arXivSub Start free trial

IROS 2025 Papers — Page 8

IEEE/RSJ International Conference on Intelligent Robots and Systems · 1984 papers

Event-Triggered Maps of Dynamics: A Framework for Modeling Spatial Motion Patterns in Non-Stationary Environments

Junyi Shi, Tomasz Kucner

Diffusion modelStochastic Differential Equation

🎯 What it does: Proposes the Event-Triggered Dynamic Mapping (ETMoD) framework for modeling spatial movement patterns in non-stationary environments.

EventSync: Joint Recovery of Temporal Offsets and Relative Orientations for Wide-Baseline Event Cameras

Wanli Xing, Jia Pan

Pose EstimationOptical Flow

🎯 What it does: Developed a software algorithm called EventSync for synchronizing multi-camera event streams and estimating the relative pose between cameras.

Evidential Uncertainty Estimation for Multi-Modal Trajectory Prediction

Sajad Marvi, A. Valada

Autonomous DrivingMultimodality

🎯 What it does: Propose a multi-modal trajectory prediction method based on evidence deep learning, which estimates position and mode probability uncertainties in real-time.

EvidMTL: Evidential Multi-Task Learning for Uncertainty-Aware Semantic Surface Mapping from Monocular RGB Images

Rohit Menon, Maren Bennewitz

SegmentationDepth EstimationSimultaneous Localization and MappingImage

🎯 What it does: Proposed the EvidMTL multi-task learning framework, combining evidence-based depth estimation with semantic segmentation to achieve uncertainty-aware reasoning from monocular RGB images.

ExAMPC: the Data-Driven Explainable and Approximate NMPC with Physical Insights

Jean Pierre Allamaa, Tong Duy Son

OptimizationExplainability and InterpretabilityPhysics Related

🎯 What it does: Propose the ExAMPC method, combining data-driven explainable AI with traditional nonlinear model predictive control (NMPC), to achieve interpretable approximate NMPC and enhance system trustworthiness.

ExFace: Expressive Facial Control for Humanoid Robots with Diffusion Transformers and Bootstrap Training

Dong Zhang, Jiahao Chen

Robotic IntelligenceTransformerDiffusion modelImage

🎯 What it does: Proposes the Expressive Facial Control (ExFace) method based on Diffusion Transformers, achieving precise mapping from facial blendshapes to biomimetic robot motor control.

Exo-ViHa: A Cross-Platform Exoskeleton System with Visual and Haptic Feedback for Efficient Dexterous Skill Learning

Xintao Chao, Wenbo Ding

Robotic IntelligenceSimultaneous Localization and MappingMultimodality

🎯 What it does: Proposed a cross-platform exoskeleton system Exo-ViHa based on 3D printing, which provides visual and tactile feedback and collects robotic hand operation data in real-time from a first-person perspective.

Exoskeleton Gait Adaptation Framework via Hm-DMP and PI2 Optimization for Dynamic Patient Mobility Matching

Qiaohuan Cao, Wei Yang

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a closed-loop mobile matching framework that integrates hybrid multi-attractor dynamic movement primitives (Hm-DMP) with path integral strategy improvement (PI2) optimization, achieving real-time adaptive control of lower-limb exoskeleton gait trajectories;

Experimental Comparison of Whole-Body Control Formulations for Humanoid Robots in Task Acceleration and Task Force Spaces

Sait Sovukluk, Christian Ott

Robotic Intelligence

🎯 What it does: An experimental comparison of Inverse Dynamics Whole-Body Control (ID-WBC) and Passivity-Based Whole-Body Control (PB-WBC) on bipedal robots, covering swing leg position and attitude control, squatting with an additional mass modeled without a model, and jumping tasks.

Experimental Evaluation of Radio-aware Semantic Map with 5G-Enabled Mobile Robots

A. Ibanez, Xavier Pérez Costa

Robotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Proposed a framework for constructing and utilizing wireless perception semantic maps based on 5G networks, supporting mobile robots in performing task offloading and real-time navigation decisions in unknown environments;

Experimental Evaluation of Safe Trajectory Planning for an Omnidirectional UAV

Mahmoud Hamandi, F. Khorrami

Safty and PrivacyRobotic Intelligence

🎯 What it does: Propose a safe trajectory planning framework for omnidirectional drones that dynamically adjusts tracking speed based on obstacle proximity, and verify its navigation capability in confined spaces on the OmniOcta UAV.

Experimental Open-Source Framework for Underwater Pick-and-Place Studies with Lightweight UVMS – An Extensive Quantitative Analysis

Nathalie Bauschmann, Daniel A. Duecker

OptimizationRobotic IntelligenceBenchmark

🎯 What it does: This paper proposes a complete open-source software framework for fully automated grasping and releasing experiments with a lightweight underwater manipulator (UVMS), extending the previous task-priority control framework by incorporating high-level decision-making and grasping detection methods, and verifying them on BlueROV2 and Alpha5 Manipulator.

eXplainable Intention Estimation in Teleoperated Manipulation Using Deep Dynamic Graph Neural Networks

Prakash Baskaran, Soshi Iba

Explainability and InterpretabilityRobotic IntelligenceGraph Neural Network

🎯 What it does: Developed and demonstrated a dynamic intent estimation framework based on graph neural networks, which can predict low-level actions and high-level tasks in dual-armed robot teleoperation, and validated the correctness and consistency of predictions using interpretability metrics.

