arXivSub Start free trial

ICRA 2025 Papers with AI Summaries

IEEE International Conference on Robotics and Automation · 1604 papers

“Hierarchy of Needs” for Robots: Control Synthesis for Compositions of Hierarchical, Complex Objectives

Ruoyu Lin, Magnus Egerstedt

Robotic Intelligence

🎯 What it does: Developed a real-time control synthesis framework for robots to achieve hierarchical complex goals.

“Oh! It's Fun Chatting with You!” a Humor-Aware Social Robot Chat Framework

Heng Zhang, Adriana Tapus

GenerationRobotic IntelligenceTransformerLarge Language ModelPrompt EngineeringTextMultimodality

🎯 What it does: Developed a humor-aware social robot chat framework that utilizes two GPT models to generate appropriate humorous dialogues based on environmental context, user profiles, and emotional states.

$\mathbf{F}^{2} \mathbf{R}^{2}$: Frequency Filtering-Based Rectification Robustness Method for Stereo Matching

Haolong Zhou, Jiamao Li

Depth EstimationImage

🎯 What it does: A robustness-enhancing method for disparity image correction based on frequency filtering (F²R²) is proposed. It removes components susceptible to correction errors in the frequency domain using a sensitive frequency filter (SFF), and restores features lost during filtering through a matching feature reconstruction module (MFRM) to improve the correction robustness of existing stereo matching networks.

$\mathcal{D}(\mathcal{R}, \mathcal{O})$ Grasp: A Unified Representation of Robot and Object Interaction for Cross-Embodiment Dexterous Grasping

Zhenyu Wei, Lin Shao

Representation LearningRobotic IntelligencePoint Cloud

🎯 What it does: Proposes a unified representation framework δ(Ρ, Ω) Grasp based on the interaction between robotic hands and objects for cross-entity multi-fingered robotic grasping.

$U^2$ Frame: A Unified and Unsupervised Learning Framework for LiDAR-Based Loop Closing

Yixin Zhang, Yulan Guo

Pose EstimationSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes U^2 Frame, a unified LiDAR loop closure framework that can simultaneously perform loop closure detection and relative pose estimation without any ground truth training data.

3D Dense Captioning via Prototypical Momentum Distillation

Jinpeng Mi, Jianwei Zhang

GenerationKnowledge Distillation

🎯 What it does: Proposes the Prototypical Momentum Distillation (PMD) method for 3D dense description, combining Momentum Distillation with uncertainty-aware Prototype-anchored Clustering to guide the student model in generating more detailed descriptions.

3D Lane Detection Based on Projection-Consistent Reference Points and Intra- & Inter-lane Context

Yiqiu Bing, Qichuan Geng

SegmentationAutonomous DrivingImage

🎯 What it does: Propose a 3D lane detection method based on projection-consistent reference points and intra- and inter-lane contexts

3D Multi-Modal Object Detection Based on Cross-Attention Feature Fusion

Sin-Ye Jhong, Yung-Yao Chen

Object DetectionAutonomous DrivingTransformerImageMultimodalityPoint Cloud

🎯 What it does: Proposes a 3D multi-modal object detection method based on cross-attention feature fusion, combined with a depth-weighted voting fusion strategy.

3D Spatial Understanding in MLLMs: Disambiguation and Evaluation

Chun-Peng Chang, Didier Stricker

Object DetectionTransformerLarge Language ModelVision Language ModelVision-Language-Action ModelMultimodality

🎯 What it does: Enhances the ability of multimodal large language models (MLLM) to locate and distinguish target objects in complex 3D environments, and provides more precise spatial instructions.

3D Whole-Body Pose Estimation Using Graph High-Resolution Network for Humanoid Robot Teleoperation

Mingyu Zhang, Yulan Guo

Pose EstimationRobotic IntelligenceConvolutional Neural NetworkGraph Neural Network

🎯 What it does: Propose the GraphHRNet framework for precise 3D full-body pose estimation, supporting teleoperation of humanoid robots.

3DSSDF: Underwater 3D Sonar Reconstruction Using Signed Distance Functions

Simon Archieri, Yvan R. Pétillot

Robotic IntelligenceSimultaneous Localization and MappingAudio

🎯 What it does: Proposed a 3D sonar reconstruction algorithm called 3DSSDF based on the signed distance function, for online localization and 3D mapping of underwater robots using forward-looking wide-aperture sonar and vehicle-integrated navigation estimation.

3DWG: 3D Weakly Supervised Visual Grounding via Category and Instance-Level Alignment

Xiaoqi Li, Yang Liu

Object DetectionVision Language ModelContrastive LearningTextMultimodalityPoint Cloud

🎯 What it does: Propose a 3D weakly supervised visual localization method that explicitly distinguishes between categories and instances. The category-level branch leverages category knowledge from pre-trained detectors and aligns it with sentence-level category features, while the instance-level branch refines candidate box features using spatial relationship descriptions in language to precisely locate objects and distinguish instances of the same category.

