arXivSub Start free trial

ICRA 2025 Papers — Page 7

IEEE International Conference on Robotics and Automation · 1604 papers

Generalizable Zero-Shot Object Pose Estimation for Bin-Picking

Zijian Zhang, S. Serikawa

Pose EstimationImageBenchmark

🎯 What it does: Proposes a zero-shot 6D pose estimation method that achieves accurate pose estimation without retraining.

Generalized Mission Planning for Heterogeneous Multi-Robot Teams via LLM-Constructed Hierarchical Trees

Piyush Gupta, Sangjae Bae

OptimizationRobotic IntelligenceLarge Language Model

🎯 What it does: Proposed a task planning strategy for heterogeneous multi-robot teams, systematically decomposing complex tasks using a hierarchical tree structure, and constructing the hierarchical tree through a large language model (LLM); subsequently further decomposing to generate optimal scheduling that meets the constraints and capabilities of each robot.

Generalizing Motion Planners with Mixture of Experts for Autonomous Driving

Qiao Sun, Hang Zhao

Autonomous DrivingTransformerMixture of Experts

🎯 What it does: Review and benchmark existing learning-based motion planning methods, and propose a scalable decoder-style motion planner, StateTransformer-2 (STR2), which integrates a Vision Transformer encoder with a mix-of-experts (MoE) causal Transformer architecture.

Generating Causal Explanations of Vehicular Agent Behavioural Interactions with Learnt Reward Profiles

Rhys Howard, L. Kunze

Autonomous DrivingExplainability and InterpretabilityReinforcement Learning

🎯 What it does: Learning the weights of reward metrics to achieve causal explanations for vehicle agent interactions

Generating Diverse Challenging Terrains for Legged Robots Using Quality-Diversity Algorithm

Arthur Esquerre-Pourtère, Jaeheung Park

Data SynthesisRobotic IntelligenceReinforcement Learning

🎯 What it does: Generate diverse and challenging unstructured terrains using the Quality-Diversity framework to test and uncover weaknesses in legged robot controllers.

Generating Out-of-Distribution Scenarios Using Language Models

Erfan Aasi, Daniela Rus

GenerationData SynthesisAutonomous DrivingTransformerLarge Language ModelVision Language ModelText

🎯 What it does: Propose and implement a generative framework based on large language models (LLMs), generating diverse out-of-distribution (OOD) driving scenarios by constructing a branch tree, automatically aligning text descriptions with scene enhancements in the CARLA simulator, and subsequently conducting simulation verification;

Generative-AI-Driven Jumping Robot Design Using Diffusion Models

Byungchul Kim, Daniela Rus

Robotic IntelligenceDiffusion model

🎯 What it does: Generating directly manufacturable 3D robot structures using diffusion models and optimizing embedded vectors to find the best design

GenTact Toolbox: A Computational Design Pipeline to Procedurally Generate Context-Driven 3D Printed Whole-Body Artificial Skins

Carson Kohlbrenner, Alessandro Roncone

GenerationRobotic IntelligenceMesh

🎯 What it does: Proposed and implemented GenTact Toolbox, a computationally generative process for creating full-body tactile skins that can be customized according to robot morphology and application scenarios.

Geometric Design and Gait Co-Optimization for Soft Continuum Robots Swimming at Low and High Reynolds Numbers

Yanhao Yang, Ross L. Hatton

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: Proposes a co-optimized framework for the design and control of soft continuum robots based on geometric kinematic analysis, achieving more efficient swimming at low to high Reynolds numbers.

Geometry and Force-Informed Robotic Assembly with Small Relative Initial Deviations for Circular Electrical Connectors

Zhenyu Wang, Han Ding

Robotic IntelligencePhysics Related

🎯 What it does: A robot assembly strategy for circular electrical connectors (CEC) with small relative initial deviation is proposed, with the core design focusing on search trajectories and heuristic mechanics strategies to perceive force/pose (F/P) discontinuity characteristics under different geometric constraints.

Geometry-Aware Volumetric Data Stitching Using Local Surface Mapping and Robot Optical Coherence Tomography

Guangshen Ma, Mark Draelos

OptimizationRobotic IntelligencePoint CloudBiomedical DataComputed Tomography

🎯 What it does: Studied the integration of OCT with a 6-DOF robot based on three targets and a geometric-aware voxel stitching framework, aiming to achieve large-area OCT scanning.

GERA: Geometric Embedding for Efficient Point Registration Analysis

Geng Li, Jianfei Yang

Pose EstimationComputational EfficiencyPoint Cloud

🎯 What it does: Proposes a fully MLP-based architecture for point cloud registration, utilizing offline-constructed geometric encoding instead of traditional complex feature extractors, significantly reducing computational and memory costs while improving inference speed and resource utilization efficiency.

Gesturing Towards Efficient Robot Control: Exploring Sensor Placement and Control Modes for Mid-Air Human-Robot Interaction

Tonia Mielke, Christian Hansen

Robotic Intelligence

🎯 What it does: Systematically compared the effects of position control and rate control modes under different sensor placement configurations in aerial human-robot interaction.

