IROS 2025 Papers — Page 3
IEEE/RSJ International Conference on Intelligent Robots and Systems · 1984 papers
Analysis and Experiment of a Pneumatic Linear Actuator Actuated by both Positive and Negative Pressures
Weijian Ni, Jianhua Zhang
Physics Related
🎯 What it does: Establish an analytical model for a dual-pressure pneumatic linear actuator, validate the model experimentally, and investigate the effects of negative pressure and mixed pressure (negative pressure + positive pressure) on the actuator's output force and displacement.
Analysis and Mitigation of Inconsistencies in Blockchain-Enabled Robot Swarms
Giada Simionato, Marco Dorigo
Robotic Intelligence
🎯 What it does: Analyze the inconsistencies caused by network partitioning in blockchain robot swarms, propose a decentralized partition detection and response mechanism, and validate its effectiveness in a robot swarm simulator.
Analysis of Compliant Torso Vibration on Passive Quadruped Walkers
Yuxuan Xiang, Isao T. Tokuda
Robotic IntelligencePhysics Related
🎯 What it does: Linearization and frequency domain analysis of passive quadruped walking robots to study trunk vibration characteristics (natural frequency, amplitude) and their impact on walking performance, verified through numerical simulation of resonance phenomena.
Analytic Conditions for Differentiable Collision Detection in Trajectory Optimization
A. Jaitly, Yuki Shirai
Optimization
🎯 What it does: Proposes an efficient method for enforcing non-penetration constraints during optimization configuration, directly applicable to collision-aware trajectory optimization.
Analyzing Human Perceptions of a MEDEVAC Robot in a Simulated Evacuation Scenario
Tyson Jordan, Adam Goodie
Safty and PrivacyRobotic Intelligence
🎯 What it does: Studied human perception of the MEDEVAC robot in a simulated evacuation scenario using a mixed-factor design, comparing the effects of three modes (autonomous slow, fast, and remote operation) on emotions, perceived safety, and social compatibility.
Annotation-Free Curb Detection Leveraging Altitude Difference Image
Fulong Ma, Jun Ma
Object DetectionAutonomous DrivingConvolutional Neural NetworkImagePoint Cloud
🎯 What it does: Proposes an unannotated road retaining wall detection method using height difference maps (ADI), combined with an automatic retaining wall annotator (ACA) and a post-processing module to achieve end-to-end detection.
Anomaly Detection in Human-Robot Interaction Using Multimodal Models Constructed from In-the-Wild Interactions
Shota Mochizuki, Ryuichiro Higashinaka
ClassificationAnomaly DetectionVideo
🎯 What it does: Created a manually annotated dataset based on real-world interaction videos, trained a classification model for anomaly detection; subsequently, presented detection results as alerts to operators within a parallel dialogue framework, verifying that alerts assist operators in intervention.
Anomaly Knowledge Learning for Patch-Agnostic Defense against Adversarial Patches
Hongmin Mu, Zhengcai Cao
Anomaly DetectionTransformerSupervised Fine-TuningImage
🎯 What it does: Proposed a patch-agnostic adversarial patch defense method based on anomaly knowledge learning;
Antagonistic Physical-Virtual Framework for the Development of Soft Actuators
Diogo Fonseca, P. Neto
Robotic Intelligence
🎯 What it does: A framework for developing and integrating soft actuator models that includes high-fidelity digital twins and mechanical integration platforms is proposed, enabling testing and validation in both real and virtual environments.
Anti-Slip AI-Driven Model-Free Control with Global Exponential Stability in Skid-Steering Robots
Mehdi Heydari Shahna, J. Mattila
Robotic Intelligence
🎯 What it does: Designed a model-free control system for a sliding-unstable sliding-steering heavy robot, utilizing neural networks to achieve robust compensation for wheel slippage and ensure global exponential stability.
Any-shape Real-time Replanning via Swept Volume SDF
Yijin Wang, Fei Gao
Robotic Intelligence
🎯 What it does: Proposes a real-time 10Hz replanning method that combines the concept of swept volumes with B-spline trajectory representation, enabling continuous collision avoidance for arbitrary rigid bodies in complex unstructured environments.
AnyBipe: An Automated End-to-End Framework for Training and Deploying Bipedal Robots Powered by Large Language Models
Yifei Yao, Jun-Guo Lu
Robotic IntelligenceTransformerLarge Language ModelReinforcement Learning
🎯 What it does: Proposes Anybipe, a fully automated end-to-end framework for training and deploying bipedal robots, leveraging large language models to generate reward functions.
AnyTSR: Any-Scale Thermal Super-Resolution for UAV
Mengyuan Li, Liangliang Yao
Super ResolutionImage
🎯 What it does: Proposed a UAV-specific arbitrary-scale thermal imaging super-resolution method called AnyTSR, which achieves image reconstruction at arbitrary scales using a single model.
