IROS 2025 Papers — Page 2
IEEE/RSJ International Conference on Intelligent Robots and Systems · 1984 papers
A Tightly Coupled Inertial-Sonar Fusion for Localization of Underwater Robots
Jibo Bai, Hongfei Li
Pose EstimationRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Propose a tight-coupled inertial-forward-looking sonar fusion method for 2D localization of underwater robots.
A Two-Stage Lightweight Framework for Efficient Land-Air Bimodal Robot Autonomous Navigation
Yongjie Li, Qingquan Li
OptimizationComputational EfficiencyRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose a two-stage lightweight framework that combines global key points prediction with local trajectory refinement to achieve efficient autonomous navigation for ground-air dual-mode robots.
A Two-Stage Swarm Planning Framework for Efficient Multi-Drone Waypoint Traversal
Kailun Cui, Hao Hu
Autonomous DrivingOptimizationRobotic Intelligence
🎯 What it does: Proposed a two-stage multi-quadcopter waypoint crossing planning framework that integrates offline global trajectory generation and online distributed local trajectory planning;
A Unified Framework to Learn Collision-Free Loco-Manipulation via Adversarial Motion Priors
Huayang Yin, Zhen Kan
OptimizationRobotic Intelligence
🎯 What it does: Proposes a unified framework for achieving collision-free walking-manipulation control in unstructured real environments; the framework includes a trajectory optimization module for generating motion priorities and an MPPI-based collision-free trajectory generation and vectorized trajectory following module.
A V-shaped In-pipe Robot Capable of Drawing Route Maps for Both 3-in and 4-in Diameters using Only Low-cost Internal Sensors
Yuma Sugizaki, A. Kakogawa
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Designed and implemented a robot capable of mapping routes inside 3-inch and 4-inch pipes using internal sensors for route mapping.
A Variable Stiffness Supernumerary Robotic Limb with Pneumatic-Tendon Coupled Actuation *
Mengcheng Zhao, Youfu Li
Robotic IntelligenceReinforcement Learning
🎯 What it does: Implemented a variable stiffness redundant robotic limb (VSSRL), achieving coordinated control of position and stiffness through pneumatic-tendon coupling drive, and verified its load capacity and daily assistive functions.
A Variable-stiffness Neck Exoskeleton with Pneumatic-driven Actuators for Prolonged Head Flexion Assistance
Tianfang Li, Qiang Lin
Robotic Intelligence
🎯 What it does: Designed and implemented a variable stiffness neck exoskeleton using pneumatic-driven tension actuators, aiming to assist head movement and reduce muscular load during prolonged forward neck flexion.
A VisuoMotor Human-Robot Interaction Framework for Attention-Motion-Integrated Training
Chen Chen, Hong Cheng
Robotic IntelligenceWorld ModelMultimodality
🎯 What it does: Proposed an integrated human-computer interaction framework combining gaze-visual games and force-motion robots for attention-motor integration training, and designed a dynamic pattern recognition scheme to online identify human attention and motion states, dynamically adjusting training parameters accordingly.
A Wearable Centaur Robot with Wheel-Legged Transformation for Enhanced Load-Carrying Assistance
Songhao Li, Jian Huang
Robotic Intelligence
🎯 What it does: Developed a wearable hybrid human-animal robot with a wheel-leg convertible structure for assisting humans in long-distance load carrying.
A Whole-Body Unified Force-Impedance Control for Non-holonomic Service Robots
Moein Forouhar, Sami Haddadin
Robotic Intelligence
🎯 What it does: This paper extends the Unified Force Impedance Control (UFIC) framework to whole-body control of service mobile robots under non-homogeneous constraints, enabling robots to perform complex service tasks requiring force and impedance control within their whole-body workspace. The task space is defined as the pose of dual end-effectors, with impedance and force tracking commands applied in this space. Enhanced energy tanks are introduced to ensure system passivity, while shape functions enable smooth transitions between force tracking and impedance control. The robot's redundancy is utilized to shape the posture while satisfying joint limits, avoiding singularities and self-collisions. The effectiveness of the proposed controller was validated through simulations and experiments on the real robot GARMI during daily tasks.
A2I-Calib: An Anti-Noise Active Multi-IMU Spatial-Temporal Calibration Framework for Legged Robots
Chaoran Xiong, Ling Pei
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed and implemented A2I-Calib, a noise-robust active multi-IMU calibration framework for arbitrary foot-mounted IMU installations.
AAOPL: Automated Articulated Object Parameter Learning for Open-World Robotics
Ziyang Feng, Jianmin Ji
Robotic Intelligence
🎯 What it does: Developed the AAOPL framework, which leverages direct interaction between robots and real articulated objects to automatically learn their joint parameters, enabling the generation of precise manipulation trajectories without requiring predefined models or large amounts of demonstration data.
