arXivSub Start free trial

IROS 2025 Papers — Page 7

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

Dynamic Obstacle Avoidance through Uncertainty-Based Adaptive Planning with Diffusion

Vineet Punyamoorty, Aniket Bera

Autonomous DrivingReinforcement LearningDiffusion model

🎯 What it does: Propose a diffusion model based on uncertainty-adaptive planning for dynamic obstacle avoidance;

Dynamic planning and assembly for constructing mortar-joint multi-leaf stone masonry walls with a robotic arm

Qianqing Wang, S. Parascho

OptimizationRobotic Intelligence

🎯 What it does: Developed a robotic system for automated construction of multi-layer stone masonry walls with mortar joints.

Dynamic Quadrupedal Legged and Aerial Locomotion via Structure Repurposing

Chenghao Wang (Northeastern University), Morteza Gharib (Northeastern University)

Robotic Intelligence

🎯 What it does: Designed and implemented a multimodal ground-air quadruped robot Husky v.2, demonstrating its ability to achieve dynamic quadrupedal walking and hovering through structural reuse, and reporting hardware design and experimental results.

Dynamic Residual Safe Reinforcement Learning for Multi-Agent Safety-Critical Scenarios Decision-Making

Kaifeng Wang, Xin Gao

Reinforcement Learning

🎯 What it does: Proposed a dynamic residual safety reinforcement learning (DRS-RL) framework for decision-making in multi-agent safety-critical scenarios.

Dynamic Risk-Aware MPPI for Mobile Robots in Crowds via Efficient Monte Carlo Approximations

Elia Trevisan, Javier Alonso-Mora

Autonomous DrivingRobotic Intelligence

🎯 What it does: Propose a dynamic risk-aware model predictive path integral controller (DRA-MPPI), which estimates the joint collision probability of multiple dynamic obstacles under real-time Monte Carlo methods, and applies it to sample rejection or cost weighting to address the robot's 'freezing' problem and enhance safety.

Dynamic Walking Corridor Generation for Visually Impaired Navigation Using Social Force Models and Convex Optimization

Qingquan Na, A. Frisoli

OptimizationSafty and Privacy

🎯 What it does: Propose a dynamic walking corridor generation algorithm to provide safe navigation for visually impaired individuals in crowded environments

Dynamic-Characteristics-Based Continuous Impact-Minimizing Rolling Locomotion for Variable Topology Truss

Hanbom Kim, Taewon Seo

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: Proposed a continuous impact minimization rolling locomotion method, CIM, enabling variable topology truss robots to achieve seamless continuous motion.

DynamicGSG: Dynamic 3D Gaussian Scene Graphs for Environment Adaptation

Luzhou Ge, Xuesong Li

Domain AdaptationGraph Neural NetworkVision Language ModelGaussian Splatting

🎯 What it does: Propose the DynamicGSG system, which utilizes high-precision Gaussian light scattering technology to construct a dynamically updatable 3D Gaussian scene graph for environmental adaptation.

Dynamicity Adaptation for Multi-object Tracking and Segmentation: Toward Improved Association Correction

Longtao Chen, Huanqiang Zeng

Object TrackingSegmentationVideoBenchmark

🎯 What it does: Designed DA-Track, which improves association correction in multi-object tracking and segmentation through dynamic adaptation;

DynamicPose: Real-time and Robust 6D Object Pose Tracking for Fast-Moving Cameras and Objects

Tingbang Liang, Boyu Zhou

Object TrackingPose EstimationSimultaneous Localization and MappingMultimodality

🎯 What it does: Proposes a real-time 6D object pose tracking framework called DynamicPose, which does not require retraining and is designed for scenarios with rapidly moving cameras and objects.

Dynamics-Invariant Quadrotor Control using Scale-Aware Deep Reinforcement Learning

Varad Vaidya, J. Keshavan

Robotic IntelligenceReinforcement Learning

🎯 What it does: Propose a quadrotor control framework based on deep reinforcement learning, achieving physical dynamics invariance by directly optimizing force/torque inputs, and employing a temporal trajectory encoder, latent dynamics encoder, and scale-aware dynamic randomization

Eagle-Scale Flapping-Wing Robot with Aggressive Roll Maneuverability: Bio-Inspired Actuation, Fluid-Structure Interaction Simulation and Flight Experiment

Haoyu Wang, W. Xu

Robotic IntelligencePhysics Related

🎯 What it does: By reverse-engineering the biomechanics of raptor flight, a biomimetic wing shoulder torsion mechanism was proposed, and a hawk-like flapping-wing aircraft was successfully developed, completing high-speed driving and extreme rolling maneuver experiments.