Anyview: General Indoor 3D Object Detection with Variable Frames
Zhenyu Wu, Haibin Yan
Object DetectionPoint Cloud
🎯 What it does: Proposed a framework called AnyView for indoor 3D object detection that can handle RGB-D inputs with variable frame numbers.
Apple detection method based on fusion of infrared thermal image and visible-light image
Yuanchen Li, Kedian Wang
Object DetectionConvolutional Neural NetworkMultimodalityAgriculture Related
🎯 What it does: Proposes an apple detection method based on the fusion of infrared thermal imaging and visible light images to address the problems of lighting variations and fruit occlusion in open orchard environments.
Applicability Analysis for Optical Cooperative Localization
Yixian Li, Zhonghu Hao
Simultaneous Localization and MappingImage
🎯 What it does: A general feasibility analysis method for optical cooperative localization is proposed, addressing the problem of whether optical beacons can be continuously captured by visual sensors throughout the positioning process.
Application of LLM Guided Reinforcement Learning in Formation Control with Collision Avoidance
Chenhao Yao, Chi Zhu
Robotic IntelligenceTransformerLarge Language ModelReinforcement Learning
🎯 What it does: Proposed a reinforcement learning framework guided by large language models for formation control and collision avoidance in multi-agent systems, accelerating the convergence of the policy network through dynamically generated and online-adjusted reward functions.
Application of soft constraints on mirror position to improve robustness of optical target positioning in shallow water
Xiangjie Zhang, Yong Cao
Pose EstimationOptimizationImagePhysics Related
🎯 What it does: Proposed and implemented a scheme to enhance optical target positioning in shallow water environments using mirror soft constraints. By establishing and optimizing the pose relationship between ArUco markers and their mirror images during the mapping phase, the locatable space was expanded, and the number of usable markers was doubled.
Approximate Convex Decomposition-based Whole-Body Trajectory Optimization for Robots in Dense Environments
Linao Gong, Ning Hao
OptimizationRobotic Intelligence
🎯 What it does: Propose a full-body trajectory optimization method based on approximate convex decomposition (ACD), which uses combinations of multiple convex polyhedra to approximate non-convex robots and obstacles, and provides a differentiable convex polyhedra collision evaluation technique;
ARC-Calib: Autonomous Markerless Camera-to-Robot Calibration via Exploratory Robot Motions
Podshara Chanrungmaneekul, Kaiyu Hang
OptimizationRobotic Intelligence
🎯 What it does: Proposes a model-based unmarked camera-robot calibration framework called ARC-Calib, achieving fully autonomous and generalizable camera-to-robot calibration.
Arc-Length-Based Warping for Robot Skill Synthesis from Multiple Demonstrations
Giovanni Braglia, L. Biagiotti
Robotic IntelligenceSequential
🎯 What it does: Proposed and implemented a trajectory alignment algorithm called Spatial Sampling (SS) based on arc length, achieving time-agnostic trajectory synchronization to extract more accurate and robust robot skill representations.
ARC: Robots Adaptive Risk-aware Robust Control via Distributional Reinforcement Learning
Junlong Wu, Houde Liu
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed an adaptive risk-aware control (ARC) strategy based on distributed reinforcement learning, and verified its robustness on various robots
Arena-Bench 2.0: A Comprehensive Benchmark of Social Navigation Approaches in Collaborative Environments
V. Shcherbyna, Eva Wiese
Robotic IntelligenceBenchmark
🎯 What it does: Proposed Arena-Bench 2.0, a fully integrated social navigation benchmark platform within the Arena framework, developed a ROS2-based plugin architecture, integrated multiple learning-based and model-based navigation methods, generated diverse social scenarios through a web interface, and subsequently evaluated all planners using navigation and social metrics.
ArtGS: 3D Gaussian Splatting for Interactive Visual-Physical Modeling and Manipulation of Articulated Objects
Qiaojun Yu, Cewu Lu
Robotic IntelligenceVision Language ModelGaussian SplattingPoint Cloud
🎯 What it does: Proposes the ArtGS framework, combining 3D Gaussian Splatting with visual-physical modeling to achieve visualization and interactive manipulation of joint objects.
Articulation-Gen: 3D Part Segmentation and Articulated Object Generation
Zhuo-Wen Xu, Yang Liu
SegmentationGenerationTransformerLarge Language ModelMesh
🎯 What it does: Propose the Articulation-Gen framework for generating physically feasible multi-jointed 3D objects and construct a dataset of 10.6K articulated assets.
Artificial Muscle: A Sarcomere-inspired Magnetic Approach
Ning Li, Xingang Zhao
Robotic Intelligence
🎯 What it does: Developed a magnetic biomimetic flexible actuator (BMAA) based on the structure of muscle spindle muscle fibers, utilizing soft magnetic composite materials in a hierarchical structure to simulate myofilament arrangement, achieving contraction and relaxation through an external magnetic field.
