arXivSub Start free trial

IROS 2025 Papers — Page 13

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

Motion Planning and Control with Unknown Nonlinear Dynamics through Predicted Reachability

Zhiquan Zhang, M. Ornik

OptimizationRobotic IntelligenceGraph

🎯 What it does: Proposed a hybrid planning-control framework by partitioning the state space, approximating the system as a piecewise affine (PWA) system, and abstracting it as a directed weighted graph. The graph edges are incrementally updated using affine system identification and reachability control theory. Heuristic weights are assigned based on predictive reachability conditions and edge determinism. During task execution, data is adaptively collected and analyzed to update the predictive graph in real-time. Controllers are synthesized online based on graph search results, and the method's effectiveness was validated in a simulation scenario involving mobile robots navigating unknown terrain.

Motion-Feat: Motion Blur-Aware Local Feature Description for Image Matching

Ye Gao, Jinhu Lu

Pose EstimationRetrievalConvolutional Neural NetworkOptical FlowImage

🎯 What it does: Proposed Motion-Feat, an end-to-end motion blur-aware local feature descriptor method.

MotionScript: Natural Language Descriptions for Expressive 3D Human Motions

Payam Jome Yazdian, Angelica Lim

GenerationLarge Language ModelTextMultimodality

🎯 What it does: Develop the MotionScript framework, which generates detailed descriptions of 3D human actions in natural language and uses these descriptions to train text-to-action models.

Motivational Cognitive Maps Allow Robot Biomimetic Autonomy

Oscar Guerrero-Rosado, P. Verschure

Robotic IntelligenceAuto EncoderMultimodality

🎯 What it does: Developed a hippocampus-inspired model called Motivational Hippocampal Autoencoder (MoHA), integrating intrinsic motivation and extrinsic visual information to simulate the firing characteristics of hippocampal place cells under different motivational states, and enable synthetic agents to learn and deploy efficient trajectories in foraging tasks.

Moving Object Segmentation via 3D LiDAR Data: A Learning-Free Real-time Online Alternative

Zinuo Yi, Darius Burschka

SegmentationPoint Cloud

🎯 What it does: Proposed a real-time online, non-learning 3D LiDAR moving object segmentation method

MovSAM: A Single-image Moving Object Segmentation Framework Based on Deep Thinking

Chang Nie, Hesheng Wang

SegmentationTransformerLarge Language ModelPrompt EngineeringImageMultimodalityChain-of-Thought

🎯 What it does: Propose the MovSAM framework to achieve moving object segmentation in single images.

MPC-based Deep Reinforcement Learning Method for Space Robotic Control with Fuel Sloshing Mitigation

Mahya Ramezani, Holger Voos

Robotic IntelligenceReinforcement LearningPhysics Related

🎯 What it does: Proposed a framework combining reinforcement learning (RL) with model predictive control (MPC) for satellite docking control of partially filled fuel tanks, validated its effectiveness on the Zero-G experimental platform and high-precision numerical simulations.

MPDG-SLAM: Motion Probability-Based 3DGS-SLAM in Dynamic Environment

Conghao Huang, Mingrui Li

Object DetectionConvolutional Neural NetworkGaussian SplattingSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes MPDG-SLAM, a 3D Gaussian point cloud rendering SLAM method based on motion probability (MP), integrating mobile-deployable YOLO for dynamic object detection, and assigning labels to Gaussian points via MP attributes, mapping dynamic points to the frontend feature tracking system to correct dynamic interference.

Mr. Virgil: Learning Multi-robot Visual-range Relative Localization

Si Wang, Yue Wang

OptimizationRobotic IntelligenceGraph Neural NetworkSimultaneous Localization and MappingMultimodalityGraph

🎯 What it does: Proposes an end-to-end learning framework named Mr. Virgil, which uses graph neural networks for data association between UWB ranging and visual detection, and combines differentiable pose graph optimization (PGO) to achieve relative localization within the visual range of multi-robot systems.

MRMT-PR: A Multi-Scale Reverse-View Mamba-Transformer for LiDAR Place Recognition

Kan Luo, Xieyuanli Chen

RecognitionTransformerPoint Cloud

🎯 What it does: Proposed a multi-scale inverse perspective Mamba-Transformer architecture MRMT-PR for place recognition based on LiDAR single-frame point clouds.

MRS-CWC: A Weakly Constrained Multi-Robot System with Controllable Constraint Stiffness for Mobility and Navigation in Unknown 3D Rough Environments

Runze Xiao, H. Asama

Robotic Intelligence

🎯 What it does: Proposed a tunable weakly constrained multi-robot system (MRS-CWC) that achieves a balance between flexibility and maneuverability among robots through dynamically adjustable constraint stiffness, enabling navigation in unknown 3D rugged environments.