4DRadDet: Cluster-Queried Enhanced 3D Object Detection with 4D Radar

Caien Weng, A. Eichberger

Object DetectionPoint Cloud

🎯 What it does: Propose the 4DRadDet model for 3D object detection using 4D radar data.

6-DoF Shape Servoing of Deformable Objects in Co-Rotated Space of Modal Graph

Bohan Yang, Yunhui Liu

Robotic IntelligenceGraph Neural NetworkGraph

🎯 What it does: Proposes a method for complete 6-degree-of-freedom shape servoing of deformable objects with unknown physical geometry models in co-rotational space.

A Bio-Inspired Sand-Rolling Robot: Effect of Body Shape on Sand Rolling Performance

Xingjue Liao, Feifei Qian

Robotic IntelligencePhysics Related

🎯 What it does: Developed a sand-rolling robot inspired by the body curling movement of mountain salamanders, investigating the impact of different body shapes (hexagon, quadrilateral, triangle) on its rolling performance on sand.

A Cane-Mounted System for Dynamic Orientation Prediction for Correcting Incorrect Cane-Tapping by Visually Challenged Persons

Gagandeep Singh, Rohan Paul

Pose EstimationTime Series

🎯 What it does: Designed and implemented an AI-based embedded system capable of real-time prediction and correction of the cane's orientation, providing proportional feedback to help users self-correct their posture and recording data for remote visualization.

A Coarse-to-Fine Event-based Framework for Camera Pose Relocalization with Spatio-Temporal Retrieval and Refinement Network

Yuhang Song, Zheng Fang

Pose EstimationConvolutional Neural NetworkSimultaneous Localization and MappingTime Series

🎯 What it does: Propose a coarse-to-fine event camera pose relocalization framework, first using fine-grained temporal information to extract similar features for accurate retrieval, then refining the pose through ESPR-Net.

A Comparative Study Between a Virtual Wand and a One-to-One Approach for the Teleoperation of a Nearby Robotic Manipulator

Alexis Poignant, N. Jarrassé

Robotic Intelligence

🎯 What it does: Compared the performance of virtual cane mapping and one-to-one position mapping on a teleoperated robotic arm near the workspace.

A Comparative Study of Pulley and Bowden Transmissions in a Novel Cable-Driven Exosuit, the Stillsuit

Matthias Jammot, Chiara Basla

OptimizationRobotic IntelligenceBiomedical Data

🎯 What it does: Evaluated and compared the efficiency and responsiveness of the novel cable-pulley transmission with traditional Bowden transmission in electric exoskeletons, and developed the Stillsuit lower-limb exoskeleton based on experimental results.

A Complete and Bounded-Suboptimal Algorithm for a Moving Target Traveling Salesman Problem with Obstacles in 3D*

Anoop Bhat, H. Choset

Optimization

🎯 What it does: Proposed and implemented FMC*-TSP, a complete and bounded-suboptimal 3D algorithm for the mobile target traveling salesman problem with obstacles (MT-TSP-O), combining high-level GTSP-TW search with low-level FMC* path planning on convex set graphs.

A Comprehensive LLM-powered Framework for Driving Intelligence Evaluation

Shanhe You, Jiangtao Gong

Autonomous DrivingTransformerLarge Language ModelText

🎯 What it does: Proposed and validated an LLM-based intelligent driving behavior assessment framework, constructed a natural language assessment dataset, and verified its effectiveness through CARLA simulation and human evaluation.

A Control Barrier Function for Safe Navigation with Online Gaussian Splatting Maps

Timothy Chen, Mac Schwager

Safty and PrivacyRobotic IntelligenceNeural Radiance FieldGaussian SplattingSimultaneous Localization and Mapping

🎯 What it does: Proposed a real-time safety filter SAFER-Splat based on control barrier functions for robot navigation, which constructs detailed maps in real-time through Gaussian Splatting, and实现了 real-time GSplat mapping with safety filtering via SplatBridge; verified its safety and robustness in simulation and drone hardware platforms.

A Control Scheme for Collaborative Object Transportation between a Human and a Quadruped Robot Using the MIGHTY Suction Cup

Konstantinos Plotas, Dimitrios Papageorgiou

Robotic Intelligence

🎯 What it does: Proposes a control scheme for human-robot collaborative object transportation using a quadruped robot equipped with a MIGHTY suction cup, which can both grasp objects and measure force/torque.