EANS: Reducing Energy Consumption for UAV with an Environmental Adaptive Navigation Strategy

Tian Liu, Kai Huang

Autonomous DrivingOptimization

🎯 What it does: Proposes a method that dynamically adjusts navigation strategies by analyzing UAV dynamic characteristics and the temporal features of autonomous navigation pipelines to reduce energy consumption.

EAROL: Environmental Augmented Perception-Aware Planning and Robust Odometry via Downward-Mounted Tilted LiDAR

Xinkai Liang, Hao Fang

OptimizationRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Propose the EAROL framework, utilizing a 20° tilted downward-mounted LiDAR to enhance environmental perception and achieve robust odometry, combined with LIO and hierarchical trajectory-yaw optimization to enable UAV autonomous localization and planning in open-top scenarios.

EASEIR: Efficient and Adaptive Safe-set Estimation via Implicit Representation for High-dimensional Motion Planning

Hojun Lee, M. Jun

Safty and PrivacyComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposed and implemented the EASEIR method, which rapidly generates safe sets for robot operations in high-dimensional configuration spaces using implicit neural representations;

EDeformNet: Estimating Fishing Net Deformations from Sparse Observations

I. Wijegunawardana, Shoudong Huang

Data SynthesisOptimizationMeshPhysics Related

🎯 What it does: Real-time 3D reconstruction of fishing nets using EDeformNet with sparse localization observations and acoustic tracking beacons.

Edge-Guided Lighting Adaptation: Real-Time Detection of Transparent Objects for Cell Culture Robot

Qingze Huang, Shimin Wei

Object DetectionData SynthesisRobotic IntelligenceConvolutional Neural NetworkTransformerImage

🎯 What it does: Construct a synthetic dataset and improve YOLOv8 to achieve real-time detection of transparent objects

EdgeSpotter: Multi-Scale Dense Text Spotting for Industrial Panel Monitoring

Changhong Fu, Liguo Zhang

RecognitionObject DetectionTransformerImageBenchmark

🎯 What it does: Proposed a multi-scale dense text detection and recognition system called EdgeSpotter for industrial panel monitoring

EdgeSR: Reparameterization-Driven Fast Thermal Super-Resolution for Edge Electro-Optical Device

Changhong Fu, Haobo Zuo

Super ResolutionComputational EfficiencyConvolutional Neural NetworkImageBenchmark

🎯 What it does: Proposed a fast thermal super-resolution model called EdgeSR for edge electro-optical devices.

Edit Distance Based Intention Estimation for Teleoperated Assembly

Aolin Xu, Behzad Dariush

Robotic IntelligenceSequential

🎯 What it does: To address intention estimation in remote operation assembly, this paper proposes using edit distance to quantify the similarity between observed action sequences and standard task sequences, and employs the nearest neighbor rule for task identification and prediction of the next action.

EDSOD: An Encoder-Decoder, Diffusion-model, and Swin-Transformer-based Small Object Detector

Junnian Li, Zhengcai Cao

Object DetectionTransformerDiffusion modelImage

🎯 What it does: Proposed a small object detection framework for aerial images based on encoder-decoder architecture, diffusion models, and Swin Transformer, redefining the SOD task as a Noise-to-Box process;

Educational SoftHand-A: Building an Anthropomorphic Hand with Soft Synergies using LEGO® MINDSTORMS®

Jared K. Lepora, Nathan F. Lepora

Robotic Intelligence

🎯 What it does: Built a fully LEGO® MINDSTORMS®-assembled humanoid hand — Educational SoftHand‑A — using a dual-motor-driven antagonistic tendon system and achieving synchronized fingertip motion through soft synergy (differential mechanism and clutch gears);

EFCWM-Mamba-YOLO: Real-Time Underwater Object Detection with Adaptive Feature Representation and Domain Adaptation

Pan Sun, Huilin Ge

Object DetectionDomain AdaptationConvolutional Neural NetworkImage

🎯 What it does: Proposed a lightweight real-time underwater object detection model named EFCWM-Mamba-YOLO, and created the UOD-SZTU-2025 dataset containing 3,133 high-quality images.

Effect of Haptic Feedback on Avoidance Behavior and Visual Exploration in Dynamic VR Pedestrian Environment

Kyosuke Ishibashi, Ko Yamamoto

Time Series

🎯 What it does: Explore the impact of tactile feedback on walking behavior in crowded dynamic pedestrian VR environments through user studies, analyzing users' collision avoidance actions, trajectory length, pelvic angle, lateral displacement, and visual exploration behavior.

Efficient and Accurate Low-Resolution Transformer Tracking

Shaohua Dong, Heng Fan

Object TrackingComputational EfficiencyKnowledge DistillationTransformerVideo

🎯 What it does: Proposed a low-resolution Transformer tracker called LoReTrack, which enhances tracking performance at low resolutions through dual knowledge distillation.