MSPA-LIO: LiDAR-Inertial Odometry with Multi-Scale Plane Adjustment

Su Yan, Song Wu

Autonomous DrivingSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposed MSPA-LIO, a LiDAR-Inertial odometry method with multi-scale plane adjustment; by applying geometric constraints and plane adjustment at both voxel plane scale and large-scale plane scale, and employing voxel-based explicit large plane extraction to fully utilize planar structures in the environment, using large planes to correct the orientation of related voxel planes, thereby improving the consistency between pose and map.

MT-Fusion: Multi-Task Learning for Degradation-Aware Infrared and Visible Image Fusion

Yuyang Gao, Kenji Hashimoto

RestorationRobotic IntelligenceImageMultimodality

🎯 What it does: Proposed an MT-Fusion deep learning workflow based on multi-task learning for fusing infrared and visible light images under various degradation scenarios, and automatically identifying degradation types to select the corresponding encoder.

Multi-Agent Combinatorial Path Finding for Tractor-Trailers in Occupancy Grids

Xuemian Wu, Zhongqiang Ren

Autonomous DrivingOptimization

🎯 What it does: Propose and solve the multi-agent cooperative path planning problem, conducting collision-free path search and optimizing arrival time for the tractor-trailer model

Multi-Agent Inverse Reinforcement Learning in Real World Unstructured Pedestrian Crowds

Rohan Chandra, Joydeep Biswas

Reinforcement LearningSequential

🎯 What it does: Proposed a multi-agent maximum entropy inverse reinforcement learning algorithm tailored for real-world unstructured crowds;

Multi-Agent Pickup and Delivery with Mobile Pickups

Benedetta Flammini, Bruno Lacerda

Optimization

🎯 What it does: Proposed an extended algorithm called TP-EL based on Token Passing to handle item exchange between suppliers and deliverers, addressing the multi-agent pickup and delivery problem (MAPD) with mobile pickup.

Multi-Agent Reinforcement Learning Guided by Signal Temporal Logic Specifications

Jiangwei Wang, Fei Miao

Reinforcement Learning

🎯 What it does: Proposed a multi-agent reinforcement learning framework based on Signal Temporal Logic (STL), guiding the learning process by using the robustness value of STL specifications for reward design.

Multi-Agent Reinforcement Learning with Transformer-based Spatio-temporal Fusion for Autonomous Driving in Mixed Traffic

Rixin Li, Tiantian Xu

Autonomous DrivingTransformerReinforcement LearningContrastive Learning

🎯 What it does: A multi-agent reinforcement learning method combining a Transformer-based spatiotemporal fusion module (TSTF) with auxiliary contrastive learning tasks is studied for autonomous driving decision-making in mixed traffic environments

Multi-Cali Anything: Dense Feature Multi-Frame Structure-from-Motion for Large-Scale Camera Array Calibration

Jinjiang You, W. Pu

OptimizationImage

🎯 What it does: Propose a multi-frame structured light reconstruction (SfM) camera array calibration method based on dense features, which directly optimizes camera intrinsic parameters from scene data without requiring additional calibration pattern capture.

Multi-Functional Granular Propulsion: Bio-Inspired Orientation Control and Local Fluidization for Crawl-To-Dig Transitions

Dongting Li, Nick Gravish

Robotic IntelligencePhysics Related

🎯 What it does: Adding adjustable body anchors to existing spiral propulsion vehicles (SPV) to achieve directional control and switching between horizontal crawling and vertical digging, and studying the impact of localized liquefaction technology on performance.

Multi-Material 3D-Printed Magnetic Millirobot for Quadrupedal Locomotion in Endoluminal Spaces

Ruichen Wang, Dong Wang

Robotic Intelligence

🎯 What it does: Developed a four-legged magnetic microrobot (beamrobot) fabricated using multi-material direct ink writing (DIW), featuring a flexible body equipped with magnetic feet that enable shape transformation and motion under an external magnetic field;

Multi-Modal Graph Convolutional Network with Sinusoidal Encoding for Robust Human Action Segmentation

Hao Xing, Gordon Cheng

SegmentationGraph Neural NetworkMultimodality

🎯 What it does: Proposes a multi-modal graph convolutional network (MMGCN) that fuses low-frame-rate visual data with high-frame-rate motion data (skeleton and object detection) to achieve robust human action temporal segmentation.

Multi-Objective Optimization of Humanoid Robot Hardware and Control for Multiple Tasks via Genetic Algorithms

Carlotta Sartore, Daniele Pucci

OptimizationRobotic Intelligence

🎯 What it does: Jointly optimize the hardware structure and hierarchical control parameters of a humanoid robot using Non-dominated Sorting Genetic Algorithm II (NSGA-II) to enhance multi-task performance in walking and carrying.

Multi-PrefDrive: Optimizing Large Language Models for Autonomous Driving Through Multi-Preference Tuning

Yun Li, Manabu Tsukada

Autonomous DrivingTransformerLarge Language ModelSupervised Fine-Tuning

🎯 What it does: Proposed the Multi-PrefDrive framework, which enhances autonomous driving performance based on large language models by pairing each selected action with multiple rejected alternative actions and employing the Plackett-Luce preference model for multi-dimensional preference tuning.

