MEKAWIRE
Research

Papers

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

72 papers
TopicAllManipulation · 65VLA · 43Navigation · 24Learning · 23Robotics · 22Perception · 20Humanoid · 15Safety · 9Simulation · 9Locomotion · 8World Action Models · 4Aerial · 2Aerial manipulation · 2Dexterous manipulation · 2Sim-to-Real Manipulation · 2Social Navigation · 2Vision-Language Navigation · 2World models · 2World-action model · 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 · 1Bimanual Manipulation · 1Byzantine localization · 1Canal inspection · 1Co-speech Humanoid · 1Compositional Manipulation · 1Construction navigation · 1Contact-Rich Manipulation · 1Contact-rich Manipulation · 1Continuum sensing · 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 · 1Event-camera VO · 1Exoskeletons · 1Failure recovery · 1Federated 3D Perception · 1Fisheye detection · 1Floorplan SLAM · 1Geometry-aware manipulation · 1Grasping · 1Grounding · 1Humanoid Control · 1Humanoid RL · 1Humanoid Tool Use · 1Humanoid control · 1Humanoid loco-manipulation · 1Humanoid safety · 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 · 1Policy Optimization · 1Preference Learning · 1Quadruped locomotion · 1Real-time VLA · 1Real-to-sim · 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 · 1Tactile Exploration · 1Tactile VLA · 1Temporal manipulation · 1Terrain traversability · 1Text-to-3D Policy · 1Tool Co-Design · 1Trajectory Correction · 1Traversability · 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 · 1Visual Generalization · 1Visual Grounding · 1Visual Reward Models · 1Visuomotor Fusion · 1Visuotactile manipulation · 1World Models · 1World-Action Models · 1
LevelAnyBeginnerIntermediateExpert
arXiv
MAV guidanceIntermediate

LiDARFlow: Real-Time Panel-Based MAV Guidance in Unknown Environments

João Machado, Zeynep Bilgin, Matthieu Verdoucq, Murat Bronz

This preprint describes a lightweight onboard guidance system for micro aerial vehicles in unknown environments, combining a guiding vector field for nominal motion with panel-method potential-flow obstacle avoidance. In indoor flight tests at 0.5 m/s, a quadrotor using only a Livox MID-360 LiDAR and Raspberry Pi 5 successfully completed waypoint-navigation and direction-following tasks while avoiding all obstacles. The result matters because it shows map-free sense-and-avoid can run in real time on constrained onboard hardware, reducing operator workload for cluttered or emergency missions.

LiDAR
Read the full paper (PDF)
arXiv
Agricultural RoboticsIntermediate

ALFRED: Requirement-driven development of an open-source mobile manipulator for long-term plant monitoring

Ciarán Miceal Johnson, Christopher Quail, Garry Ellard, Alistair McConnell, Steve Tonneau, Fernando Auat Cheein

Preprint. The authors built ALFRED, an open-source mobile manipulator made from commercial parts, and ran it for a year of monthly forest surveys (528 traversals) with no missed collection. Model-based comparisons across three builds show usable arm workspace rising from 34.0% to 66.1% of reachable poses and only the final build keeps the frame-mounted LiDAR clear forward and backward. This matters because it provides a reproducible, requirement-driven platform and release package for repeated close-range plant monitoring.

LiDAR
Read the full paper (PDF)
arXiv
VLA ConditioningIntermediate

Continuous Conditioning of VLAs with Augmenting EMG and Visual Task Descriptors

Edward W. Staley, Connor O. Pyles, Rahul Hingorani, Frank Camargo, Griffin Milsap, Jared Markowitz, Matthew S. Fifer, Michael Wolmetz

This preprint tunes action-expert vision-language-action models to accept non-language task cues: 8-channel EMG envelopes or visual segmentation annotations. On a real cube-picking task, EMG conditioning lifts cluttered success from 30% to 50% and cuts failure rate from 50% to 4%; on 16 simulated RoboCasa pick-and-place tasks, visual annotations lift cluttered success from 11% to 20% and cut failure from 47% to 25%. The findings suggest downstream VLA policies can usefully condition on continuous cues beyond fixed language prompts, especially in clutter.

VLA
Read the full paper (PDF)
arXiv
Articulation EstimationIntermediate

Query-Conditioned Articulation Estimation from a Single Image

Abdelrhman Werby, Fabio Scaparro, Kai O. Arra

QueryArt is a preprint model that estimates revolute or prismatic articulation parameters from a single calibrated RGB image, a 2D query point, and camera intrinsics, then uses one depth reading at the query to recover metric scale. It outperformed recent baselines on four held-out benchmarks and two out-of-distribution datasets, and achieved 70.2% success across 57 real-robot manipulation trials on unseen furniture. This matters because it decouples where to interact from how the part moves, making single-view explicit kinematics practical for mobile manipulators.