Efficient and Precise Drone Rephotography for Video Sequences

Hao-Liang Xu, Kuan-Wen Chen

Pose EstimationSimultaneous Localization and MappingOptical FlowVideo

🎯 What it does: Developed a video-based precise drone rephotography system that recovers camera poses using video sequences and achieves precise alignment of image sequences. The system combines visual SLAM with a dense flow prediction model for continuous pose optimization and introduces a Dynamic Frame Alignment Error (DFAE) evaluation metric.

Efficient and Real-Time Motion Planning for Robotics Using Projection-Based Optimization

Xuemin Chi, Sylvain Calinon

OptimizationComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposed an Augmented Lagrangian Spectral Projected Gradient Descent (ALSPG) method based on geometric projection for efficient real-time motion planning of robots interacting with objects of different shapes.

Efficient Collision Detection for Long and Slender Robotic Links in Euclidean Distance Fields: Application to a Forestry Crane

Marc-Philip Ecker, Wolfgang Kemmetmüller

Computational EfficiencyRobotic IntelligencePoint CloudAgriculture Related

🎯 What it does: Proposed a collision detection algorithm for long and slender robotic arm links, significantly enhancing the computational efficiency of motion planning.

Efficient End-to-end Visual Localization for Autonomous Driving with Decoupled BEV Neural Matching

Jinyu Miao, Diange Yang

Pose EstimationAutonomous DrivingImage

🎯 What it does: Proposed an end-to-end visual localization network that directly estimates vehicle pose from surrounding images, and introduced a BEV-based decoupled neural matching pose solver.

Efficient Hitting with different links of a Redundant Robotic Manipulator

Harshit Khurana, Aude Billard

Robotic Intelligence

🎯 What it does: Built a non-grasping manipulation capability based on impact perception, utilizing the striking actions of a redundant manipulator by selecting appropriate arm links to interact with the environment to complete the striking task.

Efficient Human-Aware Task Allocation for Multi-Robot Systems in Shared Environments

Maryam Kazemi Eskeri, Tomasz Kucner

OptimizationRobotic Intelligence

🎯 What it does: This paper proposes a human-aware task allocation method (HATA) for multi-robot systems in shared environments, utilizing human motion pattern prediction to influence task execution time.

Efficient Instance Motion-Aware Point Cloud Scene Prediction

Yiming Fang, Huimin Lu

Autonomous DrivingComputational EfficiencyConvolutional Neural NetworkPoint Cloud

🎯 What it does: Proposed an instance motion-aware network called IMPNet for predicting future 3D point cloud scenes through historical LiDAR scans, explicitly integrating motion and instance-level information to enhance prediction accuracy.

Efficient Learning of A Unified Policy For Whole-body Manipulation and Locomotion Skills

Dianyong Hou, Yong Liu

Robotic IntelligenceReinforcement Learning

🎯 What it does: In scenarios integrating quadruped robots with robotic arms, a method is proposed to achieve unified whole-body control and locomotion strategy learning by embedding the explicit kinematic model of the robotic arm into a reinforcement learning framework.

Efficient Multimodal 3D Object Detector via Instance-Level Contrastive Distillation

Zhuoqun Su, Xieyuanli Chen

Object DetectionAutonomous DrivingTransformerContrastive LearningImageMultimodalityPoint Cloud

🎯 What it does: Propose a fast and effective multi-modal 3D object detector that combines the Instance-level Contrastive Distillation (ICD) framework and the Cross Linear Attention Fusion Module (CLFM).

Efficient Navigation Among Movable Obstacles using a Mobile Manipulator via Hierarchical Policy Learning

Taegeun Yang, Sung-Eui Yoon

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a hierarchical reinforcement learning framework for mobile manipulators, integrating obstacle attribute estimation with structured pushing strategies to achieve dynamic obstacle navigation while maintaining pre-planned paths;

Efficient Navigation for Quadruped Robots in Post-Disaster Scenarios

Christyan Cruz Ulloa, Antonio Barrientos

Computational EfficiencyRobotic Intelligence

🎯 What it does: Proposes a navigation framework for quadruped robots that combines a high-fidelity simulation environment with fuzzy logic control, designed for post-disaster search and rescue scenarios.

Efficient Prediction of Dense Visual Embeddings via Distillation and RGB-D Transformers

Söhnke Benedikt Fischedick, Horst-Michael Gross

SegmentationKnowledge DistillationRepresentation LearningTransformerVision Language ModelImageText

🎯 What it does: Propose DVEFormer, an efficient model based on RGB-D Transformer that predicts dense visual embeddings aligned with text through knowledge distillation.

Efficient Swept Volume-Based Trajectory Generation for Arbitrary-Shaped Ground Robot Navigation

Yisheng Li, Fu Zhang

Autonomous DrivingOptimizationRobotic IntelligenceBenchmark

🎯 What it does: Proposes a coarse-to-fine navigation framework for efficiently generating swept volume trajectories for ground robots with arbitrary shapes in complex environments, achieving continuous collision avoidance (CCA).