A Control Strategy for an Orbital Manipulator Equipped with an External Actuator at the End-Effector

Francesco Sena, M. Stefano

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: A control strategy for an orbital manipulator equipped with an external drive module of the end-effector is proposed, which repositions the system's center of mass through reconfiguring the manipulator's posture and utilizing the end-effector's thrust mechanism during close-range operations.

A Cooperative Bearing-Rate Approach for Observability-Enhanced Target Motion Estimation

Canlun Zheng, Shiyu Zhao

Object Tracking

🎯 What it does: Proposed a new collaborative estimator STT-R to enhance the observability of visual target motion estimation

A Data-Driven Aggressive Autonomous Racing Framework Utilizing Local Trajectory Planning with Velocity Prediction

Zhouheng Li, Hongye Su

Autonomous DrivingOptimizationHyperparameter Search

🎯 What it does: Proposes a local trajectory planning and speed prediction method based on VPMPCC, learning optimal parameters through Bayesian optimization and a specially designed racing objective function OFR.

A Data-Driven Contact Estimation Method for Wheeled-Biped Robots

Bora Gökbakan, S. Caron

Robotic IntelligenceTime Series

🎯 What it does: Designed a contact estimator for wheeled bipedal robots using a Bayesian filtering framework, where the update step is learned from real robot torque measurements, and the prediction step relies on inertial measurements; this method enables real-time estimation of contact states.

A Data-Driven Velocity Estimator for Autonomous Underwater Vehicles Experiencing Unmeasurable Flow and Wave Disturbance

Jinzhi Cai, Fumin Zhang

Robotic IntelligenceRecurrent Neural NetworkTime SeriesSequential

🎯 What it does: Proposes a data-driven speed estimation method that utilizes inertial measurement units (IMU) and gated recurrent unit (GRU) neural networks to capture temporal dependencies and suppress external disturbances; subsequently integrates this estimator with an asynchronous Kalman filter to form a sensor fusion framework, enabling the fusion of onboard and external sensor information; validates the effectiveness of this method in improving speed and position estimation accuracy in confined environments with significant fluctuations and flow disturbances through experiments on a mini AUV.

A Data-Efficient Progressive Learning Framework for Robot Scooping Task

Shuai Wang, Yu Zheng

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed an advanced learning framework that first learns a data collection strategy using limited human demonstration data, then enhances this strategy as a pre-trained model within curriculum-based reinforcement learning to obtain a robot data collection strategy that generalizes across different task settings, and experimentally evaluated it on both simulated and real robots.

A Dataset and Benchmark for Shape Completion of Fruits for Agricultural Robotics

Federico Magistri, C. Stachniss

ImagePoint CloudBenchmarkAgriculture Related

🎯 What it does: Released the first publicly available 3D shape completion dataset for agricultural vision systems, providing RGB-D data and high-precision point clouds as annotations, supporting shape completion tasks for bell peppers in laboratory and greenhouse environments.

A Direct-Drive Gripper Designed by Ellipse Synthesis Across Two Output Modes

Shashank Ramesh, Mark M. Plecnik

OptimizationRobotic Intelligence

🎯 What it does: Designed and manufactured a direct-drive gripper capable of switching between two modes (sensing mode and grasping mode), utilizing elliptical synthesis to optimize the Jacobian for improving the trade-off between force estimation and grasping force performance.

A Fairness-Oriented Control Framework for Safety-Critical Multi-Robot Systems: Alternative Authority Control

Lei Shi, Xiong Li

Robotic Intelligence

🎯 What it does: Proposed a fairness-oriented multi-robot control framework that combines Alternative Authority Control (AAC) and Flexible Control Barrier Function (F-CBF), achieving dynamic allocation of control authority to enhance system collaboration and safety.

A Full-Cycle Assembly Operation: From Digital Planning to Trajectory Execution Using a Robotic Arm

Dror Livnat, Dan Halperin

Robotic Intelligence

🎯 What it does: Proposed an end-to-end tight assembly operation framework, taking a digital model as input and generating a complete execution plan (trajectory placement and grasping), with physical demonstrations completed on the UR5e robotic arm.

A Full-Optical Pretouch Dual-Modal and Dual-Mechanism (PDM2) Sensor for Robotic Grasping

Cheng Fang, Jun Zou

Robotic IntelligenceUltrasound

🎯 What it does: Designed, manufactured, and characterized an all-optical pre-contact dual-mode dual-mechanism (PDM2) sensor based on a gas-coupled fiber tip surface micro-electro-mechanical opto-acoustic ultrasound transducer (SMOUT)

A Generalized Control Revision Method for Autonomous Driving Safety

Zehan Zhu, Jianqiang Wang

Autonomous Driving

🎯 What it does: Propose a general control revision method that utilizes vectorized perception and occupancy grid map inputs to correct autonomous driving safety;

A Haptic Feedback Device Actuated by Electromagnetic Torque

Xionghuan Luo, Hongbin Liu

Physics Related

🎯 What it does: Propose an electromagnetic haptic feedback device based on magnetic torque to achieve real-time tactile perception in a large workspace.