ExpliDrive: Bridging Model Predictive Control and Transformers for Interactive Autonomous Driving

Zhexi Lian, Jia Hu

Autonomous DrivingOptimizationExplainability and InterpretabilityTransformer

🎯 What it does: Propose ExpliDrive, which utilizes Transformer to encode vehicle interactions and embeds them into MPC motion planning, achieving active interaction perception and response.

Exploiting Policy Idling for Dexterous Manipulation

Annie S. Chen, Dushyant Rao

Robotic IntelligenceReinforcement Learning

🎯 What it does: Propose a method to alleviate the 'stagnation' problem in robots during high-precision tasks such as grasping and insertion by applying perturbations (Pause-Induced Perturbations, PIP) in detected paused states, and evaluate it on simulated dual-arm tasks and real-world insertion tasks.

Exploratory Movement Strategies for Texture Discrimination with a Neuromorphic Tactile Sensor

Xingchen Xu, B. Ward-Cherrier

RecognitionComputational EfficiencyRobotic Intelligence

🎯 What it does: Developed a robot texture classification framework based on neuromorphic tactile sensors, investigating the impact of different exploration actions on classification performance.

Exploring Spontaneous Social Interaction Swarm Robotics Powered by Large Language Models

Yitao Jiang, Devin J. Balkcom

Robotic IntelligenceTransformerLarge Language Model

🎯 What it does: Built a decentralized multi-robot system that utilizes large language models (LLMs) for reasoning, enabling robots to discover each other, share information, and dynamically coordinate through natural language.

Exploring Stiffness Gradient Effects in Magnetically Induced Metamorphic Materials via Continuum Simulation and Validation

Wentao Shi, Hongliang Ren

Physics Related

🎯 What it does: Constructed graded stiffness magnetically induced deformable materials (GMCs) and developed a numerical model for their bending performance, while training an efficient and accurate bending prediction extension model.

Exploring the Virtual Pivot Point in Unilateral Transfemoral Amputee Locomotion: Implications for Prosthetic Development

Omid Mohseni, A. Seyfarth

Biomedical Data

🎯 What it does: First study on the virtual pivot point (VPP) in unilateral above-knee amputees, analyzing changes in quality and position of VPP in longitudinal and planar dynamics

Explosive Jumping with Rigid and Articulated Soft Quadrupeds via Example Guided Reinforcement Learning

Georgios Apostolides, Jiatao Ding

Robotic IntelligenceReinforcement Learning

🎯 What it does: This paper proposes a deep reinforcement learning framework based on example guidance, which learns controlled, variable-distance jumps for quadruped robots equipped with passive elastic structures, and enhances robustness to uneven terrain through advanced training.

Exponentially Weighted Instance-Aware Repeat Factor Sampling for Long-Tailed Object Detection Model Training in Unmanned Aerial Vehicles Surveillance Scenarios

Taufiq Ahmed, Sasu Tarkoma

Object DetectionData-Centric LearningConvolutional Neural NetworkImage

🎯 What it does: Developed an exponentially weighted instance-aware resampling method (E-IRFS) to address long-tailed class imbalance in UAV monitoring scenarios.

Extraction of Robotic Surface Processing Strategies from Human Demonstrations

Thomas Eiband, Alin Albu-Schäffer

Robotic IntelligenceReinforcement Learning from Human FeedbackMultimodalitySequential

🎯 What it does: This paper proposes a system that records human actions and contact forces during surface processing using electric sanding tools, provides a public dataset, and extracts robot execution strategies based on multi-user demonstrations through Learning from Demonstration (LfD).

Extreme-Hydrostatic-Pressure Resilient Dielectric Elastomer Actuator for Propeller Propulsion

Boyuan Du, Huichan Zhao

Robotic IntelligencePhysics Related

🎯 What it does: Designed and tested a rotational dielectric elastomer actuator capable of sustained operation under extremely high hydrostatic pressure for deep-sea propulsion.

Eye-In-Finger: Smart Fingers for Delicate Assembly and Disassembly of LEGO

Zhenran Tang, Changliu Liu

Robotic IntelligenceImage

🎯 What it does: Proposed the 'Eye-In-Finger' tool design concept, embedding low-cost high-resolution visual perception directly into the tool tip, and achieving real-time fine calibration for LEGO assembly and disassembly tasks, enhancing the robot's precise manipulation capabilities in cluttered environments.