AS2FM: Enabling Statistical Model Checking of ROS 2 Systems for Robust Autonomy
Christian Henkel, M. Morelli
Computational EfficiencyRobotic Intelligence
🎯 What it does: Propose a tool called AS2FM that applies statistical model checking (SMC) to ROS 2 systems, enabling design-time verification of properties in autonomous robot systems.
Assembly Sequence Planning Considering Robotic Motion Costs and Multi-Operation Constraints
Haruto Nagai, Kensuke Harada
OptimizationRobotic Intelligence
🎯 What it does: Propose an assembly sequence planning method that considers multiple operational constraints and robot motion cost.
Assessing Trust and Cognitive Load in Teleoperated Robotic Systems Across Different Information Conditions
Juan Jose Garcia Cardenas, Adriana Tapus
Robotic IntelligenceBiomedical Data
🎯 What it does: Evaluate trust and cognitive load in teleoperation robot systems under different information conditions
AssistantX: An LLM-Powered Proactive Assistant in Collaborative Human-Populated Environments
Nan Sun, Huaping Liu
Robotic IntelligenceLarge Language ModelAgentic AIText
🎯 What it does: Proposed a LLM-driven proactive assistant robot named AssistantX, employing a multi-agent framework (perception, planning, decision-making, reflection) to achieve autonomous high-precision operations, validated on 210 real-world tasks;
Assistive Guidance System Based on Online Path Structure Recognition for the Visually Impaired
Jae-Yeong Lee, Beomsu Seo
RecognitionAutonomous DrivingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Propose an auxiliary robot navigation system based on LiDAR for online path structure recognition, enabling navigation according to directional instructions from visually impaired individuals without relying on pre-existing maps.
ASV-Aided AUV Navigation: A Field Study on Nonlinear Estimation for Localization of Low-Cost, Scalable Systems
Raymond Turrisi, Michael Benjamin
Autonomous DrivingSimultaneous Localization and MappingTime Series
🎯 What it does: The study uses multiple autonomous surface vehicles (ASVs) as communication/navigation assistance devices (CNA) to enhance the navigation and state estimation capabilities of autonomous underwater vehicles (AUVs).
Asynchronous Harmony-based Decentralized Auctions Method for Scalable UAV Swarm
Runfeng Chen, Zehao Xiong
Optimization
🎯 What it does: Proposes an Asynchronous Harmony-based Distributed Auction (AHDA) method for task and time scheduling in UAV swarms, reducing communication load and scheduling time through neighboring communication, rapid conflict resolution, and information diffusion constraints.
Asynchronous Rectification-Based Fast Local Imaging and Estimation Scheme for High-Speed Rotating States Observation of MM
Zhiyong Sun, Bo Song
Object TrackingRobotic IntelligenceImagePhysics Related
🎯 What it does: Developed a tracking-based optimal local imaging and estimation scheme for observing the high-speed rotation state of magnetic microrobots, combined with an asynchronous calibration method to achieve precise measurement of the rotation state.
ATARS: An Aerial Traffic Atomic Activity Recognition and Temporal Segmentation Dataset
Zihao Chen, Yan-Tsung Peng
RecognitionSegmentationVideoBenchmark
🎯 What it does: Proposed and constructed the ATARS dataset, defined the multi-label temporal atomic activity recognition task, and conducted experimental evaluation on the performance of existing SOTA models in atomic activity recognition and temporal segmentation.
Attention-Based Higher-Order Reasoning for Implicit Coordination of Multi-Robot Systems
Jonathan Reasoner, Nicola Bezzo
Robotic IntelligenceTransformerLarge Language Model
🎯 What it does: Propose a theory-of-mind-based implicit coordination method for multi-robot systems, leveraging higher-order reasoning, epistemology, and active inference to coordinate robot behaviors, while reducing computational overhead through LLM attention selection.
AugInsert: Learning Robust Visual-Force Policies via Data Augmentation for Object Assembly Tasks
Ryan Diaz, K. Desingh
Robotic IntelligenceTransformerMultimodality
🎯 What it does: Proposed and implemented a factor-based evaluation framework using the multi-sensor Perceiver IO to learn the plug-in hole task, and investigated robustness to out-of-distribution conditions; evaluated using a simulated environment with simple data augmentation on multi-sensor data, and validated through real-world experiments.
Augmented Bridge Spinal Fixation: A New Concept for Addressing Pedicle Screw Pullout via a Steerable Drilling Robot and Flexible Pedicle Screws
Y. Kulkarni, F. Alambeigi
Robotic IntelligenceBiomedical DataComputed Tomography
🎯 What it does: Drilling a J-shaped channel on a vertebral prosthesis using a steerable drilling robot and flexible screw nails, and injecting bone cement to form a bridging fixation;
Augmenting robotic disassembly skill: combining compliance control strategy with reinforcement learning for twist-pulling disassembly *
Y. Zang, Yongjing Wang
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose a robotic torsion-pulling disassembly strategy that integrates compliance control with reinforcement learning (RL), which can adapt to unknown connection geometries and system errors, thereby enhancing disassembly performance.