GET-Zero: Graph Embodiment Transformer for Zero-Shot Embodiment Generalization

Austin Patel, Shuran Song

Robotic IntelligenceTransformer

🎯 What it does: Introduces GET-Zero, an embodiment-aware control strategy architecture and training process that can immediately adapt to new hardware changes without retraining.

GHIL-Glue: Hierarchical Control with Filtered Subgoal Images

Kyle Hatch, Benjamin Burchfiel

Robotic IntelligenceImageVideoBenchmark

🎯 What it does: Proposes the GHIL-Glue method, which filters non-progressive subgoals and enhances the robustness of low-level goal-conditioned policies based on generated subgoals.

GMF: Gravitational Mass-Force Framework for Parametric Multi-Level Coordination in Multi-Robot and Swarm Robotic Systems

Michael Starks, Ramviyas Parasuraman

Robotic IntelligencePhysics Related

🎯 What it does: Proposed a parameterized multi-layer coordination framework based on a gravitational field (GMF) for distributed cooperative control in multi-robot and swarm robot systems.

Gnd: Global Navigation Dataset With Multi-Modal Perception and Multi-Category Traversability in Outdoor Campus Environments

Jing Liang (George Mason University), Xuesu Xiao (George Mason University)

Robotic IntelligenceImageMultimodalityPoint CloudBenchmark

🎯 What it does: Proposed and made publicly available the Global Navigation Dataset (GND), which integrates multi-modal sensing data (3D LiDAR point clouds, RGB, and 360° images) along with multi-class traversability maps, and demonstrated the dataset's feasibility in applications such as global robot navigation, map-free navigation, and global location recognition.

Goal-Driven Robotic Pushing Manipulation Under Uncertain Object Properties

Yongseok Lee, Keehoon Kim

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes a goal-driven precise robotic pushing framework based on model predictive path integral (MPPI), which can complete pushing tasks under uncertain object properties.

Goal-Guided Reinforcement Learning: Leveraging Large Language Models for Long-Horizon Task Decomposition

Ceng Zhang, G. Chirikjian

TransformerLarge Language ModelReinforcement Learning

🎯 What it does: Proposes a goal-based reinforcement learning framework that utilizes a large language model (LLM) to first generate subgoals and plan action sequences, forming a hierarchical process of task decomposition and strategy generation, followed by guiding RL to gradually complete each subgoal through these LLM-generated strategies;

GPU-Accelerated Subsystem-Based ADMM for Large-Scale Interactive Simulation

Harim Ji, Dongjun Lee

OptimizationComputational Efficiency

🎯 What it does: Implemented a GPU-accelerated subsystem-based ADMM algorithm for interactive simulation

GRACE: Generating Socially Appropriate Robot Actions Leveraging LLMs and Human Explanations

Fethiye Irmak Doğan, Hatice Gunes

Robotic IntelligenceTransformerLarge Language ModelText

🎯 What it does: By integrating common sense knowledge from large language models with human explanations, generating robot behaviors that conform to social norms and providing explanations for human-specified actions.

Gradient Guided Search for Aircraft Contingency Landing Planning

H. E. Tekaslan, Ella M. Atkins

OptimizationTabular

🎯 What it does: A three-dimensional discrete search path planner is proposed for emergency landing planning of fixed-wing aircraft, which utilizes cost gradients to ensure the descent flight path angle aligns with the runway direction, while considering the maximization of flight envelope boundary margins under steady wind conditions; robust solutions are achieved through a multi-objective cost function combined with gradient guidance and population risk measures.

Gradient-Based Adversarial Attacks on Deep LiDAR Odometry

Zhenbo Song, Weiqing Li

Pose EstimationAutonomous DrivingAdversarial AttackPoint Cloud

🎯 What it does: Proposed an adversarial attack method based on gradient optimization targeting deep LiDAR localization networks.

Gradient-Based Trajectory Optimization with Parallelized Differentiable Traffic Simulation

Sanghyun Son, Ming C. Lin

Autonomous DrivingOptimizationWorld ModelVideoTime Series

🎯 What it does: Proposed a parallel differentiable traffic simulator based on the Intelligent Driver Model (IDM), utilizing the simulator to achieve trajectory denoising, dense trajectory reconstruction, and future trajectory prediction, and optimizing IDM parameters through gradient methods.

Graph2Nav: 3D Object-Relation Graph Generation to Robot Navigation

Tixiao Shan, Han-Pang Chiu

Robotic IntelligenceGraph Neural NetworkLarge Language ModelGraph

🎯 What it does: Propose Graph2Nav, a real-time 3D object relationship graph generation framework for enabling robot autonomous navigation in indoor and outdoor scenes.

GraspSAM: When Segment Anything Model Meets Grasp Detection

Sangjun Noh, Kyoobin Lee

SegmentationRobotic IntelligenceTransformerSupervised Fine-TuningPrompt EngineeringImage

🎯 What it does: Propose the GraspSAM model, applying the Segment Anything Model (SAM) to grasp detection, achieving prompt-driven and category-agnostic grasp recognition.

