arXivSub Start free trial

IROS 2025 Papers — Page 16

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

REFINE-bot: Furnace Cleaning Robot for Heat-transfer Efficiency Improvement

Sanpoom Punapanont, P. Manoonpong

Robotic IntelligenceImageUltrasoundPhysics Related

🎯 What it does: Developed and tested the REFINE-bot furnace cleaning robot to enhance heat transfer efficiency in furnace radiation coils.

Refined Policy Distillation: From VLA Generalists to RL Experts

T. Jülg, Florian Walter

Knowledge DistillationRobotic IntelligenceSupervised Fine-TuningReinforcement LearningVision-Language-Action ModelMultimodality

🎯 What it does: Proposes Refined Policy Distillation (RPD), which integrates reinforcement learning (RL) with behavioral cloning to convert vision-language-action models (VLA) into more efficient expert policies

Region-Aware 6D Grasping for Industrial Bin-Picking: A Sim2Real Label Self-Generation and Hybrid Evaluation Framework

Xungao Zhong, Huosheng Hu

Data SynthesisPose EstimationDomain AdaptationRobotic IntelligenceLarge Language ModelMultimodalityPoint Cloud

🎯 What it does: This paper proposes a region-aware 6D grasping framework, combining the Sim2Real dataset and Semantic-GraspNet to generate and evaluate grasping poses for industrial box-packed objects.

Region-Centric 6-Dof Grasp Detection: A Data-Efficient Solution for Cluttered Scenes

Siang Chen, Guijin Wang

Pose EstimationRobotic IntelligenceImage

🎯 What it does: Propose a data-efficient 6-DoF grasping detection framework RCGD, adopting region center retrieval and local grasping prediction methods;

Registration After Completion: Towards Sparse and Partial Point Set Registration for Computer-Assisted Orthopedic Surgery

Xinzhe Du, Zhe Min

Point Cloud

🎯 What it does: Proposed the Registration After Completion framework, which first progressively completes sparse, partial intraoperative point sets and then registers the complete point sets; adopted a two-stage progressive completion strategy, bidirectional hybrid model, dual-path cross-attention, and clustering mechanism for correspondence, and computed the transformation through bidirectional registration.

REGRACE: A Robust and Efficient Graph-based Re-localization Algorithm using Consistency Evaluation

D'ebora N.P. Oliveira, Stefan Leutenegger

Pose EstimationAutonomous DrivingGraph Neural NetworkSimultaneous Localization and MappingPoint CloudGraph

🎯 What it does: Proposed a graph-based relocalization algorithm called REGRACE for loop closure detection and pose correction in large-scale navigation.

Regrasp Maps for Sequential Manipulation Planning

Svetlana Levit, Marc Toussaint

OptimizationRobotic Intelligence

🎯 What it does: Proposes using a regrasp map to assist task and motion planning (TAMP) in solving operational challenges requiring multiple regrasps in constrained and cluttered environments.

Reinforcement Learning Assist-As-Needed Control Promotes Recovery of Walking Speed Following Ankle Weight Perturbations

Andy Li, D. Zanotto

Robotic IntelligenceReinforcement Learning

🎯 What it does: Developed a reinforcement learning-based on-demand assistive controller for ankle exoskeletons to enhance gait speed training effectiveness; conducted a proof-of-concept study on subjects

Reinforcement Learning-Based Autonomous Control Methodology of Hydraulic Excavators

Bobo Helian, Marcus Geimer

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Propose a fully autonomous control method for the entire digging cycle based on reinforcement learning, develop a co-simulation tool, and train an agent using PPO to control the proportional valve, combining reward shaping with adaptive control frequency to improve training efficiency.

Reinforcement Learning-Based Energy-Efficient and Obstacle-Free Path Planning for Magnetic Microrobots in Dynamic Environments

Hongwei Wang, Tiantian Xu

OptimizationRobotic IntelligenceTransformerReinforcement Learning

🎯 What it does: Proposes an end-to-end path planner for energy-efficient and obstacle-avoiding navigation of magnetic helical microrobots in dynamic flow fields.

Reinforcement Learning-Based Microrobotic Swarm Navigation and Obstacle Avoidance in Partially Observable Environments

Shengming Luo, Qianqian Wang

Domain AdaptationRobotic IntelligenceTransformerReinforcement Learning

🎯 What it does: Proposed a Transformer-based reinforcement learning strategy integrating Proximal Policy Optimization (PPO) to achieve autonomous control and navigation for micro-robot swarms in obstacle-filled environments;

Reinforcement Learning-based Optimization of Humanoid Joint Motion Control via Text-driven Human Motion Mapping

Zihan Xu, Qi Chen

OptimizationRobotic IntelligenceReinforcement LearningText

🎯 What it does: Using a text-driven approach to map human motion to a humanoid robot, enabling motion redirection and controlling the robot to perform actions.