A Helping (Human) Hand in Kinematic Structure Estimation

Adrian Pfisterer, Oliver Brock

Pose EstimationRobotic IntelligenceVideo

🎯 What it does: Propose a probabilistic real-time method that utilizes human hands as a prior to online estimate the kinematic model of objects under visual uncertainty.

A Hierarchical Graph-Based Terrain-Aware Autonomous Navigation Approach for Complementary Multimodal Ground-Aerial Exploration

Akash Patel, G. Nikolakopoulos

Robotic IntelligenceGraph Neural NetworkSimultaneous Localization and MappingGraph

🎯 What it does: Propose a terrain-aware autonomous navigation method based on hierarchical graphs, cooperating ground and aerial robots for exploration.

A Highly Robust Contact Sensor for Precise Contact Detection of Fabric

Zhengrong Ling, Yajing Shen

Robotic Intelligence

🎯 What it does: Developed a contact sensor with high robustness and high resolution for fabric contact force;

A Human-in-the-Loop Simulation Framework for Evaluating Control Strategies in Gait Assistive Robots

Yifan Wang, Wei Tech Ang

Robotic IntelligenceReinforcement Learning from Human Feedback

🎯 What it does: Proposed a human-in-the-loop (HITL) simulation framework for gait-assisting robots to evaluate control strategies.

A Hybrid Approach to Indoor Social Navigation: Integrating Reactive Local Planning and Proactive Global Planning

Arnab Debnath (George Mason University), Jana Kosecka

Robotic IntelligenceReinforcement LearningBenchmark

🎯 What it does: Proposes a modular indoor social navigation framework that combines classical global planning with deep reinforcement learning (DRL) local planning, enabling rapid goal-reaching while avoiding collisions with people in crowded and confined environments.

A Hybrid User Interface Combining AR, Desktop, and Mobile Interfaces for Enhanced Industrial Robot Programming

Jan Krieglstein, Michael Sedlmair

Robotic Intelligence

🎯 What it does: Proposed a hybrid user interface combining AR, desktop, and mobile devices for subtask selection in industrial robot programming.

A Kinematics Optimization Framework with Improved Computational Efficiency for Task-Based Optimum Design of Serial Manipulators in Cluttered Environments

Nikolay Petkov, R. Skilton

OptimizationComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposes an optimization framework capable of performing task-driven optimal kinematic design for serial manipulators in cluttered environments, which simultaneously optimizes link dimensions and joint topology through a generic parameterized kinematic model;

A Large-Scale Dataset for Humanoid Robotics Enabling a Novel Data-Driven Fall Prediction

Oliver Urbann, Alice Kirchheim

Anomaly DetectionRobotic IntelligenceConvolutional Neural NetworkTime SeriesSequential

🎯 What it does: Collected 37.9 hours of robot sensor data and proposed a model called RePro-TCN to predict humanoid robot falls.

A Learning Framework for Diverse Legged Robot Locomotion Using Barrier-Based Style Rewards

Gijeong Kim, Hae-Won Park

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed and implemented a model-free reinforcement learning framework that supports various locomotion modes, including quadrupedal, tripodal, and bipedal walking, and completed multiple tasks on the KAIST HOUND robot.

A Light-Weight Framework for Open-Set Object Detection with Decoupled Feature Alignment in Joint Space

Yonghao He, Song Liu

Object DetectionConvolutional Neural NetworkTransformerVision Language ModelMultimodality

🎯 What it does: Proposed a lightweight open-set object detection framework called DOSOD, combining YOLO-World with a visual language model and an MLP adapter to align cross-modal features in a joint space, enhancing real-time performance.

A Magnetic-Actuated Vision-Based Whisker Array for Contact Perception and Grasping

Zhixian Hu, Yu She

Robotic IntelligenceVideo

🎯 What it does: Designed and evaluated a vision-based magnetic actuation whisker array sensor for tactile perception and fine grasping.

A Map-Free Deep Learning-Based Framework for Gate-to-Gate Monocular Visual Navigation Aboard Miniaturized Aerial Vehicles

Lorenzo Scarciglia, D. Palossi

Autonomous DrivingComputational EfficiencyRobotic IntelligenceConvolutional Neural NetworkImage

🎯 What it does: Proposed a map-agnostic gated visual navigation framework based on a monocular camera, integrating a real-time deep learning door detection frontend and a visual servo control backend for ultra-lightweight drones.