Ground-Aware Automotive Radar Odometry

Daniel Casado Herraez, C. Stachniss

Pose EstimationAutonomous DrivingOptimizationSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Vehicle self-localization using a single automotive radar, proposing a heuristic-based ground plane extraction and matching method, and performing windowed factor graph optimization to improve pose estimation accuracy.

Ground-Level Viewpoint Vision-and-Language Navigation in Continuous Environments

Zerui Li, Qi Wu

Robotic IntelligenceVision Language ModelVision-Language-Action ModelWorld ModelImageTextGraph

🎯 What it does: Proposes the Ground-level Viewpoint Navigation (GVNav) method, which constructs enhanced spatiotemporal context using weighted historical observations to address the matching problem between low-height quadruped robots and human instructions, and transfers connectivity graphs from HM3D and Gibson to improve spatial priors.

Ground-Optimized 4D Radar-Inertial Odometry Via Continuous Velocity Integration Using Gaussian Process

Wooseong Yang, Ayoung Kim

Autonomous DrivingOptimizationSimultaneous Localization and MappingPoint CloudTime Series

🎯 What it does: Proposes a ground-optimized noise filtering and continuous velocity pre-integration method based on radar to achieve more accurate radar-inertial odometry

GS-EVT: Cross-Modal Event Camera Tracking Based on Gaussian Splatting

Tao Liu, L. Kneip

Object TrackingPose EstimationGaussian SplattingMultimodality

🎯 What it does: Motion tracking using an event camera, and achieving cross-modal localization through the efficient high-quality view synthesis technique Gaussian Splatting.

Guaranteed Reach-Avoid for Black-Box Systems through Narrow Gaps via Neural Network Reachability

Long Kiu Chung, Shreyas Kousik

Autonomous DrivingReinforcement Learning

🎯 What it does: Proposed the NeuralPARC method, extending the previous Piecewise Affine Reach-avoid Computation (PARC) to black-box systems modeled by ReLU neural networks, and computing reachable sets to ensure reaching the target while avoiding collisions in narrow gap obstacle environments.

Guiding Long-Horizon Task and Motion Planning with Vision Language Models

Zhutian Yang, L. Kaelbling

Robotic IntelligenceVision Language ModelMultimodality

🎯 What it does: Propose a hierarchical planning algorithm called VLM-TAMP, which utilizes a vision-language model to generate semantically meaningful intermediate subgoals that shorten the planning horizon and guide task and motion planning.

H2O+: An Improved Framework for Hybrid Offline-and-Online RL with Dynamics Gaps

Haoyi Niu, Xianyuan Zhan

Reinforcement Learning

🎯 What it does: Designed a new hybrid offline-online reinforcement learning framework H2O+, which can integrate limited offline data with imperfect simulators while considering the dynamics differences between real and simulated environments to achieve transferable policy learning.

H3O: Hyper-Efficient 3D Occupancy Prediction with Heterogeneous Supervision

Yunxiao Shi, F. Porikli

Autonomous DrivingComputational EfficiencyPoint CloudBenchmark

🎯 What it does: Proposed an efficient 3D occupancy prediction method called H30, which significantly reduces computational cost through a streamlined architecture design, and achieves multi-source heterogeneous supervision for 3D occupancy labels by leveraging auxiliary tasks such as multi-camera depth estimation, semantic segmentation, and surface normal estimation;

Hand-Object Interaction Pretraining from Videos

Himanshu Gaurav Singh, Jitendra Malik

Robotic IntelligenceReinforcement LearningVideo

🎯 What it does: Using 3D hand-object interaction trajectories from natural scene videos, generate robot perceptual-motor trajectories, learn task-agnostic baseline policies through generative models, and further fine-tune with reinforcement learning and behavioral cloning to achieve high sample efficiency, robustness, and generalization for downstream tasks.

Haptic Shoulder for Rendering Biomechanically Accurate Joint Limits for Human-Robot Physical Interactions

Elizabeth Peiros, Michael C. Yip

Robotic Intelligence

🎯 What it does: Proposed the SHULDRD shoulder joint simulator for human-machine physical interaction testing without human subjects

Hardware-Accelerated Ray Tracing for Discrete and Continuous Collision Detection on GPUs

Sizhe Sui, Andrew Bylard

Robotic IntelligenceMesh

🎯 What it does: Proposed a robot collision detection algorithm based on GPU hardware-accelerated ray tracing, used for precise mesh-to-mesh discrete collision detection and spherical robot-to-mesh sweeping continuous collision detection.

HARP: Human-Assisted Regrouping With Permutation Invariant Critic for Multi-Agent Reinforcement Learning

Huawen Hu, Shu Zhang

Reinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: Proposes the HARP framework, integrating multi-agent automatic reorganization and deployment with human assistance, enabling non-experts to provide effective guidance with minimal intervention; during the training phase, agents dynamically adjust groupings, and during deployment, they proactively seek human assistance, utilizing permutation-invariant group evaluators to assess and refine human-proposed groupings.