EfficientEQA: An Efficient Approach to Open-Vocabulary Embodied Question Answering

Kai Cheng, Aniket Bera

Computational EfficiencyRobotic IntelligenceVision Language ModelVision-Language-Action ModelImageTextMultimodalityRetrieval-Augmented Generation

🎯 What it does: Propose the EfficientEQA framework to address robot-assisted open-vocabulary embodied question answering tasks, integrating efficient exploration with free-text answer generation.

EffiTune: Diagnosing and Mitigating Training Inefficiency for Parameter Tuner in Robot Navigation System

Shiwei Feng, Xiangyu Zhang

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Propose the EffiTune framework, which improves the training efficiency of parameter tuners through robot behavior-guided diagnosis and targeted upsampling

EFFOcc: Learning Efficient Occupancy Networks from Minimal Labels for Autonomous Driving

Yining Shi, Diange Yang

SegmentationAutonomous DrivingKnowledge DistillationConvolutional Neural Network

🎯 What it does: Proposed the EFFOcc framework, which includes an efficient fused OccNet and a multi-stage occupancy-guided distillation, for training lightweight occupancy networks under scenarios with extremely few labels.

EIC Framework for Hand Exoskeletons Based on a Multimodal Large Language Model

Houcheng Li, Long Cheng

Robotic IntelligenceTransformerLarge Language ModelVision-Language-Action ModelImageMultimodalityAudio

🎯 What it does: Built a hand exoskeleton interactive control framework (EIC framework) based on multimodal large language models (MLLM), achieving the inference of user intent through the fusion of voice and image modalities, generating corresponding motion plans, and executing them to support undefined gestures and grasping actions.

Elastic Motion Policy: An Adaptive Dynamical System for Robust and Efficient One-Shot Imitation Learning

Tianyu Li, Nadia Figueroa

Robotic IntelligenceOrdinary Differential Equation

🎯 What it does: Propose an Elastic Motion Policy (EMP) framework based on a single demonstration, which can adjust the robot's behavior in real-time according to environmental changes while maintaining task specifications; adopt the dynamical systems paradigm, achieving motion planning and control through first-order differential equations, and verify its robustness and efficiency in dynamic environments, including obstacle avoidance and multi-step tasks, in real-world robot experiments;

ELMAR: Enhancing LiDAR Detection with 4D Radar Motion Awareness and Cross-modal Uncertainty

Xiangyuan Peng, Robert Wille

Object DetectionAutonomous DrivingMultimodalityPoint Cloud

🎯 What it does: Develop a LiDAR detection framework that utilizes 4D radar motion information and cross-modal uncertainty enhancement

ELPTNet: An Efficient LiDAR-based 3D Pedestrian Tracking Network for Autonomous Navigation Social Robots

Jinzheng Guang, Jingtai Liu

Object TrackingAutonomous DrivingPoint CloudBenchmark

🎯 What it does: Proposed ELPTNet, a LiDAR-based efficient 3D pedestrian tracking network for autonomous navigation social robots, achieving real-time accurate tracking.

Embodied Domain Adaptation for Object Detection

Xiangyu Shi, Feras Dayoub

Object DetectionDomain AdaptationContrastive LearningBenchmark

🎯 What it does: Proposed a Source-Free Domain Adaptation method, which refines pseudo labels through temporal clustering, employs multi-scale threshold fusion, and integrates the Mean Teacher framework with contrastive learning to adapt object detection in dynamic indoor environments.

Embodied Escaping: End-to-End Reinforcement Learning for Robot Navigation in Narrow Environment

Han Zheng, Ming Yang

Robotic IntelligenceReinforcement LearningPoint CloudTime Series

🎯 What it does: Proposes an embodied escape model based on reinforcement learning, combined with an effective action mask to enable robots to escape from dead-ends in narrow environments, and experiments are conducted on real robots for validation.

Embodied Instruction Following in Unknown Environments

Zhenyu Wu, Haibin Yan

Representation LearningRobotic IntelligenceLarge Language ModelVision-Language-Action ModelWorld ModelMultimodality

🎯 What it does: Proposed an embodied instruction following method for unknown environments, adopting a hierarchical framework comprising a high-level task planner and a low-level exploration controller, and utilizing a multimodal large language model to construct a semantic representation graph.

Embodied multi-modal sensing with a soft modular arm powered by physical reservoir computing

Jun Wang, Suyi Li

Robotic IntelligenceMultimodality

🎯 What it does: Embed simple bending strain gauges in a soft modular arm to achieve multimodal perception through physical reservoir computing.