A Method for Constructing Building Structure Grid Map Based on a Climbing Algorithm

Xidong Zhou, Yaonan Wang

OptimizationRobotic Intelligence

🎯 What it does: Propose a building structure grid map construction method based on the hill climbing algorithm, which enables autonomous navigation for drone-ground amphibious robots in indoor environments.

A Modeling and Control Strategy for the Gaze-Guided Teleoperation of Robotic Manipulators via Smart Glasses

A. Lawson, H. Saeidi

Robotic IntelligenceVision-Language-Action ModelOptical FlowImage

🎯 What it does: Developed a model-based pupil tracking filtering and control strategy for directly guiding a robot arm to manipulate objects through smart glasses, with feasibility validated via a case study involving six participants.

A Modified Resistance Model for Magnetic Honeycomb Robots to Navigate in Low Reynolds Number Fluids

Leyao Zou, Zhongxue Gan

Robotic IntelligencePhysics RelatedOrdinary Differential Equation

🎯 What it does: Designed and manufactured the H-robot, adopting a honeycomb porous spherical structure, proposed a dynamic model under low Reynolds numbers, implemented motion control based on retreat sliding mode control, and completed magnetic field-driven tracking experiments and maze path planning.

A Neural Network-Based Framework for Fast and Smooth Posture Reconstruction of a Soft Continuum Arm

Tixian Wang, Prashant G. Mehta

Pose EstimationComputational Efficiency

🎯 What it does: Designed and experimentally validated a neural network-based framework for estimating the shape of soft continuous arms from noisy pose measurements at limited positions.

A New Framework for Repetitive Control of Robot Manipulators via Operator-Theoretic Robust Stabilization

G. Song, Jung Hoon Kim

Robotic Intelligence

🎯 What it does: A robust stability framework based on operator theory was established to achieve repetitive control for uncertain robotic manipulators; by using inverse dynamics, the nonlinear input/output behavior was transformed into LTI equations, followed by constructing operator theory methods to prove that the necessary and sufficient condition for exponential stability of the closed-loop system is the spectral radius of the corresponding single-track operator being less than 1. Based on this, control parameter selection criteria were provided, and finally, experiments validated the effectiveness of this framework.

A New Semidefinite Relaxation for Linear and Piecewise-Affine Optimal Control with Time Scaling

Lujie Yang, Russ Tedrake

Optimization

🎯 What it does: Proposed a semi-definite relaxation method for optimal control of linear systems under time scaling, and extended it to piecewise affine (PWA) systems; by converting the PWA optimal control problem into a shortest path problem in a convex set graph, and integrating relaxation for unified solution.

A New Stereo Fisheye Event Camera for Fast Drone Detection and Tracking

D. Costa, Fabio Morbidi

Object DetectionObject Tracking

🎯 What it does: A novel dual fisheye event camera stereo camera SFERA was developed and evaluated for rapid drone detection and tracking, providing a 360° panoramic view and employing a Bayesian estimator for real-time target tracking.

A New Variable-Gain Sliding Mode Filter and Its Application to Velocity Filtering

Myo Thant Sin Aung, Haoyong Yu

Ordinary Differential Equation

🎯 What it does: Proposed a new variable gain sliding mode filter and applied it to speed filtering

A Novel Computational Framework of Robot Trust for Human-Robot Teams

Bhavana Nare, Kyle Johnsen

Robotic Intelligence

🎯 What it does: Proposed a novel trust model based on the probability of positive human experiences with robots, and constructed a dynamic trust-building process; simultaneously introduced the concept of kinship to explore its impact on trust and distrust, and presented an algorithm for decision-making utilizing kinship-mediated trust; verified the framework's feasibility in real-world human-robot collaboration scenarios through simulation tasks.

A Novel Control Strategy for Offset Points Tracking in the Context of Agricultural Robotics

Stephane Ngnepiepaye Wembe, R. Lenain

Robotic IntelligenceAgriculture Related

🎯 What it does: Proposes a new method for controlling fixed points on agricultural machinery tools and presents two implementation schemes

A Novel Decomposed Feature-Oriented Framework for Open-Set Semantic Segmentation on LiDAR Data

Wenbang Deng, Huimin Lu

SegmentationAnomaly DetectionAutonomous DrivingPoint Cloud

🎯 What it does: Proposed a feature-based decomposed dual-decoder framework to achieve open-set semantic segmentation for LiDAR data.

A Novel Dynamic Motion Primitives Framework for Safe Human-Robot Collaboration

Andrea Pupa, Cristian Secchi

OptimizationSafty and PrivacyRobotic Intelligence

🎯 What it does: Proposed a new dynamic motion primitive framework for achieving safe human-robot collaboration

A Novel Hybrid Hysteresis Modeling Method for Multiloop-Asymmetry Hysteresis Behavior of Nonlinear Compliant Actuators

Libo Zhou, S. Bai

Physics Related

🎯 What it does: A new hybrid hysteresis model is proposed, decoupling the nonlinear reference line from the symmetric hysteresis loop, and separately fitting them using a power function and the Maxwell slip model; the multi-loop asymmetric hysteresis behavior of nonlinear compliant actuators is modeled and experimentally validated.