Hash-GS: Anchor-Based 3D Gaussian Splatting with Multi-Resolution Hash Encoding for Efficient Scene Reconstruction

Yijia Xie, Jiajun Lv

Gaussian Splatting

🎯 What it does: Proposes Hash-GS, an efficient method for large-scale scene reconstruction that utilizes anchor-based 3D Gaussian Splatting and multi-resolution hash encoding.

HBTP: Heuristic Behavior Tree Planning with Large Language Model Reasoning

Xinglin Chen, Ji Wang

TransformerLarge Language Model

🎯 What it does: Developed the Heuristic Behavior Tree Planning (HBTP) framework, combining large language model (LLM) task reasoning with behavior tree (BT) planning, achieving efficient BT expansion based on heuristic paths, and proposed two heuristic variants (optimal planning and sufficient planning), as well as action space pruning and reflection feedback methods to address LLM reasoning inaccuracies.

Helical Structured Soft Growing Robot for Hazardous Gas Suction in Inaccessible Environments

Sanghun Lee, Jee-Hwan Ryu

Robotic Intelligence

🎯 What it does: Developed a soft-growing robot integrated with an inflatable spiral structure, capable of extending in complex environments and absorbing harmful gases through internal channels.

Helios: Heterogeneous Lidar Place Recognition via Overlap-Based Learning and Local Spherical Transformer

Minwoo Jung, Ayoung Kim

RetrievalTransformerContrastive LearningPoint Cloud

🎯 What it does: Developed HeLiOS, a deep network for heterogeneous LiDAR, used for LiDAR location recognition, leveraging small local windows and spherical transformers along with clustering assignment based on optimal transport to generate robust global descriptors, and improving performance through overlapping foundation data mining and guided triplet loss.

HelmetPoser: A Helmet-Mounted IMU Dataset for Data-Driven Estimation of Human Head Motion in Diverse Conditions

Jianping Li, Lihua Xie

Pose EstimationRecurrent Neural NetworkTransformerTime SeriesSequential

🎯 What it does: Proposed a helmet-mounted IMU dataset with ground truth, and used LSTM and Transformer networks to correct IMU bias to improve localization accuracy, while evaluating performance across various activities of ten participants.

HeRCULES: Heterogeneous Radar Dataset in Complex Urban Environment for Multi-Session Radar SLAM

Hanjun Kim, Ayoung Kim

Robotic IntelligenceSimultaneous Localization and MappingMultimodalityBenchmark

🎯 What it does: Proposed and released a heterogeneous radar dataset named HeRCULES, which includes 4D radar, spinning radar, FMCW LiDAR, IMU, GPS, and camera, for multi-session, multi-robot radar SLAM research in complex urban environments.

Here's your PDDL Problem File! On Using VLMs for Generating Symbolic PDDL Problem Files

Victor Aregbede, Achim J. Lilienthal

Robotic IntelligenceTransformerVision Language ModelImageText

🎯 What it does: Designed and implemented a hybrid system ViPlan, which uses Vision Language Models (VLM) to extract high-level semantic information from visual and textual inputs, generates syntactically and semantically correct PDDL problem files, and combines classical planners to generate executable plans; the entire process is embedded in a behavior tree framework to improve efficiency, reactivity, replanning, modularity, and flexibility.

HeRo: A State Machine-Based, Fault-Tolerant Framework for Heterogeneous Multi-Robot Collaboration

R. Tang, Jun Wei

Robotic IntelligenceLarge Language Model

🎯 What it does: Designed and implemented a state machine-based fault-tolerant framework called HeRo for heterogeneous multi-robot collaboration, supporting development, state synchronization, and automatic fault detection and recovery.

HEROES: Unreal Engine-based Human and Emergency Robot Operation Education System

A. Chaudhary, Aniket Bera

Data SynthesisRobotic IntelligenceReinforcement Learning

🎯 What it does: Developed the HEROES simulator based on Unreal Engine to create urban search and rescue training environments for human rescuers and emergency robots, and to generate synthetic datasets for robot navigation learning.

Heterogeneous Exploration and Monitoring with Online Free-Space Ellipsoid Graphs

Brennan Brodt, Alyssa Pierson

Autonomous DrivingOptimization

🎯 What it does: Propose a heterogeneous team solution using agile sensing探测器 to explore unknown non-convex environments, and employing slow monitoring/service agents to accomplish target discovery and monitoring; generate collision-free elliptical maps via the IRIS algorithm, and hand over these maps to monitoring agents for polynomial complexity allocation and cruise planning, enabling high-quality path servicing for all targets.

Heterogeneous Sensor Fusion and Active Perception for Transparent Object Reconstruction with a PDM2 Sensor and a Camera

Fengzhi Guo, Dezhen Song

OptimizationMultimodality

🎯 What it does: Proposed a sensor fusion method that combines a conventional camera with a PDM2 sensor for shape reconstruction of transparent household objects.