Multi-Robot Assembly of Deformable Linear Objects Using Multi-Modal Perception

Kejia Chen, Rüdiger Daub

Robotic IntelligenceMultimodality

🎯 What it does: Propose a multi-robot assembly framework based on multi-modal perception, completing the entire process from picking up to assembly;

Multi-Robot Coordination in an Adversarial Graph-Traversal Game

J. Berneburg, Daigo Shishika

OptimizationRobotic IntelligenceReinforcement LearningGraph

🎯 What it does: Studies coordinated behaviors of robots in adversarial graph traversal games, establishes a non-cooperative stochastic game model, solves for Nash equilibrium, and conducts theoretical analysis to provide performance bounds.

Multi-Robot Ergodic Trajectory Optimization with Relaxed Periodic Connectivity

Yongce Liu, Zhongqiang Ren

OptimizationRobotic Intelligence

🎯 What it does: Proposes a multi-robot trajectory planning method combining intermittent connectivity maintenance with traversable search for information gathering tasks.

Multi-Robot Motion Planning with Cooperative Localization

Anne Theurkauf, Morteza Lahijanian

Robotic Intelligence

🎯 What it does: Proposes a multi-robot cooperative localization motion planning problem under the presence of motion and measurement noise, and presents an algorithm with safety guarantees

Multi-Sets Trees (MST*): Accelerated Asymptotically Optimal Motion Planning Optimization Informed by Multiple Domain Subsets

Liding Zhang, A. Knoll

OptimizationRobotic Intelligence

🎯 What it does: Proposed and implemented a sampling-based motion planner based on multi-set tree (MST*), utilizing GuILD sets and Beacon selectors to accelerate path search and optimization;

Multi-Step Deep Koopman Network (MDK-Net) for Vehicle Control in Frenet Frame

Mohammad Abtahi, S. Nazari

Autonomous DrivingOptimization

🎯 What it does: Propose a deep learning-based Koopman model that describes the full dynamics of a vehicle from pedal and steering inputs to chassis states in the Frenet coordinate system, and apply it to MPC control.

Multi-target Association and Localization with Distributed Drone Following: A Factor Graph Approach

Kaixiao Ye, Tao Yang

Object TrackingAutonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: A factor graph-based multi-target association and localization method is proposed, integrating sensor measurements and control constraints from distributed UAVs to achieve joint association and localization of multiple targets. The effectiveness and robustness of the method are validated through simulations and real-world experiments.

Multi-UAV Deployment in Obstacle-Cluttered Environments with LOS Connectivity

Yuda Chen, Meng Guo

Optimization

🎯 What it does: Propose an efficient restricted search method based on minimal edge RRT* to find a spanning tree topology requiring the fewest drones; and propose a distributed model predictive control (MPC) strategy to achieve online motion coordination, ensuring collision safety and line-of-sight connectivity among drones.

Multi-UAV Formation Control with Static and Dynamic Obstacle Avoidance via Reinforcement Learning

Yuqing Xie, Yu Wang

Robotic IntelligenceTransformerReinforcement Learning

🎯 What it does: A two-phase reinforcement learning process is used to achieve multi-drone formation control, with obstacle avoidance implemented in both static and dynamic obstacle environments.

Multi-UAV-UGV Collision-Free Tracking Control via Control Barrier Function-Based Reinforcement Learning

Haojie Xia, Housheng Su

Autonomous DrivingOptimizationReinforcement Learning

🎯 What it does: Propose a three-tier hierarchical control scheme to achieve multi-unmanned aerial vehicle (UAV)-unmanned ground vehicle (UGV) cooperative tracking, including feature matching, reinforcement learning (RL) real-time tracking, and collision avoidance based on control barrier functions.

Multi-View Normal and Distance Guidance Gaussian Splatting for Surface Reconstruction

Bo Jia, Lin Cao

Gaussian SplattingImage

🎯 What it does: Propose a multi-view regularized Gaussian Splatting method, achieving geometric depth unification and high-precision surface reconstruction for small indoor and outdoor scenes through a distance projection re-projection regularization module and a multi-view normal enhancement module.

Multimodal Anomaly Detection with a Mixture-of-Experts

C. Willibald, Dongheui Lee

Anomaly DetectionRobotic IntelligenceMixture of ExpertsVision Language ModelMultimodality

🎯 What it does: Proposed a hybrid expert framework that fuses vision-language models and Gaussian Mixture Regression detectors to achieve multimodal anomaly detection

Multimodal Autonomous Robotic Long-Horizon Task Planning via Embodied Language Model and Behavior Trees

Hongpeng Chen, Pai Zheng

Robotic IntelligenceTransformerLarge Language ModelVision Language ModelMultimodality

🎯 What it does: Propose a multimodal framework based on large language models and vision-language models for planning in long-term robotic manipulation tasks.