EmbodiedAgent: A Scalable Hierarchical Approach to Overcome Practical Challenge in Multi-Robot Control

Hanwen Wan, Xiaoqiang Ji

Robotic IntelligenceAgentic AITextBenchmark

🎯 What it does: Designed a scalable hierarchical framework, EmbodiedAgent, for heterogeneous multi-robot control, integrating next-step action prediction and structured memory systems to decompose tasks into executable robot skills and dynamically verify action compliance with environmental constraints.

Embracing Dynamics: Dynamics-aware 4D Gaussian Splatting SLAM

Zhicong Sun, Jinxing Hu

Autonomous DrivingOptimizationGaussian SplattingSimultaneous Localization and MappingVideo

🎯 What it does: Propose a dynamic SLAM method called D4DGS-SLAM based on 4D Gaussian Splatting, which utilizes the time dimension to achieve high-quality reconstruction of dynamic scenes and filters unstable points through a dynamic perception module.

Emergency Avoidance: Model Predictive Control Based Path Tracking for Unmanned Ground Vehicles with Active Obstacle Avoidance

Zongliang Chen, Xiaocong Li

Autonomous DrivingOptimization

🎯 What it does: Proposed a model predictive control-based active obstacle avoidance path tracking method called SOAMPC, integrating safety distance constraints to achieve real-time obstacle avoidance for UGVs.

Emergent Cooperative Strategies for Pursuit-Evasion in Cluttered Environments: A Knowledge-Enhanced Multi-Agent Deep Reinforcement Learning Approach

Yihao Sun, Jie Jiang

Reinforcement Learning

🎯 What it does: Propose a knowledge-enhanced multi-agent deep reinforcement learning method for pursuit-evasion tasks in complex environments, utilizing a team reward function to encourage collaboration.

EmoRLTalk: Speech-Driven Emotional Facial Animation With Offline Reinforcement Learning

Gaofeng Liu, Tao Fang

GenerationReinforcement LearningDiffusion modelVideoAudio

🎯 What it does: Introduce the EmoRLTalk framework, integrating offline reinforcement learning, conditional diffusion models, multi-task learning, and ControlNet to achieve voice-driven emotional facial animation;

Empirical Analysis of Sim-and-Real Cotraining of Diffusion Policies For Planar Pushing from Pixels

Adam Wei, Russ Tedrake

Domain AdaptationRobotic IntelligenceDiffusion modelImage

🎯 What it does: This paper investigates the method of using simulation and real data co-training in planar pushing tasks, and systematically analyzes its principles through large-scale experiments.

Enabling On-Chip Adaptive Linear Optimal Control via Linearized Gaussian Process

Yuan Gao, Yini Fang

Optimization

🎯 What it does: A linearized Gaussian Process (GP) is used to model external aerodynamic forces, combined with a linear Model Predictive Control (MPC) to achieve real-time computability, while designing an active data acquisition strategy based on Bayesian optimization and additive GP to reduce performance loss caused by linearization and the computational cost of the GP on the chip.

Encoding Robot Behavior as Sensory-Based Adaptation of Learned Skillful Trajectories

Jonathan Madera, A. M. Fey

Robotic Intelligence

🎯 What it does: Motion control of a dual-arm surgical robot by combining a learning control strategy based on Gaussian Mixture Models (GMM) with a reactive control strategy triggered by online perception

EndoMUST: Monocular Depth Estimation for Robotic Endoscopy via End-to-end Multi-step Self-supervised Training

Liangjing Shao, Xinrong Chen

Depth EstimationRobotic IntelligenceSupervised Fine-TuningOptical FlowImageBiomedical Data

🎯 What it does: Propose a multi-step efficient fine-tuning framework for end-to-end training in robotic endoscopy monocular depth estimation, where the training process is divided into three steps: optical flow registration, multi-scale image decomposition, and multi-transformation alignment.

Energy-constrained multi-robot exploration for autonomous map building

Sambhu H. Karumanchi, Abraham P. Vinod

Robotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Exploring unknown environment map building using multi-robot with limited battery power

Energy-Efficient Motion Planner for Legged Robots

Alexander Schperberg, S. Cairano

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Developed an online gait planner focused on energy efficiency, utilizing a set of foot placements based on the robot's body position to determine when and how to execute steps.

Energy-Efficient Obstacle Avoidance via Iterative B-Spline Optimization for a Mobile Manipulator in Dynamic Environments

Kai-Tai Song, Yuan-Shuo Hsieh

OptimizationRobotic Intelligence

🎯 What it does: Proposes a collision avoidance system for autonomous mobile manipulators in dynamic environments, achieving low energy consumption by generating safe and smooth paths through B-Spline iterative optimization;

Energy-Efficient Omnidirectional Locomotion for Wheeled Quadrupeds via Predictive Energy-Aware Nominal Gait Selection

Xu Yang, Yilin Mo

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Propose a hierarchical control framework that optimizes the omnidirectional walking energy efficiency of wheeled quadruped robots through a power prediction network and residual reinforcement learning.