Heuristically Guided Compilation for Task Assignment and Path Finding

Zheng Chen, Junhao Wang

Optimization

🎯 What it does: Proposed a compilation-based heuristic method to solve multi-agent task allocation and path planning problems, approximately solving for optimal solutions.

Hey Robot! Personalizing Robot Navigation Through Model Predictive Control with a Large Language Model

Diego Martinez-Baselga, Luis Montano

OptimizationRobotic IntelligenceVision Language ModelImageText

🎯 What it does: Propose a zero-shot method that leverages existing visual language models to interpret users' text or image instructions, automatically generates cost functions, and reconfigures model predictive controllers to convert natural language instructions into robot motion behaviors.

HFUS-NeRF: Hybrid Representation for Fast Ultrasound Reconstruction in Robotic Ultrasound System

Shuai Zhang, Bo Ouyang

RestorationNeural Radiance FieldGaussian SplattingBiomedical DataUltrasound

🎯 What it does: Propose a hybrid representation method called HFUS-NeRF that combines multi-resolution hash grids with triplane representations for fast, high-quality ultrasound image reconstruction

HGAT-CP: Heterogeneous Graph Attention Network for Collision Prediction in Autonomous Driving

Yongzhi Jiang, Zhongxia Xiong

Autonomous DrivingRecurrent Neural NetworkGraph Neural NetworkGraph

🎯 What it does: Proposes a framework named HGAT-CP, combining Heterogeneous Graph Attention Network (HGAT) and LSTM to predict potential collision events in autonomous driving.

HGS-Planner: Hierarchical Planning Framework for Active Scene Reconstruction Using 3D Gaussian Splatting

Zijun Xu, Wenchao Ding

GenerationGaussian SplattingPoint Cloud

🎯 What it does: Proposes a hierarchical planning framework based on 3D Gaussian Splatting for fast and high-fidelity active scene reconstruction.

HGSLoc: 3DGS-Based Heuristic Camera Pose Refinement

Zhongyan Niu, Dewen Hu

Pose EstimationOptimizationGaussian SplattingImage

🎯 What it does: Propose a lightweight, plug-and-play HGSLoc framework that combines 3D reconstruction with heuristic optimization to improve the accuracy of camera pose estimation.

Hide-in-Motion: Embedding Steganographic Copyright Information into 4D Gaussian Splatting Assets

Hengyu Liu (Chinese University of Hong Kong), Yixuan Yuan (Chinese University of Hong Kong)

Gaussian SplattingPoint Cloud

🎯 What it does: Proposed a 4D steganography method called Hide-in-Motion, which hides information through deformation in Gaussian splatting assets.

Hier-SLAM: Scaling-Up Semantics in SLAM with a Hierarchically Categorical Gaussian Splatting

Boying Li, Hamid Rezatofighi

Large Language ModelGaussian SplattingSimultaneous Localization and Mapping

🎯 What it does: Propose Hier-SLAM, a semantic SLAM method that utilizes hierarchical category 3D Gaussian splatting to achieve accurate global semantic mapping, scalability, and explicit semantic label prediction.

Hierarchical Contact-Rich Trajectory Optimization for Multi-Modal Manipulation Using Tight Convex Relaxations

Yuki Shirai, Devesh K. Jha

OptimizationRobotic Intelligence

🎯 What it does: Proposes a hierarchical optimization framework for efficiently designing robot, object, and contact trajectories to achieve multimodal contact-rich operations.

Hierarchical End-to-End Autonomous Driving: Integrating BEV Perception with Deep Reinforcement Learning

Siyi Lu, Keqiang Li

SegmentationAutonomous DrivingReinforcement Learning

🎯 What it does: Propose an end-to-end deep reinforcement learning (DRL) driving framework based on bird's-eye view (BEV) representation, directly mapping the DRL feature extraction network to the perception stage and enhancing interpretability through semantic segmentation.

Hierarchical Reinforcement Learning for Safe Mapless Navigation with Congestion Estimation

Jianqi Gao, Yanjie Li

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes a safe mapless navigation framework based on hierarchical reinforcement learning, which generates subgoals using a high-level policy combined with environmental congestion estimation, employs low-level safe reinforcement learning to generate real-time control commands, and introduces a novel obstacle encoding method to enhance perception capabilities.

Hierarchical Spatiotemporal Fusion for Event-Visible Object Detection

Sin-Ye Jhong, Yung-Yao Chen

Object DetectionMultimodality

🎯 What it does: Proposes a hierarchical spatiotemporal fusion method based on event cameras to improve visible light object detection performance under varying weather and lighting conditions

Hierarchical Tri-Manual Planning for Vision-Assisted Fruit Harvesting with Quadrupedal Robots

Zhichao Liu, Konstantinos Karydis

Robotic IntelligenceSimultaneous Localization and MappingImagePoint CloudAgriculture Related

🎯 What it does: Developed the first three-arm quadruped robot based on Spot, named LocoHarv3, and proposed a hierarchical three-hand planning method for automated fruit harvesting, achieving collision-free trajectory planning between the robot's body and custom-built dual-arm manipulators.