Auto-calibration of Camera Intrinsics and Extrinsics using Lidar and Motion
Stepán Obdrzálek, J. Matas
Autonomous DrivingOptical FlowPoint Cloud
🎯 What it does: A method is proposed that automatically calibrates any camera model without using calibration targets, utilizing LiDAR and motion features to calibrate the camera's intrinsic and extrinsic parameters.
Automated 3D-GS Registration and Fusion via Skeleton Alignment and Gaussian-Adaptive Features
Shiyang Liu, Mengyin Fu
OptimizationConvolutional Neural NetworkGaussian SplattingPoint Cloud
🎯 What it does: Propose an automated 3D Gaussian Splatting submap alignment and fusion method that eliminates the need for manual selection of reference submaps and enhances registration accuracy and fusion quality.
Automated Behaviour-Driven Acceptance Testing of Robotic Systems
Minh Nguyen, Nico Hochgeschwender
Robotic Intelligence
🎯 What it does: This paper extends Behavior-Driven Development (BDD) to robot systems, using domain-specific modeling and knowledge graphs to define and verify acceptance test criteria, and generate executable test models, particularly focusing on pick-and-place applications.
Automated Dual-Micropipette Coordination Microinjection for Batch Zebrafish Larvae Based on Pose Estimation
Can Wang, Mingzhu Sun
Pose EstimationRobotic IntelligenceImage
🎯 What it does: Proposed an automated dual micro-injector collaborative injection system, which randomly arranges zebrafish larvae and uses pose estimation algorithms to locate the yolk for injection.
Automated Manipulation of Magnetic Microswarms for Temporal Logic Cargo Delivery Tasks in Complex Environments
Naifu Zhang, Rongrong Ji
Robotic IntelligencePhysics Related
🎯 What it does: Proposed an automated control strategy for magnetic microrobot swarms to execute finite linear temporal logic tasks in complex environments, achieving automatic selection and delivery of multiple particles in static and dynamic environments.
Automated Repositioning from Supine to Lateral with a Humanoid Robot Based on Body Modeling
Misa Matsumura, Etsuko Kobayashi
Pose EstimationDepth EstimationOptimizationRobotic IntelligenceImageMultimodalityPoint Cloud
🎯 What it does: Automated implementation of patient position change from supine to lateral decubitus
Automated UAV-based Wind Turbine Blade Inspection: Blade Stop Angle Estimation and Blade Detail Prioritized Exposure Adjustment
Yichuan Shi, Ximin Lyu
ImagePhysics Related
🎯 What it does: Developed a UAV detection platform and proposed a blade stop angle estimation method based on the Fermat point and a detail-oriented exposure adjustment method.
Automatic Alignment of the Micropipette for Efficient and Precise Cell Micromanipulation
Shuai Cui, Wei Tech Ang
Robotic IntelligenceImage
🎯 What it does: Developed a vision-guided robot control strategy to achieve automatic alignment and precise positioning of micropipettes during cell micromanipulation.
Automatic Generation of Aerobatic Flight in Complex Environments via Diffusion Models
Yuhang Zhong, Fei Gao
GenerationDiffusion model
🎯 What it does: Proposes a framework based on diffusion models to automatically generate high-difficulty aerial stunt flight trajectories. The method constructs complex maneuvers by decomposing them into short-frame stunt primitives, and employs conditional generation using historical trajectory dynamic priors, target waypoints, and optional action constraints. During inference, obstacle avoidance is achieved through classifier guidance and batch sampling, while post-processing uses spatiotemporal trajectory optimization to ensure dynamic feasibility.
Automatic Machinability Evaluation and Recommendation for Reconfigurable Manufacturing Systems*
Yuzhe Wang, T. Ng
Recommendation System
🎯 What it does: Automatically evaluate the machinability of product features using fuzzy logic and provide reconfiguration recommendation schemes for CNC machine tools.
Automatic MILP Model Construction for Multi-Robot Task Allocation and Scheduling Based on Large Language Models
Mingming Peng, Liang Gao
OptimizationAI Code AssistantTransformerLarge Language ModelText
🎯 What it does: Proposes a knowledge-enhanced mixed-integer linear programming (MILP) automated modeling framework that generates executable code from natural language descriptions using local large language models and domain knowledge bases.
Automatic Real-to-Sim-to-Real System through Iterative Interactions for Robust Robot Manipulation Policy Learning with Unseen Objects
Minjae Kang, Songhwai Oh
Robotic IntelligenceReinforcement Learning
🎯 What it does: This paper proposes a fully automatic Real-to-Sim-to-Real framework called ARIC, which utilizes a robot to continuously change object poses through reinforcement learning to observe real objects. Subsequently, it learns task-specific manipulation strategies in a simulated environment and directly applies them to the real world without human intervention.