Multimodal Deformation Estimation of Soft Pneumatic Gripper During Operation

Changheng Cai, Yuan Gao

Representation LearningRobotic IntelligenceMultimodality

🎯 What it does: Proposed a multi-modal learning-based sensing method that integrates a camera and inertial measurement unit (IMU) for real-time perception of whole-body deformation of soft pneumatic grippers.

Multimodal Human Activity Recognition with a Large Language Model for Enhanced Human-Robot Interaction

G. Khodabandelou, Y. Amirat

RecognitionRobotic IntelligenceTransformerLarge Language ModelMultimodality

🎯 What it does: Propose a framework that unifies all sensor streams (visual, audio, inertial) into the text domain and directly uses GPT-3 for multimodal activity recognition;

Multimodal Integrated Prediction and Decision-making with Adaptive Interaction Modality Explorations

Tong Li, Shaojie Shen

Autonomous DrivingWorld ModelMultimodality

🎯 What it does: Proposed the MIND framework, which can simultaneously generate predictions and decisions for multi-modal interactions.

Multimodal Obstacle Detection and Adaptive Neural Control for Autonomous Drones

Theerawath Phetpoon, P. Manoonpong

Object DetectionAutonomous DrivingRobotic IntelligenceImageMultimodalityPoint Cloud

🎯 What it does: Proposed an integrated multi-modal obstacle detection and adaptive neural control unmanned aerial vehicle (UAV) system capable of autonomous navigation during both day and night.

Multimodal Point Cloud Registration Method Based on Centerline-Guided Expansion and Contraction: An Optimization Strategy Applied in Bronchial Lumen Map Building

Le Ren, Lining Sun

OptimizationSimultaneous Localization and MappingMultimodalityPoint CloudComputed Tomography

🎯 What it does: Proposes a multi-modal point cloud registration method that utilizes CT and intraoperative video frames, leveraging centerline guidance for bronchial cavity modeling.

Multimodal Upstream Motion of Magnetically Controlled Micro/Nano Robots in High-Viscosity Fluids

Chan Li, Lin Feng

Robotic IntelligenceBiomedical DataPhysics Related

🎯 What it does: Study the effects of high blood viscosity and blood cell interference on the upstream movement of magnetic microrobots, combining theoretical modeling, simulation, and experimental validation, and derive a velocity formula for microrobots in non-Newtonian fluids.

MultiNash-PF: A Particle Filtering Approach for Computing Multiple Local Generalized Nash Equilibria in Trajectory Games

Maulik Bhatt, Negar Mehr

OptimizationComputational Efficiency

🎯 What it does: Propose the MultiNash-PF algorithm, which uses the particle filter method to compute multiple local generalized Nash equilibria in multi-objective trajectory games.

Multiphysics Model and Hysteresis Compensation for Control of Electroactive Actuators

Ferradj Imane, N. Mechbal

OptimizationPhysics Related

🎯 What it does: Propose a model-based control architecture that combines an analytically inverted generalized Prandtl-Ishlinskii hysteresis model with model predictive control (MPC) to linearize the nonlinear hysteresis of PVDF-TrFE-CTFE polymer actuators and optimize control inputs under actuator constraints; validate the method through experimental measurements of actuator deflection and electric displacement, as well as numerical simulations;

Multiple Object Tracking with Dynamic Adaptive Object Motion Estimation

B. Cheng, Yunzhou Zhang

Object TrackingAutonomous DrivingImageMultimodalityPoint Cloud

🎯 What it does: Proposed the DA-MOT multi-object tracking algorithm, which calculates the dynamic and static states of objects using LiDAR and camera information, dynamically adjusts Kalman filter parameters based on object motion, and designs a re-association mechanism to correct incorrect ID associations.

Multiple-scale augmented reality markers for positioning of robotic micromanipulation

Shuzhang Liang, Fumihito Arai

Pose EstimationRobotic IntelligenceImage

🎯 What it does: This study designs and implements multi-scale augmented reality markers for localizing robotic manipulators at different scales, and ultimately performs cell loading operations at the end of the microfluidic chip through camera intrinsic calibration, marker detection, and distance compensation.

Multistream Network for LiDAR and Camera-based 3D Object Detection in Outdoor Scenes

Muhammad Ibrahim, Ajmal Mian

Object DetectionAutonomous DrivingConvolutional Neural NetworkImagePoint CloudBenchmark

🎯 What it does: Propose a multi-stream network (MuStD) for fusing LiDAR and RGB data to enhance outdoor 3D object detection accuracy.