Abdominal Undulation with Compliant Mechanism Improves Flight Performance of a Biomimetic Robotic Butterfly
Xuyi Lian, Tiefeng Li
Robotic Intelligence
🎯 What it does: Designed, modeled, and experimentally validated a bio-inspired robotic butterfly that integrates an elastic mechanism to achieve wing-abdomen coupled motion
Absolute Localization through Vision Transformer Matching of Planetary Surface Perspective Imagery from a Digital Twin
P. Ludivig, Maciej Zurad
Data SynthesisAutonomous DrivingTransformerSupervised Fine-TuningImage
🎯 What it does: Developed a machine learning framework based on vision transformers for absolute positioning on planetary surfaces without satellite navigation.
Accelerating Focal Search in Multi-Agent Path Finding with Tighter Lower Bounds
Yimin Tang, Sven Koenig
Optimization
🎯 What it does: Proposes the Dual ECBS (DECBS) algorithm, which accelerates focus search in multi-agent path finding by first determining the maximum lower bound value and using it as a guide for best-first search.
Accelerating Inverse Kinematic Solutions for a Cable-Driven Soft Robotic Manipulator via Physics-Informed Neural Network
Rui Lin, Xiang Zhang
OptimizationRobotic Intelligence
🎯 What it does: Propose a framework based on physics-informed neural networks (PINN) that utilizes spatiotemporal data to achieve rapid inverse kinematics solutions for cable-driven soft robotic manipulators.
Accelerating Layered Manufacturing-Based 3D Printing through Optimized Non-Printing Travel-Path Planning and Infill Strategies
Liuyin Wang, Yantao Shen
OptimizationComputational Efficiency
🎯 What it does: Propose a fast 3D printing framework that improves the efficiency of 3D printers based on layered manufacturing by optimizing non-printing travel paths and filling strategies.
Accelerating Real-World Overtaking in F1TENTH Racing Employing Reinforcement Learning Methods
Emily Steiner, Henry Williams
Autonomous DrivingReinforcement Learning
🎯 What it does: Proposed a racing and overtaking agent capable of reliably driving on tracks and completing overtaking maneuvers in both simulation and real environments.
AccidentX: A Large-Scale Multimodal BEV Dataset for Traffic Accident Analysis and Prevention
Muyang Zhang, Xiaopeng Zhang
Data SynthesisAutonomous DrivingLarge Language ModelVideoMultimodalityBenchmark
🎯 What it does: Proposed and constructed AccidentX—a large-scale multi-modal bird's-eye view (BEV) dataset designed for traffic accident analysis and prevention, containing over 10,000 video clips generated using CARLA simulation with detailed annotations.
Accurate Simulation and Parameter Identification of Deformable Linear Objects using Discrete Elastic Rods in Generalized Coordinates
Qi Jing Chen, Quang-Cuong Pham
Physics Related
🎯 What it does: Implement fast and accurate simulation of deformable linear objects (DLO) in the MuJoCo physics simulator, and perform parameter identification for bending and torsional stiffness.
ACGD: Visual Multitask Policy Learning with Asymmetric Critic Guided Distillation
K. Srinivasan, Animesh Garg
Knowledge DistillationReinforcement LearningMixture of ExpertsImage
🎯 What it does: Proposes a multi-task visual manipulation policy learning framework called ACGD, which uses image inputs and integrates multiple expert policies into a single visual base policy through student-teacher distillation.
Achieving Lift-to-Weight Ratio >3.5 in Piezoelectric Direct-Driven Insect-Scale Flapping-Wing MAVs
Xiang Lu, D. Xiao
OptimizationPhysics Related
🎯 What it does: Designed and experimentally validated a new lift enhancement strategy for piezoelectric direct-driven insect-scale flapping-wing MAVs, and redesigned the X-type structure prototype.
Achieving Precise and Reliable Locomotion with Differentiable Simulation-Based System Identification
Vyacheslav Kovalev, Roman Gorbachev
Robotic IntelligenceReinforcement Learning
🎯 What it does: Integrate system identification into the reinforcement learning (RL) training loop, using differentiable simulation to estimate system parameters based solely on trajectory data.
ACoL: From Abstractions to Grounded Languages for Robust Coordination of Task Planning Robots
Yu Zhang
Robotic Intelligence
🎯 What it does: This paper studies how to automatically construct a language that maximizes flexibility while providing sufficient explanations for coordination in task-planning robots; by mapping time-state constraints into vocabulary, the paper reverse-engineers the language from the bottom up, expressing it as 'plan sketches' that only convey necessary details, maintaining implementation flexibility and achieving robust coordination with optimality guarantees.
Acoustic Neural 3D Reconstruction Under Pose Drift
Tianxiang Lin, Michael Kaess
GenerationPose EstimationNeural Radiance FieldAudio
🎯 What it does: This paper proposes an algorithm for jointly optimizing neural implicit surfaces and sonar pose to achieve 3D reconstruction under drifted poses.