FABG : End-to-end Imitation Learning for Embodied Affective Human-Robot Interaction

Yang Zhang, Jianwei Zhang

Robotic IntelligenceVideo

🎯 What it does: Proposed FABG (Facial Affective Behavior Generation) — an end-to-end imitative learning system for generating natural and fluent human-robot interaction facial emotional behaviors. High-quality demonstrations were collected using an immersive VR demonstration system, a predictive-driven delay compensation strategy was proposed, and the system was deployed and validated on a 25-degree-of-freedom humanoid robot, demonstrating effectiveness in four interaction tasks: emotional interaction, dynamic tracking, gaze focusing, and gesture recognition.

FABRIC: FAbricating Bodily-Expressive Robots for Inclusive and Low-Cost Design

A. Arabi, Jeeeun Kim

Robotic IntelligenceVideo

🎯 What it does: Developed an end-to-end toolkit called FABRIC for low-cost 3D printed robot fabrication and body language programming. The robot captures and translates users' body expressions through learning from demonstrations, and is equipped with a visual programming interface to integrate multi-source demonstrations.

Failure Forecasting Boosts Robustness of Sim2Real Rhythmic Insertion Policies

Yuhan Liu, Abdeslam Boularias

Pose EstimationDomain AdaptationReinforcement Learning

🎯 What it does: Proposed a sim-to-real framework for rhythmic insertion tasks, combining reinforcement learning insertion strategies with a failure prediction module, utilizing 6D joint pose tracking to achieve high-precision insertion and rotation, and performing simple recovery operations when failures occur.

FalconGym: A Photorealistic Simulation Framework for Zero-Shot Sim-to-Real Vision-Based Quadrotor Navigation

Yan Miao, Sayan Mitra

Domain AdaptationRobotic IntelligenceTransformerReinforcement LearningNeural Radiance FieldImageMultimodality

🎯 What it does: Built a realistic simulation framework FalconGym for training visual control policies in NeRF environments, achieving zero-shot transfer from simulation to real-world flight;

Fast Policy: Accelerating Visuomotor Policies without Re-training

Tong Wu, Zheng Wang

Computational EfficiencyRobotic IntelligenceConvolutional Neural NetworkReinforcement LearningDiffusion model

🎯 What it does: Propose Fast Policy (FP), which accelerates visual motion policies by reusing UNet encoder features in non-critical denoising steps and dynamically selecting key steps.

Fast Real-Time Neural Network-Based Kinematics Solving of the Cosserat Rod Model for a Parallel Continuum Surgical Manipulator

Xipeng Wu, Changsheng Li

Computational EfficiencyRobotic IntelligenceBiomedical DataPhysics Related

🎯 What it does: Propose a real-time inverse kinematics solver based on neural networks for miniaturized parallel continuum surgical manipulators, using the Cosserat rod model for kinematic modeling.

Fast-Revisit Coverage Path Planning for Autonomous Mobile Patrol Robots Using Long-Range Sensor Information

Srinivas Kachavarapu, Reinhard Gerndt

OptimizationRobotic Intelligence

🎯 What it does: Proposes the Fast-Revisit Coverage Path Planning (FaRe-CPP) algorithm for autonomous mobile patrolling robots to achieve rapid revisit coverage path planning under limited sensing ranges.

Fast(er) Robust Point Cloud Alignment Using Lie Algebra

Jean-Thomas Sexton, Jonathan Gaudreault

Pose EstimationComputational EfficiencyPoint CloudBenchmark

🎯 What it does: Propose an iterative reweighted least squares (IRLS) algorithm based on Lie algebra for robust 3D point cloud registration.

FASTEX: Fast UAV Exploration in Large-Scale Environments Using Dynamically Expanding Grids and Coverage Paths

Xiaoxun Zhang, Hui Cheng

Computational EfficiencyRobotic Intelligence

🎯 What it does: Proposes the FASTEX framework, which utilizes dynamic grid expansion and coverage paths to achieve rapid exploration of unmanned aerial vehicles (UAVs) in large environments.

FastTrack: GPU-Accelerated Tracking for Visual SLAM

Kimia Khabiri, Steven Y. Ko

Simultaneous Localization and MappingVideo

🎯 What it does: Accelerating the tracking module of visual-inertial SLAM systems using GPU, particularly stereo feature matching and local map tracking, to enhance tracking performance.

Fault-Tolerant Model Predictive Control for Safety of Unmanned Surface Vessels Berthing under Multimodal disturbances and Various Constraints

Jiangteng Shi, Yujing Chen

Autonomous DrivingOptimizationSafty and Privacy

🎯 What it does: A fault-tolerant model predictive control (FTMPC) framework is proposed to achieve safe berthing operations of unmanned surface vessels under multi-mode disturbances (external ocean disturbances EODs and internal thruster faults ITFs) and various constraints (underactuated nonlinear dynamics, actuator saturation, obstacle avoidance constraints).