AutoMisty: A Multi-Agent LLM Framework for Automated Code Generation in the Misty Social Robot
Xiao Wang, Venugopal Govindaraju
Robotic IntelligenceAI Code AssistantTransformerLarge Language ModelAgentic AIText
🎯 What it does: Automatically convert natural language instructions into executable Misty robot code, and achieve subtask generation and integration through a multi-agent framework.
Autonomous 3D Moving Target Encirclement and Interception with Range Measurement
Fen Liu, Rong Su
Robotic Intelligence
🎯 What it does: Proposes a strategy for 3D target circling and interception using autonomous drones, capable of tracking and attacking non-cooperative enemy drones in non-line-of-sight, GPS-denied, and radar-jammed environments.
Autonomous Adjustment of Tracking Position in Dynamic Environments for Human-Following Robots Using Deep Reinforcement Learning
Cong-Thanh Vu, Yen-Chen Liu
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose a reinforcement learning-based framework that enables robots to autonomously adjust their following positions in dynamic environments according to workspace constraints.
Autonomous Dissection in Robotic Cholecystectomy
Ki Hwan Oh, P. Giulianotti
SegmentationPose EstimationRobotic IntelligenceVideoBiomedical Data
🎯 What it does: Proposed a vision-based autonomous robotic cholecystectomy architecture integrating real-time segmentation, keypoint detection, left-arm grasping and stretching of the gallbladder, and right-arm dissection.
Autonomous Hiking Trail Navigation via Semantic Segmentation and Geometric Analysis
Camndon Reed, Yu Gu
SegmentationRobotic IntelligenceConvolutional Neural NetworkImagePoint Cloud
🎯 What it does: Proposed a traversability analysis module that combines camera semantic segmentation with LiDAR geometric information, achieving an autonomous pedestrian path navigation method that can follow trails while deviating from them when necessary to avoid obstacles or find safe shortcuts.
Autonomous Human-Robot Interaction via Operator Imitation
S. Christen, Moritz Bacher
Robotic IntelligenceTransformerDiffusion model
🎯 What it does: A robot capable of autonomously performing human-robot interaction was built by training on data that imitates operators.
Autonomous Obstacle Avoidance for a Snake Robot with Surface Pressure Sensing
Yongjun Sun, Hong Liu
Robotic Intelligence
🎯 What it does: Developed a 16-jointed serpentine robot with full-body thin-film pressure sensing; investigated four intelligent obstacle-avoidance actions based on pressure perception, and experimentally validated their feasibility in different scenarios.
Autonomous Subtask Generation for Indoor Search and Rescue Mission via Large-Language-Model and Behavior-Tree Integration
Junfeng Shi, Hui Zhang
Robotic IntelligenceLarge Language Model
🎯 What it does: AutoExpand proposes a high-level framework that tightly couples a large language model (LLM) with behavior trees to automatically generate responsive, context-aware subtasks for indoor search and rescue missions;
Autonomous Surface Selection For Manipulator-Based UV Disinfection In Hospitals Using Foundation Models
Xueyan Oh, U-Xuan Tan
Robotic IntelligenceTransformerVision Language Model
🎯 What it does: Utilize foundation models to simplify surface selection for operator UV disinfection, reducing manual intervention and eliminating the need for model training.
Autonomous Suturing Method for Robot-Assisted Minimally Invasive Surgery
Mei Feng, Xiuquan Lu
OptimizationRobotic Intelligence
🎯 What it does: Proposed an autonomous suturing method for robot-assisted minimally invasive surgery, which was deployed on their own surgical robot.
Autonomous UAV Control for Maritime Applications using Deep Reinforcement Learning-based Image Optimisation
Yuanqing Yang, Yansha Deng
OptimizationRobotic IntelligenceReinforcement LearningImage
🎯 What it does: Proposes a system for autonomous drone control in maritime environments, specifically for patrolling detected suspicious vessels and capturing informative images.
Autonomous Vehicle Controllers From End-to-End Differentiable Simulation
Asen Nachkov, L. V. Gool
Autonomous DrivingRecurrent Neural NetworkReinforcement Learning
🎯 What it does: An end-to-end autonomous vehicle controller was trained using a differentiable simulator and analytical policy gradient (APG) method.
AutoSpatial: Visual-Language Reasoning for Social Robot Navigation through Efficient Spatial Reasoning Learning
Yangzhe Kong, Xuesu Xiao
Robotic IntelligenceVision Language ModelMultimodalityChain-of-Thought
🎯 What it does: Propose the AutoSpatial method, combining minimum manual supervision with large-scale VQA auto-annotation, adopting hierarchical two-stage VQA training to enhance spatial reasoning capabilities of vision-language models in social robot navigation.
Autotuning Bipedal Locomotion MPC with GRFM-Net for Efficient Sim-to-Real Transfer
Qianzhong Chen, Quan Nguyen
Hyperparameter SearchRobotic Intelligence
🎯 What it does: Using DiffTune automatic parameter tuning combined with GRFM-Net to improve the simulation-to-physical transfer of MPC in bipedal robots.