ACP-MVS: Efficient Multi-View Stereo with Attention-based Context Perception
Hao Jia, Xin Yang
Depth EstimationImage
🎯 What it does: Proposed ACP-MVS, an efficient multi-view stereo network that improves pixel correspondence accuracy and reduces noise by constructing a context-aware cost volume.
Action Recognition for Underwater Gesture Communication in Human Diver and Robot Teaming
Zi-Hao Zhang, Jane Shin
RecognitionTransformerVideo
🎯 What it does: Developed an underwater diver gesture recognition algorithm based on a spatio-temporal transformer for diver-robot team collaboration.
Action Tokenizer Matters in In-Context Imitation Learning
Vuong Dinh An, Ian D. Reid
Meta LearningAuto Encoder
🎯 What it does: Systematically evaluate the performance of existing action segmenters in ICIL, finding that they fail to maintain temporal smoothness, and propose LipVQ-VAE which generates smoother action encodings by imposing Lipschitz conditions in the latent space.
Active Disturbance Rejection Control for Trajectory Tracking of a Seagoing USV: Design, Simulation, and Field Experiments
Jelmer van der Saag, Javier Alonso-Mora
Autonomous Driving
🎯 What it does: Implement an ADRC-based trajectory tracking controller on the DUS V2500, and validate its performance through custom wave and current simulations as well as port and sea field experiments.
Active Modeling and Compensation Control of Yoshimura Manipulator Using Koopman Operator
Jiaqing Qi, Jianda Han
Robotic Intelligence
🎯 What it does: This paper proposes and implements an active modeling compensation control method based on the Koopman operator and Kalman filter for the Yoshimura origami manipulator, and completes trajectory tracking experiments under load and different postures.
Active Probing with Multimodal Predictions for Motion Planning
D. Gadginmath, Jovin D'sa
Autonomous DrivingReinforcement Learning
🎯 What it does: Proposes a unified framework that integrates trajectory planning, multi-modal prediction, and active exploration to enhance decision-making in uncertain environments.
Active Prostate Phantom with Multiple Chambers
Sizhe Tian, J. Dequidt
ImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: A prostate phantom driven by pneumatic actuation with multiple independently controllable chambers was constructed to simulate different morphologies of benign prostatic hyperplasia (BPH).
Active Training Data Selection for Gaussian Process-based Robot Dynamics Learning and Control
Feng Han, Jingang Yi
Robotic IntelligenceTabular
🎯 What it does: Propose an active training data selection strategy based on Gaussian processes to reduce the training set size.
ad-trait: A Fast and Flexible Automatic Differentiation Library in Rust
Chen Liang, Daniel Rakita
Computational Efficiency
🎯 What it does: Proposed ad-trait, an automatic differentiation library implemented in Rust, which efficiently accumulates required information by overloading floating-point types and using flexible traits; supports both forward-mode and reverse-mode automatic differentiation.
Adapt-VRPD: Vehicle Routing Problem with Drones Under Dynamically Changing Traffic Conditions
N. Imran, Myounggyu Won
Optimization
🎯 What it does: Propose the Adapt-VRPD problem, design a cost model combining machine learning-based travel time prediction, and solve the collaborative delivery path planning for vehicles and drones under dynamic traffic conditions using the Variable Neighborhood Descent (VND) algorithm.
Adapting Pre-Trained Vision Models for Novel Instance Detection and Segmentation
Ya Lu, Yu Xiang
Object DetectionSegmentationTransformerImage
🎯 What it does: Propose the NIDS-Net framework for few-shot novel instance detection and segmentation
Adapting Robot’s Explanation for Failures Based on Observed Human Behavior in Human-Robot Collaboration
A. Naoum, Christian Smith
Explainability and InterpretabilityRobotic IntelligenceMultimodality
🎯 What it does: Interpreting human behavior to predict user confusion when robots fail and adaptively adjusting the level of explanation based on observed behavior
Adaptive Anomaly Recovery for Telemanipulation: A Diffusion Model Approach to Vision-Based Tracking
Haoyang Wang, Zhengxiong Li
Object TrackingAnomaly DetectionDiffusion modelVideo
🎯 What it does: Proposed a diffusion model-based teleoperation anomaly recovery framework named DET, which utilizes frame difference detection (FDD) to identify anomalies in video streams and employs diffusion models for reconstruction and replacement, thereby enhancing the continuous tracking stability of teleoperation systems under visual anomalies such as occlusion and insufficient illumination.
Adaptive Cartesian Position Control with a Switching Strategy for Robotic Manipulator with Mixed Rigid-Elastic Joints
T. Hua, Filippo Sanfilippo
Robotic Intelligence
🎯 What it does: An adaptive Cartesian position control method is proposed for robot manipulators with hybrid rigid-elastic joints, and a strategy is designed to switch between Cartesian space and joint space when the end-effector approaches the target point.