Hierarchical Visual Policy Learning for Long-Horizon Robot Manipulation in Densely Cluttered Scenes

Hecheng Wang, Yunquan Sun

Robotic IntelligenceReinforcement Learning

🎯 What it does: A visually driven hierarchical strategy, HCLM, is proposed for long-term stacking tasks in dense cluttered scenes, achieving object manipulation through high-level policies and three action primitives (pushing, grasping, placing).

Hierarchically Accelerated Coverage Path Planning for Redundant Manipulators

Yeping Wang, M. Gleicher

OptimizationRobotic Intelligence

🎯 What it does: Proposes a coverage path planning method that exploits the redundancy of robotic arms and task tolerances to minimize costs in joint space.

High Accuracy Aerial Maneuvers on Legged Robots using Variational Integrator Discretized Trajectory Optimization

S. Beck, Quan Nguyen

OptimizationRobotic IntelligenceOrdinary Differential Equation

🎯 What it does: Propose a whole-body trajectory optimization method using variational integration for aggressive maneuvers during long flight periods

High Speed Robotic Table Tennis Swinging Using Lightweight Hardware with Model Predictive Control

David Nguyen, Sangbae Kim

OptimizationRobotic Intelligence

🎯 What it does: Designed and implemented a five-degree-of-freedom robot table tennis platform with lightweight and high torque, utilizing optimal control problems and model predictive control to generate and execute various ball striking trajectories for precise hits with different spin styles.

High-Force Electroadhesion Based on Unique Liquid-Solid Dielectrics for UAV Perching

Junjie Luo, Jian Zhu

Autonomous Driving

🎯 What it does: Designed and experimentally verified a high-force electro-adhesive pad embedded with liquid and solid media for drone landing

High-Performance Reinforcement Learning on Spot: Optimizing Simulation Parameters with Distributional Measures

AJ Miller, Farbod Farshidian

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Developed and deployed a high-performance reinforcement learning control strategy based on the Spot RL Researcher Development Kit, achieving low-level motor access and end-to-end training and deployment;

High-Precision Object Pose Estimation Using Visual-Tactile Information for Dynamic Interactions in Robotic Grasping

Zicai Peng, Yufeng Yue

Pose EstimationRobotic IntelligenceImageMultimodality

🎯 What it does: Estimate object pose using visual-tactile information through particle filters, real-time track the pose of contact objects, and estimate changes in the pose of grasped objects via tactile displacement.

High-Quality 3D Creation From a Single Image Using Subject-Specific Knowledge Prior

Nan Huang, Shanghang Zhang

GenerationNeural Radiance FieldImageMesh

🎯 What it does: Proposed a two-stage method to generate high-quality 3D models from a single image.

High-Quality Unknown Object Instance Segmentation via Quadruple Boundary Error Refinement

Seunghyeok Back, Kyoobin Lee

SegmentationComputational Efficiency

🎯 What it does: Propose a novel instance segmentation method for unknown objects based on quadruple boundary error correction (QuBER), which estimates four types of errors on the instance boundaries of initial segmentation results and performs error-guided fusion correction to achieve high-quality segmentation;

High-Resolution Reconstruction of Non-Planar Tactile Patterns From Low-Resolution Taxel-Based Tactile Sensors

Cheng Zhou, Qian Liu

Super ResolutionConvolutional Neural NetworkDiffusion modelGenerative Adversarial NetworkImage

🎯 What it does: Studied the high-resolution reconstruction of non-planar tactile patterns captured by low-resolution tactile sensors

Highly Dynamic Physical Interaction for Robotics: Design and Control of an Active Remote Center of Compliance

Christian Friedrich, Matthias Haag

Robotic IntelligenceBenchmarkPhysics Related

🎯 What it does: Propose a hybrid control scheme that combines the advantages of active and passive interaction control, and design a novel Active Remote Compliance Center (ARCC) for direct control of interaction forces; validate its performance through dynamic comparisons with simulation models and pure robot control schemes, integrate the scheme into the robot controller; evaluate it on different industrial benchmark tasks (e.g., peg-in-hole, circular cap track assembly, and contour tracking).

Hook-Based Aerial Payload Grasping from a Moving Platform

Péter Antal, R. T'oth

OptimizationRobotic Intelligence

🎯 What it does: A computation-efficient trajectory optimization method based on complementary constraints is proposed to determine the optimal grasping timing, and a physics-based simulation model is used to predict the future motion of the load. Subsequently, the success rate of grasping under model uncertainty and external disturbances is verified using a robustness analysis method based on integral quadratic constraints (IQC). The feasibility of the algorithm is evaluated through experiments in a high-fidelity physics simulator and on a custom-built aerial manipulation platform.

HOVER: Versatile Neural Whole-Body Controller for Humanoid Robots

Tairan He, Yuke Zhu

Knowledge DistillationRobotic Intelligence

🎯 What it does: A multi-mode strategy distillation framework called HOVER is proposed to integrate various whole-body control modes, enabling seamless switching among tasks such as navigation, motion control, and desktop manipulation in humanoid robots.