Multitask Reinforcement Learning for Quadcopter Attitude Stabilization and Tracking using Graph Policy

Y. Liu, M. Basiri

OptimizationRobotic IntelligenceGraph Neural NetworkReinforcement Learning

🎯 What it does: Proposed a multi-task deep reinforcement learning framework for quadrotor attitude tracking and aggressive stabilization from arbitrary states; the framework leverages IsaacGym parallel simulation and graph convolutional network (GCN) strategies to unify the processing of two tasks, achieving 400 Hz control on Pixhawk flight controllers;

Muscle-on-a-Chip: A Self-Healing Actuator Platform in Robotic Systems*

Hongze Yin, Tao Yue

Robotic IntelligenceBiomedical Data

🎯 What it does: Designed and implemented a programmable microfluidic chip platform to construct functional muscle tubes using C2C12 cells, and simulated exercise-induced muscle damage and its regeneration process through controllable strain.

MuSPaCSA: Multi-Scale Parallel-Channel Self-Attention Network for Point Cloud Classification and Segmentation

Xuran Yao, Xutao Li

ClassificationSegmentationTransformerPoint Cloud

🎯 What it does: Proposed the MuSPaCSA network for point cloud classification and segmentation tasks, employing a multi-scale parallel channel self-attention mechanism.

MuxHand: A Cost-Effective and Compact Dexterous Robotic Hand Using Time-Division Multiplexing Mechanism

Jianle Xu, Chongkun Xia

Robotic Intelligence

🎯 What it does: Proposed a low-cost, compact Dexterous Robotic Hand (MuxHand) using a time-division multiplexing motor mechanism, and verified its grasping and manipulation performance through experiments.

MV2: A Large-Scale 360-degree Multi-View Maritime Vision Dataset for Object Detection and Segmentation

Junseok Lee, Kyoobin Lee

Object DetectionSegmentationImageBenchmark

🎯 What it does: Constructed and released the Multi-View Maritime Vision (MV2) dataset, and conducted benchmark tests on existing object detection and panoptic segmentation models on it.

Mysteric-Net: MIMO Hysteretic Friction-aware Lagrangian-based Network for Legged Robot

Hoyeong Yeo, Sehoon Oh

Robotic Intelligence

🎯 What it does: Proposes Mysteric-Net, a MIMO hysteresis friction perception network that integrates the Lagrangian formulation and TCN, for accurately identifying robot leg dynamics and enhancing inverse dynamics estimation and tracking performance.

NailTact: Single-camera based Tactile Fingertip with Nail

Hao Zhou, Kazuhiro Shimonomura

Robotic IntelligenceImage

🎯 What it does: Designed and verified the NailTact fingertip tactile sensor based on a single camera, which can simultaneously detect forces applied to both the fingertip and the nail. The response characteristics of the sensor were validated through experiments with a robotic finger prototype under conditions of nail load, grasping thin objects, and touching a tabletop. Additionally, a simplified model was proposed to explain the relationship between nail force and marker displacement.

NanoMVG: USV-Centric Low-Power Multi-Task Visual Grounding based on Prompt-Guided Camera and 4D mmWave Radar

Runwei Guan, Yutao Yue

Object DetectionSegmentationComputational EfficiencyRobotic IntelligencePrompt EngineeringImageMultimodality

🎯 What it does: Designed and implemented a low-power multi-task visual localization model, NanoMVG, for unmanned surface vessels (USV) to perform box-level and mask-level visual localization tasks simultaneously through natural language guidance using a camera and 4D millimeter-wave radar in aquatic environments.

Natural Humanoid Robot Locomotion with Generative Motion Prior

Haodong Zhang, Rong Xiong

Robotic IntelligenceAuto Encoder

🎯 What it does: Proposed a Generative Motion Prior (GMP) that generates natural reference actions through full-body motion remapping and conditional variational autoencoders, and provides fine-grained trajectory-level supervision during policy training using a frozen online motion generator.

NavHD: Low-Power Learning for Micro-Robotic Controls in the Wild

Chae Young Lee, Zerina Kapetanovic

Computational EfficiencyRepresentation LearningRobotic IntelligenceReinforcement Learning

🎯 What it does: Proposed and implemented NavHD, a low-power learning model suitable for micro-robot navigation

Navi2Gaze: Leveraging Foundation Models for Navigation and Target Gazing

Jun Zhu, Tao Zhang

Autonomous DrivingOptimizationVision Language ModelBenchmark

🎯 What it does: Proposed a navigation and target gaze method called Navi2Gaze based on vision-language models, designed to automatically select the optimal posture in open-vocabulary tasks.

NaviDiffuser: Tackling Multi-Objective Robot Navigation by Weight Range Guided Diffusion Model

Xuyang Zhang, Jianmin Ji

Knowledge DistillationRobotic IntelligenceTransformerDiffusion model

🎯 What it does: Introduce diffusion models for multi-robot navigation, using classification labels to guide the model in learning the relationship between navigation and human preferences.

