BotPapers

Lihtsas keeles kokkuvõtted viimastest robootikauuringutest — inimkujulised robotid, kaasrobotid, tööstusautomaatika, droonid ja tehisintellekt.
85 paberid
OpenReLoc: Object-Level Camera Relocalization with Open-Vocabulary Understanding
Roboticstäna

OpenReLoc: Object-Level Camera Relocalization with Open-Vocabulary Understanding

OpenReLoc uses open-vocabulary object matching and LLM descriptions to robustly estimate camera pose from a single image, outperforming closed-vocabulary methods on real indoor scenes.

Zhaopeng Cui, Jiarui Hu, Jingbo Liu, Boming Zhao +6

Vision-Language Model Gives Warehouse Robots Context-Aware Semantic Maps
Warehousetäna

Vision-Language Model Gives Warehouse Robots Context-Aware Semantic Maps

Researchers built a pipeline combining SLAM, SAM, and a vision-language model to create contextual semantic maps that tell warehouse robots what objects are and whether they can be moved.

Marvin Rüdt, Hao Pang, Constantin Enke +2

InSight: How Robots Learn New Skills Without Human Help
AItäna

InSight: How Robots Learn New Skills Without Human Help

InSight enables robots to autonomously acquire new manipulation skills by identifying missing primitives through VLM reasoning and generating training data.

Maggie Wang, Lars Osterberg, Stephen Tian +3

Robots Learn Physical Reasoning from Scalable Human Hand Data
Roboticseile

Robots Learn Physical Reasoning from Scalable Human Hand Data

Jiaming Liu, Yinxi Wang, Chenyang Gu +15

LIBERO-Safety Benchmark Puts Vision-Language-Action Robots Through Physical and Semantic Safety Tests
Roboticseile

LIBERO-Safety Benchmark Puts Vision-Language-Action Robots Through Physical and Semantic Safety Tests

Rongxu Cui, Zongzheng Zhang, Jingrui Pang +11

AutoDex: Fully Automated Dexterous Grasping Data Collection at 75+ Trials Per Hour
AIeile

AutoDex: Fully Automated Dexterous Grasping Data Collection at 75+ Trials Per Hour

AutoDex is an end-to-end autonomous system that collects physically labeled dexterous-grasp trials without any human intervention—generating 3,593 real-world…

Mingi Choi, Gunhee Kim, Jisoo Kim +4

New FEM Framework Simulates Tactile Sensors with Unprecedented Force Accuracy
Roboticsüleeile

New FEM Framework Simulates Tactile Sensors with Unprecedented Force Accuracy

TaCauchy integrates FEM-based force computation into Isaac Sim, extracting full Cauchy stress tensors for vision-based tactile sensors.

Hengfei Zhao, Yifan Xie, Junhao Gong +6

ARC: Adaptive Robust Estimation Handles Outliers and Unknown Noise in Real-Time Localization
Roboticsüleeile

ARC: Adaptive Robust Estimation Handles Outliers and Unknown Noise in Real-Time Localization

Alexandre Hadji-Thomas, Andrew Stirling, James R. Forbes

Slow Brain, Fast Planner: Keeping Robots Safe When AI Vision Takes Its Time
Roboticsüleeile

Slow Brain, Fast Planner: Keeping Robots Safe When AI Vision Takes Its Time

Zhenghao "Mark'' Peng, Honglin He, Quanyi Li +2

GroundControl: Anticipating Navigation Failures in Vision-Language Agents Using Trajectory-Consistent Uncertainty
Robotics3 p eest

GroundControl: Anticipating Navigation Failures in Vision-Language Agents Using Trajectory-Consistent Uncertainty

GroundControl anticipates navigation failures in vision-language agents by detecting statistically significant deviations from expected goal-directed motion using trajectory-consistent uncertainty estimates.

Nastaran Darabi, Divake Kumar, Sina Tayebati +2

Fast Human Attention Prediction Enables Real-Time Fixation-Guided Drone Navigation
Robotics3 p eest

Fast Human Attention Prediction Enables Real-Time Fixation-Guided Drone Navigation

GazeLNN predicts human visual attention in real time to guide drone camera control, achieving state-of-the-art fixation prediction at 45 FPS on a Jetson Orin NX.

Fatma Youssef Mohammed, Grzegorz Malczyk, Kostas Alexis

New Motion Planning Algorithms Make Continuum Robots More Resilient
Robotics3 p eest

New Motion Planning Algorithms Make Continuum Robots More Resilient

Oxana Shamilyan, Ievgen Kabin, Zoya Dyka +2

New Attention Mechanism Treats Robot Poses as Group Elements, Boosts Performance
AI5 p eest

New Attention Mechanism Treats Robot Poses as Group Elements, Boosts Performance

Przemyslaw Musialski

Generating Task-Specific Robot Hands from Human Fingertip Demonstrations
Robotics5 p eest

Generating Task-Specific Robot Hands from Human Fingertip Demonstrations

We optimize robot hands so that human thumb-index fingertip motions are reproducible under inverse kinematics

Sha Yi, Nicklas Hansen, Xueqian Bai +3

MemoryWAM: Persistent Memory Makes Robot Action Models Faster and Smarter
Robotics5 p eest

MemoryWAM: Persistent Memory Makes Robot Action Models Faster and Smarter

Most robot action models forget what happened more than a few seconds ago, causing them to fail at tasks that require remembering past events. MemoryWAM intr…

Sizhe Yang, Juncheng Mu, Tianming Wei +8

New Algorithm UBP2 Uses Uncertainty to Learn Robot Rewards from Preferences
AI6 p eest

New Algorithm UBP2 Uses Uncertainty to Learn Robot Rewards from Preferences

Mohamed Nabail, Leo Cheng, Jingmin Wang +1

Do as I Do: Turning Everyday Human Videos Into Dexterous Robot Data
Robotics6 p eest

Do as I Do: Turning Everyday Human Videos Into Dexterous Robot Data

Bhawna Paliwal, Haritheja Etukuru, William Liang +3

Zero-Shot Long-Horizon Dexterous Manipulation with Multi-View 3D Grounded VLM Reasoning
Robotics6 p eest

Zero-Shot Long-Horizon Dexterous Manipulation with Multi-View 3D Grounded VLM Reasoning

A zero-shot framework for long-horizon dexterous manipulation using multi-view 3D grounded VLM reasoning and reusable atomic action primitives.

Jisoo Kim, Sangwon Baik, Taeksoo Kim +4

EBench: A New Benchmark Diagnoses Mobile Manipulation Robots' Core Capabilities
Robotics7 p eest

EBench: A New Benchmark Diagnoses Mobile Manipulation Robots' Core Capabilities

EBench is a 26-task benchmark that diagnoses mobile manipulation policies across five capability dimensions instead of a single success rate.

Ning Gao, Jinliang Zheng, Xing Gao +22

Robotics7 p eest

MOCHI Cleans Up Noisy Multi-Human Object Interaction Data

Jiye Lee, Yonghun Choi, Jungdam Won

🍪 Küpsiste eelistused

Kasutame küpsiseid, et mõõta toimivust. Privaatsuspoliitika