Adaptive Gaze Modulation in Social Robots: A Reinforcement Learning Approach to Attention Regulation
Nipuni H. Wijesinghe, D. Herath
Robotic IntelligenceReinforcement Learning
🎯 What it does: Developed a reinforcement learning-based gaze regulation framework for social robots, consisting of two modules: GGM and GAM.
Adaptive Invariant Extended Kalman Filter for Legged Robot State Estimation
Kyung-Hwan Kim, Dong Jin Hyun
Robotic Intelligence
🎯 What it does: Propose an Adaptive Invariant Extended Kalman Filter (AI-EKF) to improve state estimation for legged robots, adaptively adjusting the noise level of the contact foot model through online covariance estimation, thereby enhancing estimation accuracy under varying contact conditions; simultaneously replace contact sensors with a contact detection algorithm to reduce hardware dependency, and validate the method through real-world experiments on the quadruped robot LeoQuad.
Adaptive Large-Scale Novel View Image Synthesis for Autonomous Driving Datasets
Yiheng Xue, Qi Hao
GenerationData SynthesisAutonomous DrivingNeural Radiance FieldGenerative Adversarial NetworkGaussian SplattingImagePoint Cloud
🎯 What it does: Propose an adaptive pipeline suitable for large-scale outdoor traffic scenarios, constructing a high-precision 3D Surfel model and real-time synthesizing realistic novel view images.
Adaptive Manipulation using Behavior Trees
Jacques Cloete, Ioannis Havoutis
Robotic IntelligenceReinforcement LearningMultimodality
🎯 What it does: Proposes a scalable and generalizable adaptive behavior tree for robots to quickly adapt and learn from visual and non-visual observations during task execution, preventing task failure or switching to more appropriate operational strategies.
Adaptive Model-Based Control of Quadrupeds via Online System Identification using Kalman Filter
Jonas Haack, Shivesh Kumar
Robotic Intelligence
🎯 What it does: Online identification of quadruped robot mass and center of gravity using Kalman filters, dynamically adjusting model parameters to achieve better load adaptability control
Adaptive Morphing and Environmental-Phase-Transition Enables Effective Locomotion inside Granular Media
Yiliang Wang, Longchuan Li
Robotic IntelligencePhysics Related
🎯 What it does: Designed and tested a tunneling robot combining high-frequency vibration-induced environmental phase transition and adaptive deformation, achieving effective movement in granular media.
Adaptive Motion Scaling in Teleoperated Robotic Surgery based on Human Intention and Attention
Yiming Zhai, Yao Guo
Robotic IntelligenceMultimodalityBiomedical Data
🎯 What it does: Proposes a Bayesian optimization-based multidimensional adaptive motion scaling strategy to enhance operator control accuracy and operational comfort in teleoperation surgery.
Adaptive Neural Uncalibrated Visual Servo with Zero-shot Transfer of Extrinsics and Scenes
An-Jen Chen, Yue Wang
OptimizationRobotic Intelligence
🎯 What it does: Propose a structured Jacobian estimator for a neural network visual servo controller to achieve zero-shot transfer for unknown extrinsic parameters and scene scale, analyze the stability of pose error; further propose an automatic control gain scheduler to accelerate convergence; analyze the scheduling behavior through greedy optimal control; validate the method in simulation and real-world scenarios.
Adaptive Sliding Window Optimization for Multi-Modal LiDAR Inertial Odometry and Mapping
G. Han, Yu Hu
Autonomous DrivingOptimizationSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposed an uncertainty-based adaptive sliding window (ASW) strategy and implemented a multi-modal LiDAR-inertial odometry and mapping framework that integrates mechanical and solid-state LiDAR.
Adaptive Step Duration for Accurate Foot Placement: Achieving Robust Bipedal Locomotion on Terrains with Restricted Footholds
Zhaoyang Xiang, Ayonga Hereid
OptimizationRobotic Intelligence
🎯 What it does: Propose a foot placement planning algorithm based on multi-step preview, which can adaptively adjust step length and swing foot trajectory to generate feasible gaits in constrained foothold environments (e.g., stone slabs).
Adaptive Viewpoint Selection for Tomato Truss Localization via Polytope Hypotheses
Gijs van den Brandt, R. V. D. Molengraft
Robotic IntelligenceAgriculture Related
🎯 What it does: Proposed a viewpoint selection method based on the polyhedral assumption for locating tomato stems in automated harvesting.
Adaptive Visual Servoing Control Barrier Function of Robotic Manipulators with Uncalibrated Camera
Jianing Zhao, Xiang Yin
Depth EstimationOptimizationRobotic IntelligenceImage
🎯 What it does: A vision-based servoing safety control method based on control barrier functions (CBF) was developed. For uncalibrated eye-in-hand cameras, a vision-based servoing control barrier function (VS-CBF) relying solely on RGB-D images and depth data was proposed, along with the design of an adaptive estimator, baseline kinematic visual servoing control law, and quadratic programming (QP) safety controller. Experimental validation was conducted on the UR3 robotic arm.