FBG-based Actuation and Data Driven Contact Detection for Smart Steerable Instruments

Syed Zain Mehdi, E. V. Poorten

Robotic IntelligenceRecurrent Neural NetworkTime Series

🎯 What it does: A compact driving system based on fiber Bragg grating (FBG) was developed to control the bending of a steerable tip, and strain was measured on the FBG to estimate thrust. Subsequently, a long short-term memory network (LSTM) was used to predict contact with the surrounding environment based on thrust.

FCRF: Flexible Constructivism Reflection for Long-Horizon Robotic Task Planning with Large Language Models

Yufan Song, Shiqiang Zhu

Robotic IntelligenceTransformerLarge Language Model

🎯 What it does: Proposed the FCRF (Flexible Constructivism Reflection Framework), a Mentor-Actor architecture that utilizes large language models (LLMs) to achieve flexible self-reflection in long-term robot task planning.

Fear-Based Behavior Adaptation for Robust Walking Robots using Unsupervised Health Estimation

Tristan Schnell, R. Dillmann

Anomaly DetectionRobotic IntelligenceMultimodality

🎯 What it does: Implementing fear-based robot behavior adaptation using unsupervised anomaly detection, enabling robots to quickly and automatically respond to any unexpected issues in unknown environments.

Feasibility Analysis of real-time Robustness Certification

Emmanouil Seferis, Stefanos D. Kollias

Safty and PrivacyComputational EfficiencyAdversarial AttackImage

🎯 What it does: Investigated the feasibility of achieving real-time robustness proofs in Randomized Smoothing (RS) by significantly reducing the number of sampling times, and proposed a self-improvement method to alleviate the certified radius degradation caused by sample reduction.

Feature Matching-Based Gait Phase Prediction for Obstacle Crossing Control of Powered Transfemoral Prosthesis

Jiaxuan Zhang, Chenglong Fu

OptimizationHyperparameter SearchRobotic IntelligenceTime SeriesBiomedical Data

🎯 What it does: For lower-limb amputees using powered knee prostheses, the study installs inertial sensors on the healthy ankle, uses a genetic algorithm-optimized neural network to predict thigh and knee angles, and employs a gait phase prediction algorithm to determine the drive angle index of the prosthetic knee motor, thereby achieving obstacle crossing control.

Feature-aligned Fisheye Object Detection Network for Autonomous Driving

Hu Cao, Alois C. Knoll

Object DetectionAutonomous DrivingConvolutional Neural NetworkImage

🎯 What it does: Proposed a feature-aligned fisheye object detection network for autonomous driving, combining the FaPM and LaDH modules to achieve feature and spatial alignment

FedEMA: Federated Exponential Moving Averaging with Negative Entropy Regularizer in Autonomous Driving

Wei-Bin Kou, Yik-Chung Wu

Autonomous DrivingFederated LearningImage

🎯 What it does: Propose the FedEMA framework, integrating server-side EMA with vehicle-side negative entropy regularization to address the time catastrophic forgetting problem in FL AD models.

Feedback Control of a Single-Tail Bioinspired 59-mg Swimmer

Conor K. Trygstad, N. O. Pérez-Arancibia

Robotic IntelligencePhysics Related

🎯 What it does: Developed a single-tail FRISSHBot based on shape memory alloy (SMA) dual-fold actuator, achieving controllable motion in 2D space, and first realized feedback-controlled trajectory tracking at the sub-millimeter level in robotic systems.

Feedback Control of a Two-Degree-of-Freedom Electromagnetic Reluctance Precision Motion System

Michael Pumphrey, M. Janaideh

OptimizationPhysics Related

🎯 What it does: Designed and verified an Xθ two-degree-of-freedom electromagnetic impedance precision motion system based on magnetic reluctance drive, combining reluctance actuator with two moving magnet drivers to achieve precise translational and rotational control.

FEG-VON: Frontier Embedding Graph for Efficient Visual Object Navigation

Yingru Dai, Guijin Wang

Autonomous DrivingComputational EfficiencyGraph Neural NetworkVision Language ModelMultimodality

🎯 What it does: Proposes a training-free framework that constructs and maintains a frontier embedding graph to achieve efficient visual goal navigation.

FGS-SLAM: Fourier-based Gaussian Splatting for Real-time SLAM with Sparse and Dense Map Fusion

Yansong Xu, Weijia Zhou

Pose EstimationAutonomous DrivingGaussian SplattingSimultaneous Localization and MappingImage

🎯 What it does: Proposes an adaptive densification method based on Fourier frequency domain analysis to achieve fast convergence of Gaussian representations, and constructs independent and unified sparse and dense maps. The sparse map is used for efficient tracking, while the dense map is used for high-fidelity visual representation.

FIELD: Fast Information-driven Autonomous Exploration using Larger Perception Distance

Yuefeng Zhang, Wenbing Tao

OptimizationRobotic Intelligence

🎯 What it does: Proposes a planner called FIELD that enables fast, information-driven autonomous exploration for UAVs by leveraging a larger perception range.