A Novel Telelocomotion Framework with CoM Estimation for Scalable Locomotion on Humanoid Robots

Anne He, K. Hamed

Robotic Intelligence

🎯 What it does: Integrate the ForceBot walking interface with the small humanoid robot HECTOR V2, using ForceBot to estimate the operator's center of mass (CoM) trajectory as the robot's tracking reference, and implement scaled trajectory tracking on the robot side using model predictive control (MPC) based on a simplified single rigid body model.

A Novel Tendon-Driven Articulated Continuum Robot with Stabilized Self-Locking Joints

Jiankun Ren, Yunquan Sun

Robotic Intelligence

🎯 What it does: Designed and implemented a tendon-driven continuous joint robot with stable self-locking joints

A Novel Twisted-Winching String Actuator for Robotic Applications: Design and Validation

Ryan Poon, Ian W. Hunter

Robotic Intelligence

🎯 What it does: Proposed a novel actuator system that combines torsion spring actuators (TSA) with a winch mechanism to address the limitations of traditional TSA in terms of travel range and force transmission ratio;

A Novel Under-Actuated Gripper Based on Passive-Locking Mechanism for Stable Gripping Under Environmental Constraints

Seok-Hui Yang, Woosung Yang

Robotic Intelligence

🎯 What it does: Proposed a two-finger joint non-driven gripper based on a passive locking mechanism, achieving stable grasping by combining three phalanges with four-bar and eight-bar linkages.

A Novel Underwater Robot with Carangiform Locomotion Achieved via Single Degree of Actuation and Magnetically Transmitted Traveling Wave

G. Manduca, Cesare Stefanini

Robotic Intelligence

🎯 What it does: Proposed a fish-shaped robot that utilizes a magnetic transmission system to convert the rotation of a single motor into phase-shift oscillations of multiple modules, enabling propulsion through traveling waves in water;

A Parameter-Efficient Tuning Framework for Language-Guided Object Grounding and Robot Grasping

Houjian Yu, Changhyun Choi

SegmentationRobotic IntelligenceTransformerVision Language ModelContrastive LearningMultimodality

🎯 What it does: Proposes a CLIP-based multimodal parameter-efficient fine-tuning framework for three tasks: language-guided target localization segmentation, grasp synthesis, and grasp feasibility assessment.

A Piezoresistive Printable Strain Sensor for Monitoring and Control of Soft Robotic Links

Claudia Sánchez, C. A. Monje

Robotic Intelligence

🎯 What it does: Designed, manufactured, and electromechanically evaluated a 3D-printed flexible strain sensor for monitoring and controlling soft mechanical linkages.

A Quantum Annealing Approach to Target Tracking

Michel Barbeau, Houman Masnavi

Object TrackingOptimization

🎯 What it does: Proposes a 3D object tracking algorithm based on the Constraint Quadratic Model (CQM), applicable to quantum computers for real-time target tracking planning

A Real-to-Sim-to-Real Approach to Robotic Manipulation with VLM-Generated Iterative Keypoint Rewards

Shivansh Patel, Yunzhu Li

Domain AdaptationRobotic IntelligenceReinforcement LearningVision Language ModelImageText

🎯 What it does: Proposes an Iterative Key Point Reward (IKER) framework based on a vision-language model (VLM) to enable robotic multi-step manipulation tasks in a real-simulate-real loop;

A Refined 3D Gaussian Representation for High-Quality Dynamic Scene Reconstruction

Bin Zhang, Zexin Peng

Gaussian SplattingPoint Cloud

🎯 What it does: Proposed a refined 3D Gaussian representation method for high-quality dynamic scene reconstruction

A Reinforcement Learning-Based Social Robot for Personalized Learning in Children with Autism

Farzaneh Askari, M. Mahoor

Robotic IntelligenceReinforcement Learning

🎯 What it does: Developed a reinforcement learning-based active learning method for personalized human-computer interaction, teaching imitation skills to children with autism and adapting instruction based on their performance and preferences

A Robust Deep Reinforcement Learning Framework for Image-Based Autonomous Guidewire Navigation

Sangbaek Yoo, D. Chang

Autonomous DrivingReinforcement LearningImage

🎯 What it does: Proposed a robust deep reinforcement learning framework for image-based catheter autonomous navigation.

A Simple Dynamics Model for Cable Driven Continuum Robots with Actuator Coupling

Connor Watson, Tania K. Morimoto

Robotic Intelligence

🎯 What it does: Propose a novel simplified dynamic model for cable-driven continuous robots (CDCR) and validate its effectiveness through hardware experiments.