Adaptive Visuo-Tactile Fusion with Predictive Force Attention for Dexterous Manipulation
Jinzhou Li, Hao Dong
Robotic IntelligenceMultimodality
🎯 What it does: Proposed a force-guided attention fusion module that can adaptively adjust the weights of visual and tactile features without requiring artificial labels, and introduced a self-supervised future force prediction auxiliary task to enhance the tactile modality, alleviate data imbalance, and promote correct weight adjustment.
Adaptive Wall-Following Control for Unmanned Ground Vehicles Using Spiking Neural Networks
Hengye Yang, Tao Sun
Autonomous DrivingSpiking Neural Network
🎯 What it does: Proposes an adaptive wall-following control method based on synaptic neural networks, capable of real-time fitting of unknown wall shapes and generating vehicle tracking trajectories;
AdaToken-3D: Dynamic Spatial Gating for Efficient 3D Large Multimodal-Models Reasoning
Kai Zhang, Xiaofeng Zhang
Computational EfficiencyTransformerLarge Language ModelMultimodality
🎯 What it does: Propose the AdaToken-3D framework, dynamically pruning redundant spatial tokens to improve the inference efficiency of 3D large-scale multimodal models.
Addressing Dimensional Scaling in Reinforcement Learning for Symbolic Locomotion Policies through Leveraging Inductive Priors
Rogier Fransen, Simon Hadfield
OptimizationReinforcement Learning
🎯 What it does: Explores the application of symbolic policy optimization in various legged locomotion tasks, and proposes Fast Symbolic Policy (FSP) for accelerated training and Dim-X method leveraging kinematic priors of locomotion.
Adjacent-view Transformers for Supervised Surround-view Depth Estimation
Xianda Guo, Long Chen
Depth EstimationAutonomous DrivingConvolutional Neural NetworkTransformerImage
🎯 What it does: Propose a supervised surround view depth estimation framework based on Transformer, named AVT-SSDepth, which can jointly predict depth maps from multiple cameras;
Adjusting Tissue Puncture Omnidirectionally In Situ with Pneumatic Rotatable Biopsy Mechanism and Hierarchical Airflow Management in Tortuous Luminal Pathways
Botao Lin, Hongliang Ren
Robotic IntelligenceBiomedical Data
🎯 What it does: Designed and evaluated a pneumatically driven robotic catheter equipped with a rotatable biopsy mechanism, enabling non-twisting, multi-directional in vivo biopsies within tortuous lumens.
Advancing Depth Anything Model for Unsupervised Monocular Depth Estimation in Endoscopy
Bojian Li, Fugen Zhou
Depth EstimationConvolutional Neural NetworkTransformerSupervised Fine-TuningBiomedical Data
🎯 What it does: Introduce a fine-tuning strategy for the Depth Anything Model and combine it with an unsupervised monocular depth estimation framework based on intrinsic parameters.
Advancing Learnable Multi-Agent Pathfinding Solvers with Active Fine-Tuning
A. Andreychuk, Alexey Skrynnik
OptimizationTransformerSupervised Fine-Tuning
🎯 What it does: Proposed MAPF-GPT-DDG, a multi-agent pathfinding solver fine-tuned using centralized expert data.
Advancing Object-Goal Navigation through LLM-enhanced Object Affinities Transfer
Mengying Lin, Zhaoran Wang
Autonomous DrivingRobotic IntelligenceTransformerLarge Language Model
🎯 What it does: Proposes the LLM-enhanced Object Affinity Transfer (LOAT) framework to improve object goal navigation by combining LLM semantics with learned object semantic relationships through a dual-module strategy, dynamically fusing them to enhance generalization in unknown environments.
Advancing Robot Interaction Safety: A Teleoperated Shared-Control Approach Using a Lightweight Force-Feedback Exoskeleton
Ruohan Wang, Geng Yang
Robotic Intelligence
🎯 What it does: Propose a remote shared control strategy based on a lightweight force feedback exoskeleton for robotic operations in remote healthcare scenarios.
Adversarial Attacks and Detection in Visual Place Recognition for Safer Robot Navigation
Connor Malone, Michael Milford
RecognitionSafty and PrivacyAdversarial AttackRobotic Intelligence
🎯 What it does: In robot navigation, this paper systematically analyzes the impact of two common adversarial attacks and two novel VPR-specific attacks on the localization performance of visual place recognition (VPR), and proposes a framework that integrates VPR, an attack detector (AAD), and an active navigation decision-making closed-loop.
Adversarial Augmentation for Task-Parameterized Underwater Skill Learning via Digital Twins*
Zhangpeng Tu, Canjun Yang
Data SynthesisRobotic IntelligenceGenerative Adversarial NetworkSequential
🎯 What it does: Proposed an adversarial trajectory augmentation method based on digital twin to improve the motion strategy of the task-parameterized hidden semi-Markov model (TP-HSMM).