AUV-WTN: AUV Water Tunnel Navigation Framework with Acoustic Perturbations and Narrow Space Constraints
Haotian Zheng, Jinyu Fu
SegmentationAutonomous DrivingConvolutional Neural NetworkImage
🎯 What it does: Proposes the AUV-WTN framework, combining improved RM R-CNN image segmentation and dynamic trajectory homotopy method (DTHM) to achieve autonomous navigation in waterways.
AVIP: Acoustic-Visual-Inertial-Pressure Fusion-based Underwater Localization System with Multi-Centric Calibration
Yuanbo Xue, Bing Wang
OptimizationMultimodality
🎯 What it does: Proposes AVIP, an underwater localization method that integrates acoustic, visual, inertial, and pressure sensors, achieving precise localization through multi-modal registration and multi-center calibration.
AVP Scene Graph: Hierarchical Visual Language Mapping and Navigation for Autonomous Valet Parking
Xiangru Mu, Tong Qin
Autonomous DrivingGraph Neural NetworkTransformerVision Language ModelImageTextMultimodalityRetrieval-Augmented Generation
🎯 What it does: Proposes the AVP Scene Graph (AVP-SG) framework, which combines vision-language models (VLM) and OCR to perform hierarchical semantic mapping on parking lot images, and utilizes LLM-enhanced graph retrieval to achieve open-vocabulary navigation goal localization.
Awakening Facial Emotional Expressions in Human-Robot
Yongtong Zhu, Jianwei Zhang
Robotic IntelligenceImage
🎯 What it does: Designed a highly biomimetic robot face equipped with physical-electronic animation facial units, and developed an end-to-end learning framework based on KAN and attention mechanisms
B4P: Simultaneous Grasp and Motion Planning for Object Placement via Parallelized Bidirectional Forests and Path Repair
Benjamin H. Leebron, Kaiyu Hang
Robotic Intelligence
🎯 What it does: Proposes the B4P framework, which can simultaneously plan grasp poses and robot motions to satisfy the target placement configuration.
Bag-of-Word-Groups (BoWG): A Robust and Efficient Loop Closure Detection Method Under Perceptual Aliasing
Xiang Fei, Lu Li
Computational EfficiencySimultaneous Localization and MappingImage
🎯 What it does: Proposed and implemented a loop closure detection method called Bag-of-Word-Groups (BoWG), aiming to improve precision-recall rate, robustness, and computational efficiency in perceptually confusing environments.
BaTCAVe: Trustworthy Explanations for Robot Behaviors
Som Sagar, Ransalu Senanayake
Explainability and InterpretabilityRobotic Intelligence
🎯 What it does: Propose a trustworthy explanation technique based on human-interpretable high-level concepts for explaining neural networks in robot decision-making.
Bayesian Morphology Optimization for Musculoskeletal Systems
Jing Zhao, Huaping Liu
OptimizationRobotic Intelligence
🎯 What it does: Optimize the stiffness parameters of muscles using the Bayesian Morphology Optimization (BMO) method on the MyoSuite platform to enhance the grasping performance of a musculoskeletal arm when handling objects of different weights.
Beacon: A Naturalistic Driving Dataset During Blackouts for Benchmarking Traffic Reconstruction and Control
Supriya Sarker, Weizi Li
Autonomous DrivingTabularTime SeriesBenchmark
🎯 What it does: Collected and analyzed natural driving data during the dark period at two major intersections in Memphis, studying traffic demand, vehicle trajectories, and density, and evaluated the impact of robotic vehicles on intersection delays and environmental effects.
BeeTLe: Blind Terrain-aware Learned Locomotion
Rogier Fransen (University of Surrey), Simon Hadfield (University of Surrey)
Robotic IntelligenceRecurrent Neural NetworkReinforcement LearningMixture of ExpertsTime Series
🎯 What it does: Developed the BeeTLe framework to achieve terrain-aware gait learning without the need for terrain sensors.
Bench4Merge: A Comprehensive Benchmark for Merging in Realistic Dense Traffic with Micro-Interactive Vehicles
Zhengming Wang, Yilun Chen
Autonomous DrivingLarge Language ModelBenchmark
🎯 What it does: Proposed a closed-loop evaluation benchmark for assessing the motion planning capability of merging in high-density traffic, using micro-interaction vehicles trained on a large-scale dataset to enhance scenario diversity, and reconstructing the evaluation mechanism through large language models (LLM).
Benchmark for Evaluating Long-Term Localization in Indoor Environments under Substantial Static and Dynamic Scene Changes
Niklas Trekel, C. Stachniss
Robotic IntelligenceSimultaneous Localization and MappingImageMultimodalityPoint CloudTime SeriesBenchmark
🎯 What it does: Proposed a new indoor long-term localization dataset and benchmark for evaluating localization methods in complex and dynamic scenarios.