How About Them Apples: 3D Pose and Cluster Estimation of Apple Fruitlets in a Commercial Orchard

A. Qureshi, Henry Williams

Pose EstimationImageAgriculture Related

🎯 What it does: Propose a novel visual system for mapping the orientation and clustering information of apple granules.

How Sound-Based Robot Communication Impacts Perceptions of Robotic Failure

J. L. Crider, Naomi T. Fitter

Robotic IntelligenceVideo

🎯 What it does: Conducted a within-subject online experiment with modern commercial robots, using different communication modes (voice vs. non-voice) in videos of successful and failed tasks, to evaluate assessments of robot capabilities and trust after failures.

How to Train Your Robots? The Impact of Demonstration Modality on Imitation Learning

Haozhuo Li, Dorsa Sadigh

Robotic Intelligence

🎯 What it does: Compare the impact of low-cost demonstration methods (kinesthetic teaching, VR controller remote operation, spacemouse remote operation) on imitation learning performance and user experience in three desktop manipulation tasks.

HPRM: High-Performance Robotic Middleware for Intelligent Autonomous Systems

Jacky Kwok, Edward A. Lee

Autonomous DrivingComputational EfficiencyRobotic IntelligenceBenchmark

🎯 What it does: Proposed a high-performance robot middleware HPRM based on Lingua Franca, addressing the non-determinism and high communication latency issues of ROS2 in multi-core platforms with big data and multiple subscribers scenarios.

Hri-Free: Cognitive Robotic Simulation for Evaluating Embodied Social Attention Models

Fares Abawi, Di Fu

Robotic IntelligenceVideo

🎯 What it does: Proposed a cognitive robot simulation framework to evaluate social attention models, assessed by projecting real-world priority maps into simulated environments.

HS-SLAM: Hybrid Representation with Structural Supervision for Improved Dense SLAM

Ziren Gong, Matteo Poggi

Autonomous DrivingOptimizationRobotic IntelligenceNeural Radiance FieldSimultaneous Localization and Mapping

🎯 What it does: Proposed the HS-SLAM system, which enhances scene representation, structural capture, and global consistency of NeRF-based SLAM through a hybrid encoding network, structural supervision, and active global bundle adjustment.

HSRL: A Hierarchical Control System Based on Spiking Deep Reinforcement Learning for Robot Navigation

Bo Yang, Huajin Tang

Robotic IntelligenceSpiking Neural NetworkReinforcement Learning

🎯 What it does: Propose a hierarchical control system based on Spiking Deep Reinforcement Learning (SDRL) for robot navigation, employing a high-level Spiking GRU decision network and a low-level continuous attractor neural network (CANN) to execute precise continuous actions.

HULK: Large-Scale Hierarchical Coordination Under Continual and Uncertain Temporal Tasks

Q. Luo, Meng Guo

Reinforcement Learning

🎯 What it does: Proposes a hierarchical coordination framework named HULK for handling temporal logic tasks in multi-agent systems, where tasks are continuously generated with an uncertain number of subtasks.

Human Activity Recognition by Using Enhanced Radar Point Cloud 2D Histograms and Doppler Feature Fusion

Guanghang Liao, Fei Luo

RecognitionPoint Cloud

🎯 What it does: A precise non-intrusive human action recognition framework based on 2D histograms of millimeter-wave radar point clouds is proposed. The traditional histograms are improved by incorporating a fixed radar perception boundary, and concatenated with velocity (Doppler) features (range-velocity and angle-velocity histograms) to construct a multi-layer hybrid network to address overfitting issues in stacked networks.

Human-Agent Joint Learning for Efficient Robot Manipulation Skill Acquisition

Shengcheng Luo, Yong-Lu Li

Robotic IntelligenceAgentic AI

🎯 What it does: Proposed a human-robot collaborative learning system that enables human operators and robots to share end-effector control, simplifying data collection and achieving simultaneous demonstration and robot learning.

Human-Like Walking Motion Generation for Self-Balancing Lower Limb Rehabilitation Exoskeletons

Ming Yang, Xinyu Wu

GenerationOptimizationRobotic Intelligence

🎯 What it does: Designed a hierarchical optimization-based gait generator capable of producing gaits with human-like characteristics such as variable hip height, heel contact, toe-off, and knee extension.

Human-Robot Collaboration for the Remote Control of Mobile Humanoid Robots With Torso-Arm Coordination

Nikita Boguslavskii, Zhi Li

Robotic Intelligence

🎯 What it does: Proposed multiple human-robot collaboration (HRC) methods to coordinate the motion of the trunk and arms of a remotely controlled mobile humanoid robot, aiming to improve system efficiency and task execution performance.

Human-Robot Cooperative Distribution Coupling for Hamiltonian-Constrained Social Navigation

Weizheng Wang, Byung-Cheol Min

Robotic IntelligenceReinforcement Learning from Human FeedbackTransformerReinforcement LearningDiffusion model

🎯 What it does: Proposes NaviDIFF, a socially aware navigation framework based on Hamiltonian constraints, to address the complexity of human-robot interaction and path planning in public spaces.