NaviFormer: A Deep Reinforcement Learning Transformer-like Model to Holistically Solve the Navigation Problem

Daniel Fuertes, N. García

Autonomous DrivingTransformerReinforcement Learning

🎯 What it does: Proposed and implemented a deep reinforcement learning model called NaviFormer, which utilizes the Transformer architecture to simultaneously predict high-level route planning and low-level trajectory planning, thereby addressing the global navigation problem.

NeuFlow-V2: Push High-Efficiency Optical Flow To the Limit

Zhiyong Zhang, H. Singh

Computational EfficiencyConvolutional Neural NetworkOptical Flow

🎯 What it does: Propose NeuFlow-V2, a method for real-time high-precision optical flow estimation, which includes a lightweight backbone network and a fast refinement module.

Neural Collision Detection for Constrained Grasp Pose Optimization in Cluttered Environments

Longyuan Lin, Yuanlong Yu

OptimizationRobotic Intelligence

🎯 What it does: Proposes a framework that integrates collision avoidance as a core constraint into grasp pose optimization for achieving robust robot grasping in cluttered environments;

Neural Configuration Distance Function for Continuum Robot Control

Kehan Long, Nikolay Atanasov

OptimizationRobotic Intelligence

🎯 What it does: Proposed and implemented a neural network-based shape modeling method for continuum robots, N-CSDF, and integrated it into the MPPI controller to generate safe trajectories.

Neural MP: A Neural Motion Planner

Murtaza Dalal, Deepak Pathak

Autonomous DrivingOptimizationKnowledge Distillation

🎯 What it does: Proposed a neural motion planner called Neural MP, which utilizes a large amount of complex scenarios constructed in simulations to collect expert data, distills them into reactive neural policies, and subsequently combines them with lightweight optimization methods to achieve safe path planning in real environments.

Neural network control method for target tracking of magnetically actuated capsule endoscopic robots with obstacle avoidance and noise-resistant capabilities

Zhiwei Cui, Zheng Li

Robotic Intelligence

🎯 What it does: A target tracking neural network control method for a magnetic-controlled capsule endoscopy robot with obstacle avoidance and noise suppression capabilities was designed.

Neural Signatures and Decoding of the Various Cognitive Processes Elicited by the Same Stimulus

Jiawei Ju, Yongjie Zou

ClassificationBiomedical Data

🎯 What it does: Constructed an experimental paradigm for different cognitive processes under the same stimulus, explored their neural features, and decoded multiple cognitive processes using EEG signals.

Neural-Link: Non-Overlapping MPC Fusion and Passive Inertial Sensing on Soft Platforms

Zijia Dai, Laurent Kneip

Robotic IntelligenceSimultaneous Localization and MappingImageTime Series

🎯 What it does: This paper proposes a method that integrates an elastic deformation model into sensor fusion on a soft elastic platform. By using a neural network to map temporal deformation sequences to mass-normalized restoring forces, and leveraging consistency constraints between a continuous-time trajectory model and Newton's second law, the method achieves camera alignment, scale recovery, and inertial alignment on a non-overlapping stereo camera system with elastic connections.

Neural-Lyapunov Fusion: Stable Dynamical System Learning for Robotic Motion Generation

Haoyu Zhang, Long Cheng

Robotic Intelligence

🎯 What it does: Proposed a novel adaptive dynamic system (ADS) algorithm utilizing neural network technology, achieving point-to-point and periodic motion learning with stability guarantees through a neural Lyapunov function;

NeuroLoc: Encoding Navigation Cells for 6-DOF Camera Localization

Xun Li, Xian Wei

Pose EstimationTransformerImage

🎯 What it does: Proposed a neurobiologically inspired camera localization method called NeuroLoc, which leverages Hebbian learning modules, head-direction cell-inspired multi-head attention embedding, and 3D grid center prediction to enhance the robustness and accuracy of single-image localization.

Never too Prim to Swim: An LLM-Enhanced RL-based Adaptive S-Surface Controller for AUVs under Extreme Sea Conditions

Guanwen Xie, Yi Li

OptimizationRobotic IntelligenceTransformerLarge Language ModelReinforcement Learning

🎯 What it does: Developed an adaptive S-surface controller for AUVs based on LLM-enhanced reinforcement learning;

New Network Protocol for Supermedia-Enhanced Telerobotics

Xinyu Liu, Ning Xi

OptimizationRobotic Intelligence

🎯 What it does: Propose a network transmission protocol TRCP for hypermedia-enhanced remote robot operation

Nezha-Morphing: Design and Experiments of a Seabird-Inspired Hybrid Aerial Underwater Vehicle

Muxierepu Aili, Lian Lian

🎯 What it does: Designed and tested a folding-arm hybrid aerial-aquatic drone named Nezha-Morphing that mimics a seabird.

