing acceleration methods treat an action chunk as a monolithic computational unit, ignoring a crucial physical reality of receding-horizon control: actions are generated jointly but consumed sequentially, resulting in inherently heterogeneous execution urgencies.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

Faster and Better? Benchmark Bugs and Design Limitations Distort the Evaluation of Vision-Language-Action Acceleration

Qiwei Chen, Kaijun Zhou, Nuohui Shi, Zhiyang Li, Yuxuan Feng, Jinyu Gu

Simulated manipulation benchmarks are the standard tool for evaluating vision-language-action (VLA) policies and the acceleration methods that reduce their inference latency for on-robot deployment. On these benchmarks, we observe that some training-free acceleration methods, which approximate the baseline policy's computation, achieve higher measured success rates than the baseline itself.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

RawVLA: Embodied Neural Image Signal Processor For Robotic Manipulation

Shuhong Liu, Heng Zhou, Lingfeng Qian, Yuhao Fang, Xianbao Hou, Qianyu Zhou, Lin Gu, Wei Sui, Jianfei Yang, Ziteng Cui

Vision-language-action (VLA) models typically operate on RGB images produced by a fixed camera image signal processor (ISP), leaving the imaging pipeline outside the learning and evaluation loop. We systematically examine the consequences of this overlooked design choice across five fundamental ISP dimensions: gain, sensor noise, chromatic response, tonal response, and bit depth.

VLA
Read the full paper (PDF)
arXiv
VLAExpert

Rho: A Foundation for Efficiently Adaptable VLA Models

Rho Team, Simran Bagaria, Daphne Chen, Dean Fortier, Jianlong Fu, Michael Harrison, Tess Hellebrekers, Neel Joshi, Andrey Kolobov, Dalton Moore, Galen Mullins, Michael Murray, Eduardo Salinas, Reuben Tan

General-purpose physical AI models must combine broad visual and linguistic capabilities with precise control across robot embodiments and efficient adaptation to downstream tasks. We introduce Rho, a family of open-weights VLA models for bimanual manipulation designed for data-light task adaptation on 3 embodiments representative of dual-arm robots across research labs and the industry -- YAM Box, UR AI Trainer, and FR3 Duo.

VLA
Read the full paper (PDF)

Topic pages

Manipulation · 100VLA · 72Navigation · 52General robotics · 41Robot learning · 32Simulation · 24Perception · 23Humanoid · 22Locomotion · 16Safety · 13Aerial · 4World Action Models · 5World-action model · 3Aerial manipulation · 3Dexterous manipulation · 3Sim-to-Real Manipulation · 3Social Navigation · 3

Papers come from arXiv robotics feeds. Where our summary is not written yet, you see the opening lines of the abstract.