HumanFT: A Human-Like Fingertip Multimodal Visuo-Tactile Sensor

Yifan Wu, Chenxi Xiao

Robotic IntelligenceMultimodality

🎯 What it does: Designed and fabricated a multimodal visual-tactile sensor named HumanFT, which mimics the shape and function of human fingertips, capable of real-time force measurement, high-frequency vibration detection, and over-temperature alerts, with its performance verified through experiments;

Humanoid Walking Stabilization via Model Predictive Control with Step Adjustment Based on the 3D Divergent Component of Motion

G. Park, Jaeheung Park

OptimizationRobotic Intelligence

🎯 What it does: A novel model predictive control (MPC) framework based on three-dimensional Divergent Component of Motion (3D-DCM) is proposed to stabilize humanoid robot walking under varying center of mass (CoM) height conditions. The method analytically formulates control inputs, virtual repulsion points (VRP), and gait adjustment constraints into quadratic constrained quadratic programming (QCCQP). Additionally, a CoM-foot distance constraint is integrated into 3D-CoM trajectory planning to enhance the robot's disturbance rejection capability across varying step lengths and complex terrains. The effectiveness of this approach is validated through simulations and real-robot experiments.

Hybrid Decentralization for Multi-Robot Orienteering with Mothership-Passenger Systems

N. Butler, Geoffrey A. Hollinger

OptimizationRobotic Intelligence

🎯 What it does: Proposed a hybrid centralized-decentralized planning algorithm for the multi-robot navigation task in Mothership and Passenger robot systems.

Hybrid Decision Making for Scalable Multi-Agent Navigation: Integrating Semantic Maps, Discrete Coordination, and Model Predictive Control

K. Vos, R. V. D. Molengraft

Autonomous DrivingOptimization

🎯 What it does: Designed and verified a multi-agent navigation framework that integrates shared semantic maps, allocation strategies, and model predictive control.

Hybrid Deep Reinforcement Learning for Radio Tracer Localisation in Robotic-Assisted Radioguided Surgery

Hanyi Zhang, Daniel S. Elson

Robotic IntelligenceReinforcement LearningBiomedical DataPositron Emission Tomography

🎯 What it does: Proposed a learning-based robot-assisted method for localizing radioactive tracers, achieving detection of radioactive targets through automatic navigation probes.

Hybrid Gripper with Passive Pneumatic Soft Joints for Grasping Deformable Thin Objects

Ngoc-Duy Tran, Tung D. Ta

Robotic Intelligence

🎯 What it does: Developed a hybrid gripper with passive pneumatic soft joints for grasping deformable thin objects

Hybrid State Estimation and Mode Identification of an Amphibious Robot

H. B. Amundsen, Michael R. Benjamin

Robotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Proposed a hybrid observer for estimating the continuous states of C-Ray during water-land transition and identifying mode switching, achieving autonomous water-land transfer.

Hyperdimensional Computing-Based Federated Learning in Mobile Robots Through Synthetic Oversampling

Hyun-Soo Lee, Yeseong Kim

Federated LearningRobotic Intelligence

🎯 What it does: Proposed a federated learning framework based on hyperdimensional computing, introducing dynamic encoding and hyperdimensional resampling techniques

Hypergraph-Based Coordinated Task Allocation and Socially-Aware Navigation for Multi-Robot Systems

Weizheng Wang, Byung-Cheol Min

OptimizationRobotic IntelligenceGraph Neural NetworkReinforcement LearningGraph

🎯 What it does: Proposes Hyper-SAMARL, a hypergraph-based multi-robot task allocation and socialized navigation system

Hypergraph-Transformer (HGT) for Interaction Event Prediction in Laparoscopic and Robotic Surgery

Lianhao Yin, Guy Rosman

Representation LearningRobotic IntelligenceTransformerVideoGraph

🎯 What it does: Propose a prediction neural network based on Hypergraph Transformer (HGT) for understanding and predicting key interaction events in minimally invasive surgery, combining endoscopic videos and surgical knowledge graphs.

IBURD: Image Blending for Underwater Robotic Detection

Jungseok Hong, Junaed Sattar

Image HarmonizationData SynthesisDepth EstimationImage

🎯 What it does: Proposed the IBURD image blending pipeline to generate realistic synthetic marine debris images and pixel-level annotations, aiding in the training of depth detection models for underwater autonomous vehicles (AUVs).

ICBSS: An Improved Algorithm for Multi-Agent Combinatorial Path Finding

Zheng Chen, Yiran Ni

Optimization

🎯 What it does: Propose an improved conflict-based Steiner search algorithm, ICBSS, which replaces multiple trees with a single constraint tree and alternates between time-dependent traveling salesman problem (TDTSP) algorithms at each node to solve optimal joint paths, while also introducing a suboptimal variant to enhance computational efficiency.