Find Everything: A General Vision Language Model Approach to Multi-Object Search

Daniel Choi, Aaron Hao Tan

Object DetectionObject TrackingVision Language Model

🎯 What it does: Proposes a novel multi-object search method called Finder, which utilizes a visual language model to locate multiple objects and achieves simultaneous tracking and reasoning through multi-channel score maps and scene-level and object-level semantic relevance score maps; validated through extensive experiments in both simulated and real environments.

Finite-time Guiding Vector Fields for Accelerated Path Following of Nonholonomic Robots

Jian Yang, Weijia Yao

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Proposed a finite-time guided vector field (finite-time GVF) based on the sign function, and designed a controller for a single-vehicle model to enable nonholonomic robots to follow any smooth n-dimensional path within finite time. The method was extended to multi-robot distributed motion coordination, and its effectiveness was validated through two unmanned ground vehicle experiments.

FLAME: A Federated Learning Benchmark for Robotic Manipulation

Santiago Bou Betran, Danica Kragic

Federated LearningRobotic IntelligenceBenchmark

🎯 What it does: Proposes the FLAME benchmark for federated learning in distributed robotic manipulation tasks.

Flexible Electronic Device with Multifunctional Tactile Perception for Enhanced Robotic Interaction

Chenhao Mao, Yancheng Wang

Robotic Intelligence

🎯 What it does: Developed a flexible electronic device for multifunctional tactile sensing to enable robot-human interaction; integrated the device into a robot interaction system to simultaneously measure force, angle, and sliding distance;

Flipping Manipulation with a Two-Fingered Parallel-Jaw Gripper

Wenxi Liao, Xin Jiang

OptimizationRobotic Intelligence

🎯 What it does: Proposed a systematic flipping strategy for parallel grippers in structured environments.

FlipWalker: Jacob’s Ladder toy-inspired robot for locomotion across diverse, complex terrain

Diancheng Li, Matthew A. Robertson

Robotic Intelligence

🎯 What it does: Designed an underactuated robot named FlipWalker inspired by Jacob’s Ladder, capable of moving forward by rolling over complex terrains.

FLOAT Drone: A Fully-actuated Coaxial Aerial Robot for Close-Proximity Operations

Junxiao Lin, Fei Gao

Robotic Intelligence

🎯 What it does: Designed and implemented a fully actuated coaxial twin-rotor UAV—FLOAT Drone—for close-range operations.

Floorplan-SLAM: A Real-Time, High-Accuracy, and Long-Term Multi-Session Point-Plane SLAM for Efficient Floorplan Reconstruction

Haolin Wang, Yihong Wu

OptimizationComputational EfficiencySimultaneous Localization and MappingImagePoint Cloud

🎯 What it does: Proposes Floorplan-SLAM, a real-time, high-precision long-term plane reconstruction framework that tightly integrates floor plane reconstruction into a multi-session SLAM system, achieved using only a stereo camera;

FloPE: Flower Pose Estimation for Precision Pollination

Rashik Shrestha, Yu Gu

Data SynthesisPose EstimationKnowledge DistillationRobotic IntelligenceGaussian SplattingImageAgriculture Related

🎯 What it does: Proposes FloPE, a real-time flower pose estimation framework for computationally constrained robotic pollination systems.

Flow-Aware Navigation of Magnetic Micro-Robots in Complex Fluids via PINN-Based Prediction

Yongyi Jia, Xiang Li

Robotic IntelligenceConvolutional Neural NetworkPhysics Related

🎯 What it does: Proposed a flow-aware navigation and control strategy for magnetic microrobots, utilizing PI-UNet to improve fluid velocity prediction and integrating it into A* path planning and control schemes

FlowMP: Learning Motion Fields for Robot Planning with Conditional Flow Matching

K. Nguyen, Minh Nhat Vu

Robotic IntelligenceFlow-based Model

🎯 What it does: By extending the flow matching method to capture second-order trajectory dynamics, learning a motion field incorporating acceleration information, directly mapping a simple prior distribution to smooth, executable robot trajectories.

FlowNav: Combining Flow Matching and Depth Priors for Efficient Navigation

Samiran Gode, Wolfram Burgard

Depth EstimationRobotic IntelligenceFlow-based Model

🎯 What it does: Proposes FlowNav, a robot navigation method that combines Conditional Flow Matching with depth prior.

FlowPlan: Zero-Shot Task Planning with LLM Flow Engineering for Robotic Instruction Following

Zijun Lin, Hong Zhang

Robotic IntelligenceLarge Language ModelText

🎯 What it does: Propose FlowPlan, a multi-stage LLM workflow for zero-shot robot instruction following task planning.

Flying on Point Clouds with Reinforcement Learning

Guangtong Xu, Fei Gao

Representation LearningRobotic IntelligenceReinforcement LearningPoint Cloud

🎯 What it does: Combining onboard 3D LiDAR and simulation-to-real reinforcement learning to achieve autonomous drone flight in complex environments.