Aerial Grasping via Maximizing Delta-Arm Workspace Utilization
Haoran Chen, Ximin Lyu
OptimizationRobotic Intelligence
🎯 What it does: Propose a drone grasping planning framework that maximizes the workspace utilization of the Delta arm, using MLP to predict feasibility probabilities and RevNet to approximate forward kinematics in order to eliminate non-convex workspace constraints.
Aerobatic Maneuver Planning for Tilt-rotor UAVs Based on Multi-Modal Consistent Dynamic Model
Hongpeng Wang, Jianda Han
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: Proposed a multi-modal consistent dynamics model based on transient CFD, and developed an adaptive feedback motion planning method utilizing third-order Bézier curves for angular velocity planning. Subsequently, the Cobra maneuver in transition mode was validated through numerical simulation, hardware-in-the-loop simulation, and outdoor flight experiments.
AeroBuoy: A Drone Deployable, 3D Printed, Autonomous Robotic Buoy for Environmental Inspection in Remote and Hazardous River Systems
Reuben O'Brien, Minas V. Liarokapis
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose a buoy system that utilizes drone deployment, is 3D-printable, and has autonomous data collection capabilities for environmental monitoring in remote or hazardous rivers.
Affine-SLAM: A closed-form solution to the landmark-SLAM via affine relaxation
Shaoran Yang, Yi Dong
OptimizationSimultaneous Localization and Mapping
🎯 What it does: Proposes a closed-form solution for the landmark SLAM problem, based on the extension of GPA with the incorporation of an affine relaxation odometry term.
Affordance-Guided Reinforcement Learning via Visual Prompting
Olivia Y. Lee, Chelsea Finn
Robotic IntelligenceReinforcement LearningPrompt EngineeringVision Language Model
🎯 What it does: Proposed the KAGI method, which utilizes vision-language models (VLM) to generate dense rewards through keypoint-based operability guidance, enabling robots to perform autonomous reinforcement learning based solely on reward signals.
AffordGrasp: In-Context Affordance Reasoning for Open-Vocabulary Task-Oriented Grasping in Clutter
Yingbo Tang, Shanghang Zhang
Pose EstimationRobotic IntelligenceVision Language ModelImageTextMultimodality
🎯 What it does: Propose the AffordGrasp framework to achieve open-vocabulary, task-oriented grasping. It uses a visual language model to perform contextual grasping capability reasoning, directly inferring tasks from implicit user instructions, identifying task-related objects, and assigning affordance labels to their components. It then generates precise grasping poses within the object's affordance regions.
Ag2x2: Robust Agent-Agnostic Visual Representations for Zero-Shot Bimanual Manipulation
Ziyin Xiong, Siyuan Huang
Representation LearningRobotic IntelligenceReinforcement LearningVideo
🎯 What it does: Propose the Ag2x2 framework, which employs coordinated visual representations to simultaneously encode object states and hand motion patterns to achieve bimanual coordination.
AGCNet: Improving Inertial Odometry via IMU Accelerometer and Gyroscope Online Compensation
H. Min, Caigui Jiang
Autonomous DrivingOptimizationConvolutional Neural NetworkSupervised Fine-TuningSimultaneous Localization and MappingTime Series
🎯 What it does: Proposed a learning-based online IMU compensation method called AGCNet to enhance the accuracy of inertial odometry.
AGENTS-LLM: Augmentative GENeration of Challenging Traffic Scenarios with an Agentic LLM Framework
Yuguang Yao, Marcel Hallgarten
Data SynthesisAutonomous DrivingTransformerLarge Language ModelAgentic AIText
🎯 What it does: Introduce the LLM-agent framework, utilizing natural language descriptions to augment real traffic scenarios
AgiBot World Colosseo: A Large-Scale Manipulation Platform for Scalable and Intelligent Embodied Systems
AgiBot-World-Contributors, Jianchao Zhu
Representation LearningData-Centric LearningRobotic IntelligenceReinforcement LearningSequentialBenchmark
🎯 What it does: Built the AgiBot World large-scale manipulation platform, containing over 1 million trajectories, 217 tasks, and 5 deployment scenarios, and introduced the general policy Genie Operator-1 (GO-1) leveraging potential action representations, while open-sourcing the dataset, tools, and models.
AirSwarm: Enabling Cost-Effective Multi-UAV Research with COTS drones
Xiaowei Li, Lihua Xie
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Developed the AirSwarm platform, enabling multi-drone collaboration using low-cost commercial drones for research and educational purposes.