Nezha-T: a Bi-floating State Lightweight Tail-sitter HAUV

Xiqiao Han, Lian Lian

Robotic Intelligence

🎯 What it does: Designed and verified an ultra-lightweight bistable floating state hybrid air-sea vehicle named Nezha-T

NGD-SLAM: Towards Real-Time Dynamic SLAM without GPU

Yuhao Zhang, Mikel Luján

Pose EstimationSimultaneous Localization and MappingOptical Flow

🎯 What it does: Propose a real-time dynamic SLAM system that runs on CPU only, combining mask propagation mechanism and hybrid tracking strategy

Noise Fusion-based Distillation Learning for Anomaly Detection in Complex Industrial Environments

Jiawen Yu, Wenqiang Zhang

Anomaly DetectionKnowledge Distillation

🎯 What it does: Proposes a noise-fusion distillation learning framework specifically designed for anomaly detection and localization in complex industrial environments.

NOLO: Navigate Only Look Once

Bohan Zhou, Zongqing Lu

Autonomous DrivingReinforcement LearningOptical FlowVideo

🎯 What it does: Proposes a method that learns navigation policies using only offline videos, which infers pseudo action labels through optical flow and is trained with offline reinforcement learning;

Non-Buoyant Microrobots Swimming with Near-Zero Angle of Attack

Leendert-Jan W. Ligtenberg, Islam S. M. Khalil

Robotic Intelligence

🎯 What it does: Analyzes and experimentally verifies a control method for magnetic-driven helical microrobots to achieve drift-free near-zero angle linear movement at low Reynolds numbers.

Non-Contact Hand-Guided Coarse Positioning of Neurosurgical Instrument Insertion End Effector Based on Magnetic Sensing

Yitian Xian, Zheng Li

OptimizationRobotic Intelligence

🎯 What it does: Proposed a non-contact hand guidance method based on magnetic sensing for the coarse positioning of the instrument insertion end effector (IIEE) in neurosurgical procedures.

Non-differentiable Reward Optimization for Diffusion-based Autonomous Motion Planning

Giwon Lee, Kuk-Jin Yoon

Autonomous DrivingOptimizationReinforcement LearningDiffusion model

🎯 What it does: Proposed a reinforcement learning-based training scheme enabling diffusion models to learn non-differentiable safety and effectiveness objectives, and introduced a reward-weighted dynamic threshold algorithm to construct dense reward signals.

Non-Overlap-Aware Egocentric Pose Estimation for Collaborative Perception in Connected Autonomy

Hongjie Huang, Peng Gao

Pose EstimationAutonomous DrivingTransformer

🎯 What it does: Proposed a NOPE method for self-coordinate system pose estimation under non-overlapping perspectives in multi-robot collaborative perception.

Nonlinear Viscoelastic Model-based Deformation Optimization for Robotic Micropuncture in Retinal Vein Cannulation

Bo Hu, Xin Zhao

OptimizationRobotic IntelligenceBiomedical Data

🎯 What it does: Proposed a robot micropuncture scheme based on a nonlinear viscoelastic model, achieving deformation optimization of retinal vessels through a preload strategy, NV model, and speed optimization framework;

Normalized Triangulation for Calibrated Dual-View 3D Human Pose Estimation

Zijian Zhang, Tianyi Ma

Pose EstimationImage

🎯 What it does: Propose decomposing dual-camera 3D human pose estimation into 2D pose estimation and 2D→3D lifting, and introduce a novel technical solution for the latter.

Novel Cable Driven Fitness Gym Devices for Whole Body Weight Training

J. Park, Seungyong Hyung

Biomedical Data

🎯 What it does: Developed cable-driven wearable and fixed exercise equipment targeting lower limb muscles and full-body strength training, respectively, with interchangeable drive modules; evaluated their effectiveness through surface electromyography (sEMG) experiments compared to traditional dumbbells.

Novel Data-Driven Repetitive Motion Control Scheme for Redundant Manipulators With Zeroing Neurodynamics

Min Yang, Hui Zhang

Data-Centric LearningRobotic Intelligence

🎯 What it does: Proposed a data-driven discrete zeroing neural dynamics (DDZN) model and developed a data-driven Jacobian matrix estimation method, achieving trajectory tracking and repeatable configuration recovery for redundant manipulators without relying on structural parameters;

Novel Diffusion Models for Multimodal 3D Hand Trajectory Prediction

Junyi Ma, Hesheng Wang

GenerationTransformerDiffusion modelImageTextMultimodalityPoint Cloud

🎯 What it does: Propose the MMTwin model, which utilizes multimodal inputs (2D RGB, 3D point clouds, historical hand trajectory, and text prompts) for 3D hand trajectory prediction.

Novel LPV System Identification for a Gantry Stage: A Global Approach with Adjustable Basis Functions

Jegwon Yoon, Sehoon Oh

OptimizationComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposed a global LPV system identification method for multi-axis robot gantry frames, enabling independent selection of basis function orders for each parameter to eliminate unnecessary high-order terms.