FOCI: Trajectory Optimization on Gaussian Splats

Mario Gomez Andreu, Marco Hutter

OptimizationRobotic IntelligenceGaussian SplattingPoint Cloud

🎯 What it does: Proposed a FOCI algorithm that directly optimizes the robot's trajectory on 3D Gaussian Splatting

Focus Bug: Learning Environmental Awareness for Efficient Mapless Navigation

Charles Dansereau, Gabriela Nicolescu

Autonomous DrivingComputational EfficiencyReinforcement LearningPoint Cloud

🎯 What it does: Propose the Focus Bug algorithm, which selects necessary sensor data through reinforcement learning and combines it with classical mapless navigation methods to achieve mapless navigation with extremely low computational load.

Focusing on Projection-Stable Patch: Cross-View Localization with Geometric-Semantic Alignment

Riyu Qin, Xia Yuan

Pose EstimationAutonomous DrivingImagePoint Cloud

🎯 What it does: Propose a new feature alignment strategy that combines the Perspective-Driven Attention Fusion (PDAF) module with the Projection-Stable Patch-Guided Pose Optimizer (PSPG) for cross-perspective (ground-to-satellite) geolocation, addressing occlusion and distortion caused by perspective changes.

FoldPath: End-to-End Object-Centric Motion Generation via Modulated Implicit Paths

Paolo Rabino, Tatiana Tommasi

GenerationRobotic Intelligence

🎯 What it does: Proposed FoldPath, an end-to-end neural field method for object-centric motion generation (OCMG).

Force Aware Branch Manipulation To Assist Agricultural Tasks

M. Rijal, Yu Gu

Robotic IntelligenceAgriculture Related

🎯 What it does: Developed a method for safely manipulating branches to assist in agricultural tasks.

Force-Sensor-free Contact Estimation for Lower Limb Exoskeleton Robots Based on Probabilistic Modeling and Fusion

Weigen Ye, Hong Cheng

Robotic IntelligenceTime Series

🎯 What it does: Propose a foot contact estimation method without force sensors

FPGA Hardware Neural Control of CartPole and F1TENTH Race Car

Marcin Paluch, Tobi Delbruck

Autonomous DrivingRobotic IntelligenceSupervised Fine-Tuning

🎯 What it does: Developed an FPGA-based neural controller for high-speed control in robotic systems such as cartpole and F1TENTH racing cars, demonstrating its ability to achieve kilohertz-level control rates.

FRANC: Feeding Robot for Adaptive Needs and Personalized Care

J.-Anne Yow, Wei Tech Ang

Robotic IntelligenceTransformerLarge Language ModelPrompt Engineering

🎯 What it does: Developed the FRANC framework, leveraging large language models and decomposed prompting strategies to dynamically adjust bite sequences, acquisition, and transfer parameters during the robot's feeding process, achieving cross-phase iterative correction;

Free-form language-based robotic reasoning and grasping

Runyu Jiao, Fabio Poiesi

Object DetectionData SynthesisRobotic IntelligenceVision Language ModelMultimodality

🎯 What it does: The study investigates using pre-trained vision-language models for robot grasping based on free-form language instructions under zero-shot conditions, and proposes the FreeGrasp method.

From 2D Underwater Imaging Sonar Data to 3D Plane Extraction

António J. Oliveira, N. Cruz

Image

🎯 What it does: 3D plane extraction using 2D imaging sonar scanning

From Learning to Mastery: Achieving Safe and Efficient Real-World Autonomous Driving with Human-in-the-Loop Reinforcement Learning

Zeqiao Li, Z. Zuo

Autonomous DrivingReinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: Proposed a reward-agnostic active human-in-the-loop learning method called H-DSAC to achieve safe and efficient real-world autonomous driving;

From Monocular Vision to Autonomous Action: Guiding Tumor Resection via 3D Reconstruction

Ayberk Acar, J. Wu

SegmentationRobotic IntelligenceImagePoint CloudBiomedical Data

🎯 What it does: This study proposes a 3D reconstruction pipeline that uses only RGB images to generate segmented point clouds of the target anatomy, thereby guiding automated navigation for tumor resection surgery.

From Satellite to Street: Semantic and Depth Information for Enhanced Geo-Localization

Yilong Zhu, Shaojie Shen

Pose EstimationDepth EstimationTransformerImage

🎯 What it does: Propose a localization method that combines perspective images and satellite images, using depth and semantic information extracted from monocular images for cross-view projection to achieve three degrees of freedom (3-DoF) precise pose estimation.

Frontier Shepherding: A Bio-inspired Multi-robot Framework for Large-Scale Exploration

John Lewis, Pedro U. Lima

Robotic Intelligence

🎯 What it does: Proposed and verified a multi-robot frontier exploration framework called Frontier Shepherding (FroShe) based on shepherd behavior for large-scale environment exploration.