AirTouch: A Low-Cost Versatile Visuotactile Feedback System for Enhanced Robotic Teleoperation
Shoujie Li, Wenbo Ding
Pose EstimationRobotic IntelligenceImage
🎯 What it does: Proposed a low-cost, multifunctional visual-tactile feedback system called AirTouch, integrating pneumatic tactile feedback with lightweight gesture pose estimation to enhance robotic remote operation.
AKF-LIO: LiDAR-Inertial Odometry with Gaussian Map by Adaptive Kalman Filter
Xupeng Xie, Boyu Zhou
Autonomous DrivingSimultaneous Localization and MappingMultimodalityPoint CloudBenchmark
🎯 What it does: Propose an adaptive Kalman filter framework that dynamically estimates time-varying noise covariance of LiDAR and IMU measurements, uses Gaussian maps to represent environmental planarity and spatial noise, and achieves accurate plane normal estimation through pseudo-merging via correlation registration, enhancing localization robustness in LiDAR failure or dynamic scenes.
AlignCAPE: Support and Query Feature Aligning for Category-Agnostic Pose Estimation
Zhuoran Chen, Jianqin Yin
Pose EstimationTransformerImageBenchmark
🎯 What it does: Proposes AlignCAPE, a two-stage pipeline for category-agnostic pose estimation, comprising a feature alignment module and a keypoint perception module.
All-in-one Defensive Network (ADNet): Trustworthy Segmentation of Complex Maritime Environments for Unmanned Surface Vessels (USVs)
Yanhong Huang, Yuanchang Liu
Object DetectionSegmentationConvolutional Neural NetworkImage
🎯 What it does: Proposed a defensive network ADNet for reliable segmentation in maritime environments under various adversarial attacks, thereby enhancing the perception reliability of unmanned surface vessels.
Along-Edge Autonomous Driving on Curvy Roads Based on Frenet Frame: A Stable Hierarchical Planning Framework
Hong-Yi Kang, YaFei Wang
Autonomous DrivingOptimization
🎯 What it does: Propose a hierarchical trajectory planning framework that integrates Cartesian and Frenet coordinate systems for edge-following on curved roads, combining a curvature-constrained optimization planner and a sampling-based lane-changing planner to achieve high-precision and stable driving.
ALVO: Adaptive Learning with Velocity Obstacles for UGV Navigation in Dynamic Scenes
Yinduo Xie, Wei Zhang
Autonomous DrivingReinforcement Learning
🎯 What it does: Proposes the ALVO adaptive learning strategy, utilizing velocity obstacles to achieve obstacle avoidance and navigation for UGVs in dynamic scenarios.
An Actionable Hierarchical Scene Representation Enhancing Autonomous Inspection Missions in Unknown Environments
V. Viswanathan, G. Nikolakopoulos
OptimizationGraph Neural NetworkMultimodality
🎯 What it does: Proposed and implemented Layered Semantic Graphs (LSG), integrating them with the FLIE multimodal task planner for autonomous inspection tasks in unknown environments, supporting real-time semantic segmentation and hierarchical path planning.
An Adaptive ROS2 Node Deployment Framework in Mobile Edge-Robot Systems
Xincheng Yang, Biao Hu
Robotic Intelligence
🎯 What it does: Developed and validated an adaptive ROS2 node deployment framework (ARDF) for dynamically deploying ROS2 nodes in mobile edge-robot systems to improve response time.
An Anytime, Scalable and Complete Algorithm for Embedding a Manufacturing Procedure in a Smart Factory
Christopher Leet, Sven Koenig
Optimization
🎯 What it does: Proposed TS-ACES, the first scalable complete algorithm capable of achieving manufacturing process embedding in smart factories with over a hundred machines.
An Easy Method for Extrinsic Calibration of Camera and Time-of-Flight Sensor
Tianyou Zhang, Xin Dong
Pose EstimationImage
🎯 What it does: Propose a simple method using a chessboard and two whiteboards for extrinsic calibration of ToF sensors and RGB cameras
An Evidence-Based Tri-Branch Cross-Pseudo Supervision Method for Semi-Supervised Medical Image Segmentation
Dongyue Li, Junzhi Yu
SegmentationConvolutional Neural NetworkBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Developed an evidence-based three-branch cross pseudo-supervised model for semi-supervised medical image segmentation.
An Improved A-Star Algorithm for Path Planning in Robot-Assisted Long Bone Fracture Reduction
Qin Gao, Ruizhi Shu
OptimizationRobotic Intelligence
🎯 What it does: An improved A* algorithm is proposed for path planning in robot-assisted long bone fracture reduction. The algorithm optimizes sampling nodes using artificial potential fields (APF), detects collisions with cylindrical bounding boxes, and smooths paths using B-splines. The method was ultimately validated on a fracture reduction robotic system, demonstrating clinically acceptable accuracy.