NuExo: A Wearable Exoskeleton Covering all Upper Limb ROM for Outdoor Data Collection and Teleoperation of Humanoid Robots

Rui Zhong, Huimin Lu

Robotic IntelligenceMultimodality

🎯 What it does: Developed a wearable upper-limb full range of motion (ROM) exoskeleton system with immersive remote control and multimodal sensing for outdoor data collection and humanoid robot teleoperation.

Numerical Optimization-based Kinematics with Pose Tracking Control for Continuum Robots

Rui Peng, Peng Lu

OptimizationRobotic IntelligenceTime Series

🎯 What it does: Using multiple IMUs to record the pose data of three consecutive robotic arm end-effectors, develop a coordinate transformation scheme, propose a numerically optimized unified forward and inverse kinematics model, and utilize real-time attitude feedback to achieve closed-loop control. Finally, validate the model and controller's accuracy and convergence performance in simulation and real robot experiments.

NUSense: Shear Based Robust Optical Tactile Sensor

Madina Yergibay, Tasbolat Taunyazov

Robotic IntelligenceImage

🎯 What it does: This paper proposes and verifies an optical tactile sensing principle based on shear strain detection, using a dyed silicone gel layer to quantify shear displacement, and employing a NUSense camera to capture in real-time the stretching of the soft pad under mechanical loading (Poisson effect); subsequently, the robustness of the sensor was tested through multiple 8 N load cycles (5 mm radius ball-ended indenter) on the outer layer.

OASIS: Real-Time Opti-Acoustic Sensing for Intervention Systems in Unstructured Environments

Amy Phung, Richard Camilli

Robotic IntelligenceImageMultimodalityAudio

🎯 What it does: Real-time 3D reconstruction in unstructured underwater workspace through the fusion of optical images and acoustic data

Object Extrinsic Contact Surface Reconstruction through Extrinsic Contact Sensing from Visuo-tactile Measurements

Yoonjin Kim, Jung Kim

Multimodality

🎯 What it does: Developed a framework for external contact surface reconstruction based on vision-based tactile sensing, capable of identifying and estimating points and line contact positions where objects interact with the environment.

Object Packing and Scheduling for Sequential 3D Printing: a Linear Arithmetic Model and a CEGAR-inspired Optimal Solver

Pavel Surynek, Petr Kubis

Optimization

🎯 What it does: Propose to formulate the sequential object placement and scheduling problem in 3D printing as a linear arithmetic formula, and improve solving efficiency by utilizing SMT solvers along with CEGAR-inspired abstraction refinement techniques.

Observability Investigation for Rotational Calibration of (Global-pose aided) VIO under Straight Line Motion

Junlin Song, M. Olivares-Méndez

Pose EstimationSimultaneous Localization and Mapping

🎯 What it does: Analyzes the observability of rotation extrinsic parameters between IMU and camera under linear motion (pure translational linear motion), proving that it leads to at least one degree of freedom being unobservable.

Observability-driven Assignment of Heterogeneous Sensors for Multi-Target Tracking

Seyed Ali Rakhshan, He Kong

Object TrackingOptimization

🎯 What it does: Proposed a heterogenous sensor allocation algorithm based on observability for multi-target tracking

Observation of Snails and a Bionic Snail Robot Crawling with Distributed Suction

Qinjie Ji, Josie Hughes

Robotic Intelligence

🎯 What it does: Designed and implemented a snail-inspired robot that combines a spiral propeller and distributed suction cups to achieve upside-down crawling on ceilings.

Observation-Graph Interaction and Key-Detail Guidance for Vision and Language Navigation

Yifan Xie, Yaohua Liu

Autonomous DrivingGraph Neural NetworkVision-Language-Action ModelMultimodality

🎯 What it does: Proposes the Observation-Graph Interaction and Key-Detail Guidance (OIKG) framework for Visual and Language Navigation (VLN), enhancing navigation performance through the Observation-Graph Interaction module and the Key-Detail Guidance module.

Occlusion-Aware 6D Pose Estimation with Visual Observation Guided Diffusion Model

Yanbin Xiong, Jun Cheng

Pose EstimationDiffusion modelImage

🎯 What it does: Estimate the 6D pose of target objects in RGB-D images through occlusion-aware guidance based on diffusion models, and adaptively refine the pose in occluded and cluttered scenes.

Occupancy-belief Planning of Plant Manipulation for Staking

Pusong Li, Rakesh Nagi

Robotic IntelligenceWorld ModelImageAgriculture Related

🎯 What it does: Developed a plant fixation grasping system utilizing occupancy-belief planning.

OceanSim: A GPU-Accelerated Underwater Robot Perception Simulation Framework

Jingyu Song, Katherine A. Skinner

Data SynthesisRobotic IntelligenceImagePhysics Related

🎯 What it does: Propose OceanSim, a high-fidelity GPU-accelerated underwater simulation framework that utilizes advanced physical rendering techniques to reduce the simulation-to-reality gap, achieving real-time imaging sonar rendering and rapid synthetic data generation.