Read the full paper (PDF)
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
Multi-Robot ExplorationIntermediate

OpenSpace Lab Solution to the IROS 2026 Indoor Exploration Competition

Yuxuan Zhang, Dong Li, Zezhou Sun, Yuxuan Xu, Siyu Teng, Yuchen Li, Jianjian Yang, Long Chen

This preprint describes the OpenSpace Lab entry for the IROS 2026 Indoor Exploration Competition. It combines neural map-completion predictions, budget-aware goal selection, and homing constraints for single-robot exploration, with a coordinated multi-robot extension using shared maps and intent penalties. The system placed first in the single-robot public track at 61.04% coverage and third in both private tracks (39.53% single, 39.91% multi).

Read the full paper (PDF)
arXiv
Endoscopic navigationIntermediate

ColoACT: Multi-Cue Action Chunking for Smooth Autonomous Colon Navigation on a Self-Propelled Endoscopic Robot

Jian Hu, Shujing He, Leixin Chang, Zongze Li, Ding Huang, Chaoyang Shi, Chengzhi Hu

This is a preprint describing ColoACT, an imitation-learning navigation system for a compact self-propelled endoscopic robot. The policy fuses RGB, relative depth, and a gradient-based pseudo-elevation map with an Action Chunking Transformer and temporal ensembling. In ex-vivo porcine colons it reached 85.4% success on straight and 72.5% on curved segments, with smooth steering and success on 90-degree turns (70%) and double bends (60%), suggesting a path toward reducing physician burden in colonoscopy.

RL
Read the full paper (PDF)
arXiv
Video world modelsIntermediate

PhysicsLENS: Diagnosing Physical Property Blindness in Video Generation Models

Isaiah Milkey, Som Sagar, Aditya Taparia, Xinyuan Liu, Jiqing Wen, Ransalu Senanayake

This arXiv preprint introduces PhysicsLENS, a benchmark that tests whether robot video generation models follow text-specified physical properties that are not visible in the starting frame. Across four video generators and 439 clips, human raters found 34 of 47 plausible-looking clips ignored the stated latent property, a gap current visual-quality evaluations miss.

World Models
Read the full paper (PDF)
arXiv
PlanningIntermediate

Cross-entropy optimization with prioritized constraints

Francisco Roldan Sanchez, Pau de las Heras Molins, David Fridovich-Keil, Georgios Bakirtzis

This preprint introduces TierCEM, a variant of the cross-entropy method that enforces an ordered sequence of constraints during elite selection, removing the need for per-constraint penalty weights. In exact and learned world-model Push-T tests, reversing the priority order switched which constraint was violated, and putting task progress first let lower-priority constraints relax so progress continued. For robot planners facing conflicting safety and task requirements, this offers a more explicit and tunable-free way to specify which rules yield first.

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

SkeleWAM: Skeleton World-Action Modeling for Efficient Robotic Manipulation

Juyi Sheng, Hua Wang, Mengyuan Liu

This preprint introduces SkeleWAM, a 57.1M-parameter world-action model for robotic manipulation that represents scenes as sparse 3D skeletons of robot joints, object centers, and interaction points. On the full LIBERO-Plus benchmark it reaches 85.9% zero-shot success, beating the 2B-parameter Cosmos-Policy by 3.7 points, and 89% average success across five real-robot tasks versus Cosmos-Policy's 87%. The result suggests explicit robot–object geometry can provide robust, parameter-efficient control without video or dense latent world prediction.

Read the full paper (PDF)
arXiv
GraspingIntermediate

Continual Learning for 6-DoF Grasp Synthesis via Experience and Demonstrations

Giulio Schiavi, Andrei Cramariuc, Michael Pantic, Roland Siegwart

This preprint presents a continual-learning grasping system that adapts through memory rather than weight updates. After pretraining in simulation, it matches strong 6-DoF baselines offline and improves during deployment from grasp outcomes and optional user demonstrations; in real-robot trials it exceeds 90% success on five of six challenging object categories after at most 50 adaptation attempts. This matters because it gives deployed arms a low-cost way to improve on unfamiliar objects without gradient-based retraining.

Read the full paper (PDF)
arXiv
Multi-UAV ExplorationIntermediate

Centralized Multi-UAV Exploration and 3D Reconstruction Using Single-UAV Planners

João Félix Mendes, Meysam Basiri, Rodrigo Ventura

This is a preprint. The authors built a centralized multi-UAV exploration framework that lets existing single-UAV sampling-based planners work in a team by sharing a common TSDF map, filtering other robots from depth data, and checking inter-robot collisions. In simulation with three UAVs, separated start positions consistently beat joint starts; for example, KAEP reached 95% coverage of the school environment in 5.35 min vs 5.86 min from a joint start. The result matters because system-level integration, not new planner logic, can speed up exploration and improve robustness.