An Improved Flexible Hand Exoskeleton with SEA for Finger Strength Estimation and Progressive Resistance Exercise
Honghao Zheng, Ningbo Yu
Robotic Intelligence
🎯 What it does: An improved flexible hand exoskeleton equipped with a series elastic actuator (SEA) is proposed for finger force estimation and progressive resistance training;
An Inflatable Deployable Origami Grasper for Adaptive and High-Load Grasping
Peng Yan, Bing Li
Robotic Intelligence
🎯 What it does: Developed an inflatable deployable origami gripper with a rigid-flexible coupling structure, achieving multiple deployment and grasping modes under a single pneumatic actuation, and supporting self-folding.
An Inflatable Soft Robotic Manipulator with Decoupled Dual-Wrist Design for Advanced Endoscopy
Hanqi Lou, George P. Mylonas
Robotic Intelligence
🎯 What it does: Designed and evaluated an inflatable soft robotic manipulator with four degrees of freedom equipped with dual independent wrists for advanced endoscopic surgery.
An insect-scale multimodal amphibious piezoelectric robot
Le Wang, Chao Xu
Robotic Intelligence
🎯 What it does: Designed and fabricated an insect-scale amphibious robot driven by a single piezoelectric actuator.
An Intelligent Skeleton Based on Liquid Metal for Biohybrid Actuator Powered by Muscle
Xiaoqi Lu, Tao Yue
Robotic Intelligence
🎯 What it does: A smart crawling skeleton based on three-dimensional liquid metal was developed to detect and feedback on the crawling process driven by C2C12 muscles, thereby evaluating the working capability of muscle actuators and achieving high-precision control.
An Intelligent Tennis Training Robot with Timely Motion Feedback
Weiming Qu, Dingsheng Luo
Robotic Intelligence
🎯 What it does: Designed and developed a smart tennis training robot with functions including ball serving, human motion perception, and real-time motion feedback;
An Online Motion Planning Framework for Navigating Torpedo-shaped Autonomous Underwater Vehicles in Unknown Underwater Environments
Tianyou Yu, Xingjie Fu
OptimizationRobotic Intelligence
🎯 What it does: Proposes an online motion planning framework specifically designed for torpedo-shaped autonomous underwater vehicles (AUVs) to navigate in unknown three-dimensional underwater environments, employing receding horizon planning and trajectory replanning mechanisms;
An Online Optimization-Based Trajectory Planning Approach for Cooperative Landing Tasks
Jingmin Chen, Peter Eberhard
OptimizationRobotic Intelligence
🎯 What it does: Propose a real-time trajectory planning scheme for heterogeneous multi-robot systems (quadrotor drones and ground mobile robots) to accomplish collaborative landing tasks, capable of autonomously determining landing positions, landing times, and collaboration strategies among robots;
An Online Reconfiguration Strategy of the Cable-Driven Parallel Robot for pHRI via APF-Adjusted Linear Approximation
Gengxi Li, Weiwei Shang
OptimizationRobotic Intelligence
🎯 What it does: Proposes an online reconfiguration strategy for 3-DOF point mass CDPRs, adjusting cable anchor positions to enhance physical human-robot interaction performance.
An Online Terrain Classification Framework for Legged Robots Based on Fusion of Proprioceptive and Exteroceptive Sensors
Weikai Ding, Guoteng Zhang
ClassificationObject DetectionObject TrackingHyperparameter SearchRobotic IntelligenceConvolutional Neural NetworkImageMultimodality
🎯 What it does: Proposes a terrain classification framework that integrates ontological sensing with external sensing.
An Open-Source Snake Hole-Digging Inspired Safety-Critical Insertion Planning and Replanning Framework for Continuum Robots
Guanglin Ji, Zhenglong Sun
OptimizationRobotic Intelligence
🎯 What it does: Proposed a real-time re-planning continuous robot follower (FTL) motion planning framework inspired by snake burrowing.
An RGB-D Camera-Based Multi-Small Flying Anchors Control for Wire-Driven Robots Connecting to the Environment
Shintaro Inoue, Kei Okada
Robotic IntelligenceSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: This study proposes an RGB-D camera control system utilizing multiple small flying anchors, enabling a cable-driven robot to autonomously identify and connect multiple cables in non-predefined environments.
An Unsupervised C-Uniform Trajectory Sampler with Applications to Model Predictive Path Integral Control
O. G. Poyrazoglu, Volkan Isler
Optimization
🎯 What it does: Proposed an unsupervised Neural C-Uniform trajectory sampler to address scalability issues in traditional C-Uniform trajectory generation, integrated into the MPPI controller (CU-MPPI), and evaluated in simulation and real-world environments.
Analysis and Design of a Bistable Tail for a Hybrid Throwbot in a Step-Overcoming Scenario
Insung Ju, Dong-Woo Yun
Robotic Intelligence
🎯 What it does: Designed and verified a reconfigurable laminated bistable tail, enabling a hybrid throwing robot to fold the tail for storage during throwing and rigidly deploy it when actuated, to meet the dual requirements of convenient throwing and obstacle overcoming.