Benchmarking Long-Horizon Mobile Manipulation in Multi-Room Dynamic Environments
Junbo Zhang, Kaisheng Ma
Robotic IntelligenceVision-Language-Action ModelWorld ModelTextGraphBenchmark
🎯 What it does: Proposes a benchmark task for multi-room long-term mobile manipulation, achieving dynamic tracking of object-furniture-room relationships through hierarchical scene graph memory.
Benchmarking Shortcutting Techniques for Multi-Robot-Arm Motion Planning
Philip Huang, Jiaoyang Li
OptimizationRobotic IntelligenceBenchmark
🎯 What it does: Quantitatively compare shortcutting methods in multi-arm motion planning, analyze their advantages and disadvantages, and propose two combination strategies to improve performance and runtime balance.
BEV-LIO(LC): BEV Image Assisted LiDAR-Inertial Odometry with Loop Closure
Haoxin Cai, Jianqi Liu
Autonomous DrivingConvolutional Neural NetworkSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: Proposes a BEV-LIO(LC) framework that combines bird's-eye view (BEV) image representations with LiDAR-Inertial Odometry, achieving loop closure through BEV image features.
BEVDiffLoc: End-to-End LiDAR Global Localization in BEV View based on Diffusion Model
Ziyue Wang, Huimin Lu
Pose EstimationAutonomous DrivingTransformerDiffusion modelPoint Cloud
🎯 What it does: Proposed the BEVDiffLoc framework, treating LiDAR localization as a problem of conditional pose generation, and achieving end-to-end localization through BEV perspective, data augmentation, maximum feature aggregation module, Vision Transformer, and diffusion model;
BEVDriver: Leveraging BEV Maps in LLMs for Robust Closed-Loop Driving
Katharina Winter, Fabian B. Flohr
Autonomous DrivingTransformerLarge Language ModelVision-Language-Action ModelImagePoint CloudBenchmark
🎯 What it does: Proposed the BEVDriver model, which achieves end-to-end closed-loop driving by leveraging BEV features and large language models (LLM).
BEVPointNet3D: Fusing Bird’s Eye View and Point Cloud Features for Robust 3D Lane Detection
Xia Yuan, Zihui Jing
Autonomous DrivingConvolutional Neural NetworkPoint Cloud
🎯 What it does: Proposed a 3D lane detection model called BEVPointNet3D, which integrates bird's-eye view (BEV) images with LiDAR point cloud features to address the limitations of traditional methods that rely solely on flat ground assumptions.
Beware of the Tablet: A Dominant Distractor in Human-Robot Interaction
Linlin Cheng, Koen V. Hindriks
Robotic Intelligence
🎯 What it does: In a face-to-face human-robot interaction experiment, researchers had participants interact with a humanoid robot using two out of three communication methods: voice, tablet, and gesture, recording task completion time, error rates, and the number and duration of participants' fixations on the robot's face, tablet, and gestures.
Beyond Anthropomorphism: Enhancing Grasping and Eliminating a Degree of Freedom by Fusing the Abduction of Digits Four and Five
Simon Fritsch, Robert K. Katzschmann
Robotic IntelligenceReinforcement LearningPoint CloudBenchmark
🎯 What it does: Proposed the SABD hand, a 16-degree-of-freedom robotic hand that employs a single large-range thumb and pinky merged additive/subtractive joint to expand the grasping workspace and reduce the number of actuators.
Bi-directional Cable-driven Ankle Exoskeleton Coupled with Series Elastic Actuator for Compliant Gait Assisting*
Yao Tu, Qingquan Li
Robotic Intelligence
🎯 What it does: A lightweight bidirectional cable-driven ankle exoskeleton system (total mass 2.6 kg) was developed, employing a series of elastic drive mechanisms and achieving bidirectional assistance through a waist drive unit and Bowden cable.
Bidirectional Task–Motion Planning Based on Hierarchical Reinforcement Learning for Strategic Confrontation
Qizhen Wu, Jinhu Lü
Reinforcement Learning
🎯 What it does: Proposes a bidirectional task-motion planning framework based on hierarchical reinforcement learning to integrate discrete commands and continuous actions for achieving intelligent decision-making.
Bimanual Robot-Assisted Dressing: A Spherical Coordinate-Based Strategy for Tight-Fitting Garments
Jian Zhao, Jihong Zhu
Robotic Intelligence
🎯 What it does: Propose a dual-arm robot-assisted dressing strategy, using spherical coordinate system encoding for dressing trajectories of tight clothing, leveraging the azimuth angle as a task-related feature to achieve adaptive dressing for different human arm postures;
Bio-Inspired Hybrid Map: Spatial Implicit Local Frames and Topological Map for Mobile Cobot Navigation
T. Dang, M. Huber
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose a hybrid map construction and navigation method based on human perceptual approaches, utilizing spatial implicit local frames and topological maps for mobile collaborative robot navigation.