Engaging Mind and Body: An Immersive BCI Paradigm with Motion-Panoramic Virtual Reality

Lianchi Zhang, Hong Cheng

MultimodalityBiomedical Data

🎯 What it does: Proposed and verified a sensory immersive brain-computer interface (BCI) paradigm based on motion panoramic virtual reality for decision-making, integrating visual, auditory, and motor multimodal stimuli, and implemented a modified extreme game through the Gait Real-time Analysis Interactive Lab system.

Enhanced Bias Correction for Fiber optic gyroscope Using an Improved Artificial Bee Colony Algorithm

Jinyue Zhao, Sixu Huang

OptimizationPhysics Related

🎯 What it does: Propose an improved artificial bee colony algorithm for bias correction of fiber optic gyroscopes in inertial navigation systems.

Enhanced Kinematic Calibration of a 4PPa-2PaR Parallel Manipulator with Subchains

Jingbo Luo, Guilin Yang

OptimizationRobotic Intelligence

🎯 What it does: Proposes a kinematic calibration method for a 4PPa-2PaR parallel manipulator sub-chain architecture based on a virtual chain.

Enhanced Motion Forecasting with Plug-and-Play Multimodal Large Language Models

Katie Luo, Mingxing Tan

Autonomous DrivingTransformerLarge Language ModelPrompt EngineeringMultimodality

🎯 What it does: Propose Plug-and-Forecast (PnF), a plug-and-play approach to inject multimodal large language models (MLLMs) into existing motion prediction models;

Enhanced Precession of a Magnetic Helical Microbot in a Viscoelastic Gel *

Meng Zhang, T. Qiu

Robotic IntelligencePhysics Related

🎯 What it does: Studied the motion of magnetic helical microrobots in viscoelastic gels, observing that the robots rupture the gel and generate 3D helical trajectories, with the precession angle increasing with rotational speed.

Enhanced Robotic Navigation in Deformable Environments using Learning from Demonstration and Dynamic Modulation

Lingyun Chen, Sami Haddadin

Robotic Intelligence

🎯 What it does: Proposes a robot navigation method that utilizes learning from demonstration and dynamic modulation in deformable environments.

Enhanced Rolling Motion of Magnetic Microparticles by Turning Interface Lubrication

Yuke Li, Xiaoming Liu

Robotic IntelligencePhysics Related

🎯 What it does: Designed and experimentally tested a magnetic micro-roller robot driven by a rotating AC magnetic field, studying the relationship between lubrication film thickness and translational speed when rolling on a surface, and controlling the lubrication film thickness and speed by applying a gradient magnetic field.

Enhancing Autonomous Driving Safety with Collision Scenario Integration

Zi Wang, José M. Álvarez

Autonomous DrivingPrompt Engineering

🎯 What it does: Propose the SafeFusion framework that trains using collision data, integrates safety metrics to achieve collision avoidance learning, and introduces CollisionGen, a data generation pipeline for creating diverse collision scenarios

Enhancing Context-Aware Human Motion Prediction for Efficient Robot Handovers

Gerard G'omez-Izquierdo, A. Garrell

Computational EfficiencyRobotic Intelligence

🎯 What it does: Utilizing the lightweight architecture siMLPe with key improvements, enhancing human motion prediction in handover tasks;

Enhancing Continuum Robot Mobility: Design and Control with Integrated Dual Rotational DOFs *

Peikang Yuan, Rongjie Kang

Robotic Intelligence

🎯 What it does: Designed and implemented a continuous robot with dual rotational degrees of freedom (DOF), capable of compensating for torsional deformation and achieving 6-DOF control of the end effector.

Enhancing Deep Reinforcement Learning-based Robot Navigation Generalization through Scenario Augmentation

Shanze Wang, Wei Zhang

Data-Centric LearningRobotic IntelligenceReinforcement Learning

🎯 What it does: Proposes the Scenario Augmentation technique to enhance the generalization performance of deep reinforcement learning-driven robot navigation in unseen environments.

Enhancing Humanoid Robot Dynamics: An Optimization Framework for Shoulder Base Angle Adjustment

Jiwon Yoon, Y. Ihn

OptimizationRobotic Intelligence

🎯 What it does: Proposes a multi-objective optimization framework to enhance the dynamic performance of humanoid robot arms by optimizing the initial shoulder angle.

Enhancing Large Vision Model in Street Scene Semantic Understanding through Leveraging Posterior Optimization Trajectory

Wei-Bin Kou, Yik-Chung Wu

Autonomous DrivingOptimizationTransformerImage

🎯 What it does: Adopt a pre-trained large visual model (LVM) as the backbone, combined with downstream perception heads to enhance semantic understanding in autonomous driving, while proposing a Posterior Optimization Trajectory (POT)-guided optimization scheme (POTGui) and a POT generator (POTGen) to achieve rapid convergence.