Frozen Triumph: Lessons from GARMI’s Bimanual Trophy Handover at the Kandahar Ski World Cup – Shaping Current Research Directions

Mario Tröbinger (Garmi), Sami Haddadin (Garmi)

Robotic Intelligence

🎯 What it does: Completed the two-handed trophy handover demonstration of GARMI in extreme cold environments, and extended the previous framework to achieve real-time trophy quality estimation, while investigating the impact of base tilt and temperature variations on quality estimation and control performance.

FruitNeRF++: A Generalized Multi-Fruit Counting Method Utilizing Contrastive Learning and Neural Radiance Fields

Lukas Meyer, Marc Stamminger

Object DetectionNeural Radiance FieldContrastive LearningImagePoint CloudAgriculture Related

🎯 What it does: Proposed FruitNeRF++, a multi-fruit counting method combining contrastive learning and neural radiance fields (NeRF), capable of counting fruits in unstructured orchard images.

FSDP: Fast and Safe Data-Driven Overtaking Trajectory Planning for Head-to-Head Autonomous Racing Competitions

Cheng Hu (Zhejiang University), Lei Xie (Zhejiang University)

Autonomous DrivingOptimization

🎯 What it does: Propose Fast and Safe Data-Driven Planner for overtaking trajectory planning in autonomous racing.

FSGlove: An Inertial-Based Hand Tracking System with Shape-Aware Calibration

Yutong Li, Cewu Lu

Pose EstimationOptimizationTime Series

🎯 What it does: Proposed the FSGlove system, which utilizes inertial measurement units (IMUs) to achieve 48 degrees of freedom (DOF) hand motion capture, and employs the DiffHCal method for personalized hand shape reconstruction.

Full Pose Tracking via Robust Control for Over-Actuated Multirotors

Mohamad Hachem, Murat Bronz

OptimizationRobotic Intelligence

🎯 What it does: Designed a robust cascade control architecture to achieve full attitude tracking for overloaded multirotors.

Fully Autonomous Dual Arm Aerial Delivery Robot for Intralogistics: the euROBIN Nancy Competition Flight Dataset

A. Suárez, Aníbal Ollero

Robotic IntelligenceSimultaneous Localization and MappingBenchmark

🎯 What it does: Designed, developed, and verified a dual-arm drone system capable of autonomously performing mapping, localization, planning, and grasping packages. The system supports task execution in logistics environments with supply points, delivery points, labeled packages, and obstacles, and can generate task plans through voice commands.

FunGraph: Functionality Aware 3D Scene Graphs for Language-Prompted Scene Interaction

Dennis Rotondi, K. Arras

Object DetectionSegmentationRobotic IntelligenceVision-Language-Action ModelImageGraph

🎯 What it does: Develop a function-aware 3D scene graph to identify the location and usage of functional interaction elements, and perform fine-grained detection and storage of function-related components beneficial for robot interaction.

FUSE: Label-Free Image-Event Joint Monocular Depth Estimation via Frequency-Decoupled Alignment and Degradation-Robust Fusion

Pihai Sun, Xianming Liu

Depth EstimationTransformerMultimodality

🎯 What it does: Proposed a frequency-decoupled unsupervised image-event joint monocular depth estimation method called FUSE.

Fusion Scene Context: Robust and Efficient LiDAR Place Recognition Across Season

Fengkui Cao, Xieyuanli Chen

RecognitionAutonomous DrivingPoint Cloud

🎯 What it does: Proposed a compact fusion view image representation based on LiDAR point clouds to extract multi-view structural information, generate a global descriptor using texture features, and design regional features to adapt to seasonal changes.

G3 CN: Gaussian Topology Refinement Gated Graph Convolutional Network for Skeleton-Based Action Recognition

Haiqing Ren, Libo Zhang

RecognitionGraph Neural NetworkGraph

🎯 What it does: Proposes the Gaussian Topology Refinement Gated Graph Convolution (G3CN) method to enhance the discriminative ability for ambiguous actions in skeleton-based action recognition.

GABRIL: Gaze-Based Regularization for Mitigating Causal Confusion in Imitation Learning

Amin Banayeeanzade, Erdem Biyik

Autonomous DrivingSequential

🎯 What it does: Regularization learning using human gaze data to guide behavior cloning models to focus on causal-related features, thereby alleviating causal confusion.

GACL: Grounded Adaptive Curriculum Learning with Active Task and Performance Monitoring

Linji Wang, Xuesu Xiao

Robotic Intelligence

🎯 What it does: Proposed an adaptive curriculum learning framework GACL specifically designed for robot tasks, aiming to automatically generate task sequences that match the robot's current capabilities.

Gait in Eight: Efficient On-Robot Learning for Omnidirectional Quadruped Locomotion

Nico Bohlinger, Jan Peters

Robotic IntelligenceReinforcement Learning

🎯 What it does: Developed an efficient reinforcement learning framework for robots, enabling quadrupedal robots to learn full-directional locomotion within just 8 minutes of original real-time training.