A Skeleton-Based Topological Planner for Exploration in Complex Unknown Environments

Haochen Niu, Peilin Liu

OptimizationRobotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Propose an autonomous exploration framework based on skeleton topological graphs, utilizing global topological information to improve exploration efficiency and reduce computational costs

A Stochastic Cloning Square-Root Information Filter with Accurate Feature Tracking for Visual-Inertial Odometry

Deshun Hu

Simultaneous Localization and MappingOptical Flow

🎯 What it does: Propose an improved square root information filter, integrating random cloning, single-precision numerics, fast Mahalanobis distance, block Householder triangularization, keyframe strategy, and affine optical flow techniques to enhance the accuracy and speed of visual-inertial odometry.

A Synergistic Framework for Learning Shape Estimation and Shape-Aware Whole-Body Control Policy for Continuum Robots

M. Kasaei, Mohsen Khadem

Robotic IntelligencePhysics RelatedOrdinary Differential Equation

🎯 What it does: Proposes a collaborative learning framework for shape estimation and shape-aware whole-body control strategies for tendon-driven continuous robots;

A System for Endoscopic Submucosal Dissection Featuring Concentric Push-Pull Manipulators

Peter Connor, Robert J. Webster

Robotic IntelligenceBiomedical Data

🎯 What it does: Proposes a system that inserts two small flexible robotic manipulators into a clinically used colonoscope for endoscopic submucosal dissection (ESD);

A Tugging Controller that Maximizes Lateral Resistive Force by Mounding Sandy Terrain

Deaho Moon, Hannah Stuart

OptimizationRobotic Intelligence

🎯 What it does: Propose an anchoring method that uses a winch system to create sand piles in front of a mobile agent, and design a controller to capture the maximum lateral sand pile resistance in real-time.

A Unified End-to-End Network for Category-Level and Instance-Level Object Pose Estimation from RGB Images

Jiale Ren, Peifeng Jiang

Pose EstimationImagePoint Cloud

🎯 What it does: Proposes an end-to-end unified network called CIPE for category-level and instance-level 6-DoF pose estimation from RGB images.

A Variable Stiffness and Transformable Entanglement Soft Robotic Gripper

Huayu Zhang, Xin Ma

Robotic Intelligence

🎯 What it does: Proposed and experimentally verified a soft gripper with variable stiffness and convertible grasp modes (wrapping and clamping), capable of handling objects with various shapes and weights;

A Virtual Gravity Controller for Efficient Underactuated Biped Robots

Despoina Maligianni, E. Papadopoulos

Robotic Intelligence

🎯 What it does: A virtual gravity controller for low-driven bipedal robots is proposed, and through design modifications, the robot can reproduce passive gait on flat ground.

A Visual Servo System for Robotic on-Orbit Servicing Based on 3D Perception of Non-Cooperative Satellite

Panpan Zhao, Xueying Qin

Object TrackingPose EstimationRobotic IntelligencePoint Cloud

🎯 What it does: Propose a non-cooperative satellite visual servo system based on 3D perception, combining reconstruction and tracking to achieve shape and pose estimation under orbital conditions.

A2DO: Adaptive Anti-Degradation Odometry with Deep Multi-Sensor Fusion for Autonomous Navigation

Hui Lai, Jian Pu

Autonomous DrivingConvolutional Neural NetworkSupervised Fine-TuningImageMultimodalityPoint Cloud

🎯 What it does: Propose A2DO, an end-to-end odometry system that leverages deep multi-sensor fusion to enhance robustness in low-light, adverse weather, or sensor degradation scenarios.

Accelerated Quasi-Static FEM for Real-Time Modeling of Continuum Robots with Multiple Contacts and Large Deformation

Hao Chen, Hongbin Liu

Computational EfficiencyRobotic Intelligence

🎯 What it does: Propose Acc-FEM for real-time modeling of continuous robots with multiple contact points and large deformations.

Accurately Modeling the Output Torque and Stiffness of Ankle-Foot Orthoses with a Compliant Linkage Model

David J. Lam, Elliott J. Rouse

Time SeriesBiomedical Data

🎯 What it does: A compliant linkage model incorporating compliant components was established to accurately predict the output torque and stiffness of ankle-foot orthoses, and to re-estimate the spring stiffness previously overlooked in published studies.

Achieving Human Level Competitive Robot Table Tennis

David B. D'Ambrosio, Pannag R. Sanketi

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a table football robot capable of competing against humans in real environments, achieving amateur-level performance.