Read the full paper (PDF)
arXiv
Multi-camera odometryIntermediate

StreamRig: Exploiting Intra-Rig Geometry for Streaming Multi-Camera Odometry

Yufei Wei, Shuhao Ye, Qi Wang, Xin Zheng, Qing Huang, Rong Xiong, Yue Wang

This preprint presents StreamRig, a freeze-and-stream system that adapts a frozen multi-view 3D foundation model into causal streaming odometry for calibrated multi-camera rigs. Across NCLT, TartanGround, KITTI-360, and a self-collected humanoid-robot dataset, it achieves the lowest translation and rotation drift among non-oracle methods, and on an RTX 4090 it processes a five-camera rig in 26.2 ms with 2.6 GiB, faster and lighter than monocular CUT3R. The result matters because it shows how lightweight trainable modules can exploit known rig geometry and temporal state without retraining the foundation front-end, including sim-to-real transfer.

Read the full paper (PDF)
arXiv
Co-speech HumanoidIntermediate

ECHO-G: Embodied Co-speech Humanoid mOtion Generation

Yizhao Li, Pusen Gao, Ming Wang, Shaojie Shen, Shuo Yang, Hao Xu

ECHO-G is a preprint framework that generates full-body speech-synchronized motion for humanoid robots from audio and timed transcripts. It models motion directly in robot space with a Speech-Grounded Diffusion Transformer and reports a Fréchet Gesture Distance of 2.278 versus 4.976 for EMAGE+GMR, with 5.96 ms/frame inference time. This matters for making expressive, deployable co-speech gestures on humanoids without human-motion retargeting.

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
Social navigationIntermediate

STARS: From Spatiotemporal Dynamics to Social Representations in Human-Robot Interaction

Nathan Tsoi, Michael J. Munje, Tejas Oberoi, Rishab Maheshwari, Pengen Zheng, Tanush Chauhan, Peter Stone, Joydeep Biswas

STARS is a preprint method that learns compact, reusable social representations from short multi-agent navigation windows by treating agents as nodes in a spatiotemporal graph and training a self-supervised graph variational autoencoder. In downstream tests, its heterogeneous variant reaches 0.503 SNS macro F1 for pedestrian action classification versus about 0.30–0.33 for unstructured baselines, while remaining competitive on SEAN-T human perception tasks. This matters because it shows structured graph representations can offset scarce human-robot-interaction data, enabling data-efficient social navigation.

Read the full paper (PDF)
arXiv
3D AssemblyIntermediate

AssemblyWorld: Rethinking 3D Assembly with General-Purpose Agents

Jiahao Zhang, Yeying Fan, Moitreya Chatterjee, Suhas Lohit, Bernhard Egger, Tim K. Marks, Anoop Cherian, Stephen Gould

This preprint introduces AssemblyWorld, an interactive 3D simulation where general-purpose multimodal agents assemble rigid parts through rendered views and pose-editing tools. The strongest evaluated agent, GPT-6 Astra, reaches 80.9% part accuracy but only 59.4% complete-assembly success across 100 tasks spanning furniture, industrial parts, and broken-object reassembly. It matters because off-the-shelf agents can reason about assembly without task-specific fine-tuning, but precision and execution reliability still fall short of deployment needs.

Read the full paper (PDF)
arXiv
UAV traffic monitoringIntermediate

Prediction is Better than Detection: Traffic Congestion Control using Drones

Samira Hayat, Christian Raffelsberger

This preprint introduces a Mesa-based multi-agent simulation where UAVs patrol road junctions to detect and predict traffic jams and trigger adaptive signal control. Across sweeps of fleet size, traffic level, and network size, detection and prediction rates plateau once drone count roughly matches junction count; prediction-triggered signal changes cut jam duration by up to 8 ticks versus up to 5 ticks for detection-triggered changes, roughly doubling benefit. It suggests onboard prediction rather than larger fleets is the higher-value path for aerial congestion management.

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
Temporal manipulationIntermediate

PhasePlan: Ordered Future-Phase Planning for Robot Brain Models

Xiaoyu Yang, Yafei Zhang, Wensheng Li, Qing Zhan, Nan Wu

This preprint introduces PhasePlan, an ordered future-phase planning method for robot brain models. On a real-robot conveyor sorting task with a pretrained π0.5 backbone, PhasePlan reduces offline joint-action error by 22.5% relative to the original policy and reaches 92.5% per-position phase accuracy. By preserving where phase transitions occur within each 16-step action chunk, it may improve action timing in dynamic manipulation tasks.

Read the full paper (PDF)

Topic pages

Manipulation · 65VLA · 42Navigation · 24Robot learning · 23General robotics · 22Perception · 20Humanoid · 15Safety · 9Simulation · 9Locomotion · 7World Action Models · 5Aerial 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.