Enhancing Multi-Task Motion Planning Based on Improved DMPs for Lower Limb Prostheses

Honglei An, Hongxu Ma

Robotic IntelligenceTime SeriesSequentialBiomedical Data

🎯 What it does: A novel multi-task gait planning method based on dynamic movement primitives (DMP) and DMPs-SVD is proposed for lower-limb prosthetics.

Enhancing Navigational Scene Understanding using Integrated Language Models in Maritime Environments

Yeongha Shin, Jinwhan Kim

TransformerLarge Language ModelVision Language ModelMultimodality

🎯 What it does: Proposes an algorithm that leverages large language models (LLM) and vision-language models (VLM) to achieve navigation scene understanding in complex maritime environments and generate safe navigation cost maps.

Enhancing Object Search in Indoor Spaces via Personalized Object-Factored Ontologies

Akash Chikhalikar, Yasuhisa Hirata

Robotic Intelligence

🎯 What it does: Proposes a framework enabling robots to infer personalized ontologies of indoor environments, and introduces an adaptive reasoning strategy integrated with dynamic belief updates to enhance multi-object search tasks.

Enhancing Robustness in Language-Driven Robotics: A Modular Approach to Failure Reduction

Émiland Garrabé, Stéphane Doncieux

Robotic IntelligenceTransformerLarge Language Model

🎯 What it does: Proposed a modular architecture to enhance the robustness of locally executable LLMs in robotics, addressing localization and alignment with robot capabilities, while introducing an 'expected outcome' module and a real-time error recovery feedback mechanism.

Enhancing Single Image to 3D Generation using Gaussian Splatting and Hybrid Diffusion Priors

Hritam Basak, Zhaozheng Yin

GenerationData SynthesisDiffusion modelGaussian SplattingImage

🎯 What it does: This paper proposes a two-stage frequency distillation loss based on high-frequency and low-frequency fusion, integrating the low-frequency geometric prior from 3D diffusion models and high-frequency texture details from 2D diffusion models into a Gaussian splatting framework, achieving high-quality 3D generation from a single image.

Enhancing Tactile Sensing in Robotics Using Null-Space Diffusion Model with EIT-based Sensors

Qilin Zhang, Xiaojie Wang

Robotic IntelligenceDiffusion modelBiomedical Data

🎯 What it does: Proposed a resistance impedance tomography (EIT) reconstruction method based on the hole space diffusion model (NSDM) to improve the spatial resolution and image quality of robot tactile sensors.

Enhancing the Flexibility of a Quadruped Robot with a 2-DOF Active Spine Using Nonlinear Model Predictive Control

Zeyi Yang, Hui Cheng

OptimizationRobotic Intelligence

🎯 What it does: This paper integrates a 2-DOF active spine on the quadruped robot Yatsen Lion II and employs nonlinear model predictive control (NMPC) combined with centroidal dynamics and full kinematics to optimize its motion.

Enhancing UAV Energy Efficiency and Versatility through Trimodal Ground, Hovering, and Fixed-Wing Locomotion Modes

Afonso Vale, F. Afonso

🎯 What it does: Proposed and implemented a tri-modal drone that integrates ground locomotion, hovering, and fixed-wing flight, utilizing a shared actuator system and achieving ground control within the ArduPilot framework.

Env-Mani: Quadrupedal Robot Loco-Manipulation with Environment-in-the-Loop

Yixuan Li, Wei Liang

Robotic IntelligenceReinforcement Learning

🎯 What it does: A unified learning-based quadruped robot motion-manipulation framework is proposed, enabling the robot to utilize external environments as support to expand the workspace and enhance manipulation capabilities. The framework generates whole-body motions to complete object manipulation using limited onboard sensors and proprioceptive inputs, guided by reward design and curriculum learning.

ESCoT: An Enhanced Step-based Coordinate Trajectory Planning Method for Multiple Car-like Robots

Junkai Jiang, Jianqiang Wang

Autonomous DrivingOptimization

🎯 What it does: Proposed an enhanced stepping coordinate trajectory planning method called ESCoT for multi-vehicle robotic systems.

ESFUSION: Enhanced LiDAR-camera Fusion Architecture for HD Mapping at Intersection

Suhui Yang, Jingjing Cui

Autonomous DrivingImageMultimodalityPoint Cloud

🎯 What it does: Proposes the ESFusion method for generating high-precision maps at intersections by integrating LiDAR and camera multimodal data

Estimating Continuum Robot Shape under External Loading using spatiotemporal Neural Networks

Enyi Wang, Jianwei Zhang

Robotic IntelligenceRecurrent Neural NetworkImageMultimodalityPoint CloudTime Series

🎯 What it does: Proposed a learning method based on spatiotemporal neural networks, using multimodal inputs (tendon displacement and RGB images) to estimate the 3D shape of flexible continuous robots under external forces.