Acoustic Wave Manipulation Through Sparse Robotic Actuation

Tristan A. Shah, Stas Tiomkin

Robotic IntelligencePhysics RelatedAudio

🎯 What it does: Developed an efficient data-driven method enabling robots to focus or suppress scattered acoustic energy through spatially sparse actuators, thereby achieving acoustic wave manipulation.

ACROSS: A Deformation-Based Cross-Modal Representation for Robotic Tactile Perception

W. Z. E. Amri, Nicolás Navarro-Guerrero

Representation LearningRobotic IntelligenceImageMesh

🎯 What it does: Propose the ACROSS framework, which converts data between different tactile sensors by utilizing sensor deformation information;

Active Illumination for Visual Ego-Motion Estimation in the Dark

Francesco Crocetti, P. Valigi

Pose EstimationConvolutional Neural NetworkSimultaneous Localization and Mapping

🎯 What it does: Proposed an active illumination framework to enhance the performance of visual odometry and visual SLAM in low-light or dark environments;

Active Semantic Mapping with Mobile Manipulator in Horticultural Environments

José Cuarán, Girish Chowdhary

Robotic IntelligenceSimultaneous Localization and MappingImagePoint CloudAgriculture Related

🎯 What it does: Proposes an efficient and scalable active semantic mapping method for horticultural environments, utilizing a mobile robot gripper and RGB-D camera. The method employs probabilistic semantic maps to detect semantic targets, generate candidate viewpoints, and calculate information gain, while introducing an efficient ray casting strategy and a novel information utility function.

Active Vision Might Be All You Need: Exploring Active Vision in Bimanual Robotic Manipulation

Ian Chuang, Iman Soltani

Robotic IntelligenceVideo

🎯 What it does: Researched and implemented the AV-ALOHA dual-arm robot manipulation system utilizing active vision assistance, combining human demonstrations for imitation learning to dynamically adjust camera perspectives and enhance environmental perception and task execution.

Actor-Critic Cooperative Compensation to Model Predictive Control for Off-Road Autonomous Vehicles Under Unknown Dynamics

Prakhar Gupta, Yunyi Jia

Autonomous DrivingOptimizationReinforcement Learning

🎯 What it does: Proposed a model predictive controller (AC^3 MPC) based on Actor-Critic cooperative compensation to address unknown system dynamics;

AdaptBot: Combining LLM with Knowledge Graphs and Human Input for Generic-to-Specific Task Decomposition and Knowledge Refinement

Shivam Singh, Madhava Krishna

Robotic IntelligenceTransformerLarge Language ModelGraph

🎯 What it does: Achieved task decomposition and knowledge refinement from general to specific by combining the general predictions of large language models (LLMs) with domain knowledge graphs (KGs) and human input when necessary, enabling robots to quickly adapt to new tasks.

Adapting Gait Frequency for Posture-Regulating Humanoid Push-Recovery via Hierarchical Model Predictive Control

Junheng Li, Quan Nguyen

OptimizationRobotic Intelligence

🎯 What it does: Adapt the gait frequency through hierarchical model predictive control (Hierarchical MPC) to enhance the push recovery performance of humanoid robots and achieve posture adjustment.

Adaptive Ankle-Foot Prosthesis with Passive Agonist-Antagonist Design

Matteo Crotti, M. Catalano

Robotic Intelligence

🎯 What it does: Designed and tested an agonist-antagonist structure ankle joint and adaptive sole for passive prosthetic feet, capable of generating torque similar to a normal ankle joint without external power, and adaptively changing the sole shape according to different terrains;

Adaptive Compliance Policy: Learning Approximate Compliance for Diffusion Guided Control

Yifan Hou, Shuran Song

Robotic IntelligenceDiffusion model

🎯 What it does: Propose Adaptive Compliance Policy (ACP), which learns to dynamically adjust the compliance of robotic systems in space and time from human demonstrations.

Adaptive Concertina Locomotion of a Robotic Snake Through Narrow Uncertain Channels

Jit Koley, H. Pillai

Robotic Intelligence

🎯 What it does: The study simulates the concertina movement of biological snakes in narrow channels with uncertain widths using a multi-chain planar snake robot, and proposes a reference trajectory generation algorithm and its modified version for straight channels.

Adaptive Deadlock Avoidance for Decentralized Multi-Agent Systems via CBF-Inspired Risk Measurement

Yanze Zhang, Wenhao Luo

Optimization

🎯 What it does: Propose a decentralized framework that unifies CLF and CBF, utilizing auxiliary CBF and CBF-inspired risk measurement to detect deadlocks in multi-agent systems and provide adaptive activation solutions.

Adaptive Distance Functions via Kelvin Transformation

Rafael I. Cabral Muchacho, Florian T. Pokorny

Physics Related

🎯 What it does: Proposed and implemented a semantic-aware distance function and its numerical computation method based on Kelvin transformation