GAT-Grasp: Gesture-Driven Affordance Transfer for Task-Aware Robotic Grasping

Ruixiang Wang, Kui Jia

Robotic IntelligenceVideoRetrieval-Augmented Generation

🎯 What it does: Proposed the GAT-Grasp framework, which utilizes human gestures to directly drive the generation of task-specific grasping poses, enabling robots to achieve precise grasping of diverse objects.

Gaussian Splatting with Reflectance Regularization for Endoscopic Scene Reconstruction

Chengkun Li, Q. Dou

Gaussian SplattingBiomedical Data

🎯 What it does: Propose a Gaussian Splatting-based endoscopic scene reconstruction framework with reflectance regularization to enhance geometric quality.

GaussianGraph: 3D Gaussian-Based Scene Graph Generation for Open-World Scene Understanding

Xihan Wang, Mengyin Fu

GenerationVision Language ModelGaussian Splatting

🎯 What it does: Propose the GaussianGraph framework, which combines 3D Gaussian Splatting with adaptive semantic clustering and scene graph generation to enhance 3D scene understanding and object localization capabilities.

GaussianPU: Color Point Cloud Upsampling via 3D Gaussian Splatting

Zixuan Guo, Fei Yu

Super ResolutionGaussian SplattingPoint Cloud

🎯 What it does: Proposed the GaussianPU framework, combining 3D Gaussian Splatting with a dual-scale rendering image recovery network to achieve high-quality upsampling of sparse colored point clouds.

Gaze-Guided 3D Hand Motion Prediction for Detecting Intent in Egocentric Grasping Tasks

Yufei He, Arno H. A. Stienen

Pose EstimationRobotic IntelligenceTransformerAuto EncoderVideoSequential

🎯 What it does: This paper proposes a 3D hand motion prediction method that integrates gaze information, historical hand motion sequences, and environmental object data to detect human intent in grasping tasks from the perspective of a head-mounted camera.

Gaze-Guided Task Decomposition for Imitation Learning in Robotic Manipulation

Ryo Takizawa, Y. Kuniyoshi

Robotic Intelligence

🎯 What it does: Decompose object manipulation tasks demonstrated by humans into subtasks using gaze transitions, enabling skill reuse and combination in imitation learning.

GazeTarget360: Towards Gaze Target Estimation in 360-Degree for Robot Perception

Zhu Dai, Chen Li

Robotic IntelligenceImage

🎯 What it does: Proposed a system called GazeTarget360 for estimating human gaze targets from panoramic scenes in images, integrating an eye contact detector, a pre-trained visual encoder, and a multi-scale fusion decoder.

GDM-Net++: Multi-robot 2D and 3D Gas Distribution Mapping Via Deep Q-Learning and Gaussian Process Regression

Iliya Kulbaka, Swapnoneel Roy

Robotic IntelligenceReinforcement LearningPoint Cloud

🎯 What it does: Proposed a multi-robot gas distribution mapping framework named GDM-Net++, combining Voronoi partitioning, deep Q-learning, and Gaussian process regression to achieve gas concentration mapping in 2D/3D areas.

GDTS: Goal-Guided Diffusion Model with Tree Sampling for Multi-Modal Pedestrian Trajectory Prediction

Ge Sun, Jun-Qi Ma

Autonomous DrivingDiffusion modelSequential

🎯 What it does: Proposes GDTS, a goal-guided diffusion model combined with tree sampling for multimodal pedestrian trajectory prediction.

GEAR: Gaze-Enabled Human-Robot Collaborative Assembly

Asad Ali Shahid, L. Roveda

Robotic IntelligenceImage

🎯 What it does: Developed a collaborative robot system called GEAR based on eye-tracking technology, which assists workers in grasping parts during assembly tasks and receives high-level instructions from workers; meanwhile, it was compared with traditional touchscreen interaction methods.

Generalizable and Actionable Part Detection and Manipulation with SAM-rectified Segmentation and Iterative Pose Refinement

Sucheng Qian, Cewu Lu

Object DetectionSegmentationPose EstimationRobotic IntelligenceTransformerSupervised Fine-TuningRectified FlowImage

🎯 What it does: Proposed the SAMIR framework for cross-category GAPart detection and manipulation.

Generalizable Category-Level Topological Structure Learning for Clothing Recognition in Robotic Grasping

Xin-Hai Zhu, Yixing Gao

RecognitionRobotic Intelligence

🎯 What it does: Proposes a topological structure representation and optimization strategy for category-level clothing structure feature learning, and constructs a multi-clothing classification framework based on multiple mask generation, capable of identifying clothing regions in scenes; achieves generalization to unseen clothing through learned structural features; simultaneously proposes fabric-specific grasping position estimation methods and develops a corresponding robotic grasping system that can select and grasp specified clothing according to user instructions.