Estimation of Aerodynamics Forces in Dynamic Morphing Wing Flight

Bibek Gupta, Alireza Ramezani

Time SeriesPhysics Related

🎯 What it does: Studied two aerodynamic force estimation methods on the Aerobat platform: a physics-based observer based on Hamiltonian mechanics and a neural network regression model based on MLP.

ET-Former: Efficient Triplane Deformable Attention for 3D Semantic Scene Completion From Monocular Camera

Jing Liang, Dinesh Manocha

Autonomous DrivingTransformerAuto EncoderImage

🎯 What it does: Proposed ET-Former, which achieves 3D semantic scene completion using a monocular camera and generates semantic occupancy maps and uncertainty estimates.

ET-Plan-Bench: Embodied Task-level Planning Benchmark Towards Spatial-Temporal Cognition with Foundation Models

Lingfeng Zhang, Jianye Hao

Large Language ModelBenchmark

🎯 What it does: Proposed a new embodied task planning benchmark called ET-Plan-Bench to evaluate LLMs' understanding of spatial, temporal, and causal relationships

ETA-IK: Execution-Time-Aware Inverse Kinematics for Dual-Arm Systems

Yucheng Tang, Björn Hein

OptimizationComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposes an execution time-aware inverse kinematics method (ETA-IK) for dual-arm systems, aiming to optimize motion execution time by leveraging the system's redundancy

ETA: Learning Optical Flow with Efficient Temporal Attention

Bo Wang, Dewen Hu

Optical FlowVideo

🎯 What it does: Proposed a multi-frame information integration method that utilizes multi-frame information fusion, employing an attention mechanism to integrate temporal information from the previous frame.

ETO+: Revisit the Refinement Stage in Efficient Feature Matching

Junjie Ni, Guofeng Zhang

Depth EstimationConvolutional Neural NetworkTransformerImage

🎯 What it does: Proposed ETO+, an algorithm introducing a lightweight bidirectional interaction module and multi-stage refinement in feature matching, enhancing matching accuracy and real-time performance.

Evaluating Computational Approaches to Metabolic Cost Estimation in Gait Assistance with a Passive Exosuit*

Vahid Firouzi, Maziar Ahmad Sharbafi

Biomedical Data

🎯 What it does: Evaluated the effectiveness of different computational methods in estimating metabolic energy expenditure during gait assistance.

Evaluating Generative Models for Inverse Kinematics of Concentric Tube Robots

Paul H. Kang, D. Podolsky

GenerationRobotic IntelligenceFlow-based Model

🎯 What it does: This paper addresses the inverse kinematics problem for cylindrical pipe robots (CTR) by proposing a diversity evaluation framework based on the workspace, and experiments are conducted to validate three generative models (INN, cINN, cVAE).

Evaluating Human-Robot Collaboration through Online Video: Perspective Matters

Leimin Tian, Dana Kulic

Robotic IntelligenceVideoText

🎯 What it does: This paper conducted an online video comparative study (N=178) to evaluate the performance of three robot delivery strategies (adaptive autonomous strategy, non-adaptive scripted strategy, and teleoperation) in collaborative assembly tasks.

Evaluating Robot Program Performance with Power Consumption–Driven Metrics in Lightweight Industrial Robots

Juan Heredia, M. Kjærgaard

Robotic Intelligence

🎯 What it does: Proposed an evaluation framework based on robot power consumption, measuring the efficiency and reliability of robot programs through analysis of power curves and a set of normalized indicators (fU, fC, fR, and α), with experimental comparisons of four different strategies on the UR5e robot during a machining material lifting task.

Evaluating the Pre-Dressing Step: Unfolding Medical Garments via Imitation Learning

David Blanco-Mulero, Carme Torras

ClassificationRobotic IntelligenceReinforcement LearningImage

🎯 What it does: Proposed a pre-opening step before robot-assisted dressing, which learned three manipulation primitives (including high and low acceleration actions) through imitation learning and used a visual classifier to identify clothing states (closed, partially open, fully open).

Evaluation and Analysis of Precision Leaf Pruning End-Effectors Within Dense Foliage Agriculture *

Quinlan T. Barthelme, Chris Lehnert

Robotic IntelligenceAgriculture Related

🎯 What it does: Evaluated the performance of three leaf-cutting end-effectors (scissors-type, curved-type, and vacuum-type) in actual unmodified dense foliage environments for leaf-cutting operations.

Event-based Depth from Focus

Wenjie Xue, Limin Shang

Depth Estimation

🎯 What it does: Propose an event depth focusing (EDFF) method based on event cameras and liquid crystal lenses, and implement a prototype system