Bio-Inspired Plastic Neural Networks for Zero-Shot Out-of-Distribution Generalization in Complex Animal-Inspired Robots
Binggwong Leung, P. Manoonpong
Domain AdaptationRobotic Intelligence
🎯 What it does: Improved a Hebbian learning-based neural network by introducing a weight normalization mechanism, evaluated its performance on a 18-DOF beetle-like robot and a 16-DOF gecko-like robot for locomotion control, demonstrating its ability to adapt to unknown terrains and morphological damage under zero-shot sim-to-real transfer.
Bio-inspired Shape Self-Assembly in Large-Scale Swarm Robots Under Information Asymmetry *
Jiali Wang, Jun Luo
Robotic Intelligence
🎯 What it does: The study addresses the problem of large-scale swarm robot shape self-assembly under information asymmetry conditions, and proposes a novel bio-inspired distributed self-assembly strategy.
Bio-Skin: A Cost-Effective Thermostatic Tactile Sensor with Multi-Modal Force and Temperature Detection
Haoran Guo, Lingfeng Tao
Robotic IntelligenceBiomedical Data
🎯 What it does: Developed a low-cost multimodal tactile sensor called Bio-Skin, capable of measuring planar normal force, two-dimensional shear force, and temperature, while achieving temperature control functionality.
Bioinspired Directional Adhesives Enable High Stiffness Layer Jamming in Soft Actuators *
Zhihuan Wang, Zhihua Zhang
Robotic Intelligence
🎯 What it does: Propose a novel interlayer damping method using a biologically inspired directional adhesive as an interlayer membrane to regulate the stiffness of soft actuators;
Biomechanically-Inspired Bipedal Robot Locomotion via Hybrid Gait Representation and Model-Guided Reinforcement Learning
Lijie Xie, Hui Cheng
Robotic IntelligenceReinforcement Learning
🎯 What it does: A bio-inspired low-level control framework is proposed, combining low-dimensional gait representation with the linear inverted pendulum model (LIPM) as a motion descriptor in reinforcement learning (RL). This framework trains control strategies that dynamically balance biomechanical realism and adaptability, achieving smooth and natural gaits for bipedal robots. The approach was validated on a physical robot, demonstrating stable and efficient locomotion.
Blind-Wayfarer: A Minimalist, Probing-Driven Framework for Resilient Navigation in Perception-Degraded Environments
Yanran Xu, Danesh S. Tarapore
Autonomous DrivingRobotic Intelligence
🎯 What it does: This paper proposes Blind-Wayfarer, an exploration-driven minimal navigation framework that primarily utilizes a compass to enable safe navigation for autonomous robots in perception-limited environments;
BlueKoi: Combining a Tuna-Inspired Tail and Koi-Inspired Body Bending for Maneuverability
Irene Sha, Radhika Nagpal
Robotic Intelligence
🎯 What it does: Designed and tested a cordless fish-shaped robot platform named BlueKoi, combining a rigid tail (similar to tuna) for efficient propulsion and a rotatable head (similar to carp) for maneuverability. Experiments showed a speed of 1.84 body lengths per second and a turning radius of 1.93 body lengths.
Body-Hand Modality Expertized Networks with Cross-attention for Fine-grained Skeleton Action Recognition
Seungyeon Cho, Tae-Kyun Kim
RecognitionGraph Neural NetworkMixture of ExpertsMultimodalityGraph
🎯 What it does: Proposes the BHaRNet framework, which adds a hand expert model to traditional full-body expert models, uses a collaborative specialized ensemble loss for joint training, and implements cross-attention through expert branches and pooling attention modules to promote feature-level interaction and information fusion;
Body-Temperature-Responsive Balloon Actuator for Adaptive In-Ear Microneedle Electrode Deployment
Ruizhou Zhao, Hongliang Ren
Biomedical Data
🎯 What it does: Designed and implemented a temperature-responsive in-ear balloon actuator (BBA) based on liquid-gas phase transition, integrated with silver needle-shaped microneedle electrodes to achieve stable electrophysiological monitoring.
BookBot: A Robotic Manipulation Benchmark for Voice-Driven Book Recognition and Grasping in Cluttered Environments
Huaqiang Wang, Shen Wang
Robotic IntelligenceLarge Language ModelImageTextBenchmark
🎯 What it does: Constructed the THU-Book dataset and developed the Voice-driven BookBot robotic system for book recognition, localization, and grasping.
Boosting Omnidirectional Stereo Matching with a Pre-trained Depth Foundation Model
Jannik Endres, Alexandre Alahi
Depth EstimationSupervised Fine-TuningImage
🎯 What it does: Proposes a panoramic stereo matching method called DFI-OmniStereo, which utilizes a pre-trained large depth foundation model for relative monocular depth estimation and embeds it into an iterative optimization stereo matching architecture.