Reinforcement Learning-Based Scheduling for Dual-Arm Cluster Tool with Multifunctional Process Modules

Langni Liu, Yan Hou

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: A reinforcement learning approach is adopted to design a scheduling scheme for dual-arm integrated tools (DACT) and multifunctional process modules (MPMs)

Rejecting Outliers in 2D-3D Point Correspondences from 2D Forward-Looking Sonar Observations

Jiayi Su, Liuqing Yang

Anomaly DetectionPoint CloudUltrasound

🎯 What it does: To address outliers in the 2D-3D point correspondence of 2D forward-looking sonar (2D FLS) observations, two compatibility checks are proposed and integrated into the outlier removal process, followed by maximum clique search to determine the maximum consistent measurement set as inliers;

Relative Tilt Suppression of a Carried Object Using Base Link Angle Adjustment on a Quadruped-Wheeled Robot

Kimikage Kanno, I. Mizuuchi

OptimizationRobotic Intelligence

🎯 What it does: Studied and verified the base link angle planning method to suppress the tilting of the carried object relative to the base link, applied to quadruped wheeled robots.

Reliable Multi-Level Optimization for Safe Predictive Control of Autonomous Vehicles to Avoid Uncertain Multimodal PLEVs

E. Alao, Philippe Martinet

Autonomous DrivingOptimization

🎯 What it does: Proposed a multi-layer optimization strategy that combines sampling-based and direct optimization methods to enhance the safety of AV decision-making and control, as well as achieve trajectory smoothness; first, sampling-based optimization is used with F-sPIDP to identify safe candidate trajectories, followed by local control optimization to satisfy kinematic and dynamic constraints;

Rendering Anywhere You See: Renderability Field-guided Gaussian Splatting

Xiaofeng Jin, Matteo Matteucci

RestorationGenerationGaussian SplattingImage

🎯 What it does: Proposed a Gaussian splatting method based on a rendering field for scene view synthesis.

REOcc: Camera-Radar Fusion with Radar Feature Enrichment for 3D Occupancy Prediction

Chaehee Song, Dongsuk Kum

Autonomous DrivingMultimodalityBenchmark

🎯 What it does: Proposed a camera-radar fusion network called REOcc for 3D occupancy prediction, enhancing performance by enriching radar features.

Repetitive Motion Control for Redundant Manipulator under False Data Injection Attacks *

Yanqiong Zhao, Yinyan Zhang

OptimizationAdversarial AttackRobotic IntelligenceRecurrent Neural Network

🎯 What it does: Developed a robust controller that enables redundant manipulators to perform repetitive motion control and non-repetitive motion tasks in the face of false data injection attacks.

Reservoir Computing for Torque-Restricted Pendulum Control

Fan Ye, Fumiya Iida

OptimizationRobotic IntelligenceRecurrent Neural NetworkReinforcement LearningPhysics Related

🎯 What it does: Propose a model-agnostic controller using the Reservoir Computing framework to control a torque-limited pendulum with minimal training data (0.5%) and under passive data collection conditions.

Reservoir Computing-Enhanced Tube-MPC: Real-Time Self-Healing Control for Robust AUV Path Following Under Dynamic Faults

Lie Xu, Marcelo H. Ang

Computational EfficiencyRobotic IntelligenceRecurrent Neural Network

🎯 What it does: Integrated reservoir computing (RC) with tube model predictive control (Tube-MPC) to achieve real-time self-healing path tracking control for quadrotor autonomous underwater vehicles (QAUVs) under sudden failure conditions.

Resilient Multi-Robot Target Tracking with Sensing and Communication Danger Zones

Peihan Li, Lifeng Zhou

Object TrackingOptimizationRobotic Intelligence

🎯 What it does: Proposes an elastic collaborative framework for multi-robot multi-target tracking in unknown perception and communication hazardous areas, and evaluates its performance in various tracking scenarios.

ResLPR: A LiDAR Data Restoration Network and Benchmark for Robust Place Recognition Against Weather Corruptions

Wenqing Kuang, Xieyuanli Chen

RecognitionRestorationPoint CloudBenchmark

🎯 What it does: Propose a LiDAR data recovery network called ResLPRNet based on wavelet transform, and evaluate local recognition under adverse weather conditions using the ResLPR dataset

Resource-Efficient Affordance Grounding with Complementary Depth and Semantic Prompts

Yizhou Huang, Kailun Yang

SegmentationPrompt EngineeringVision Language ModelMultimodality

🎯 What it does: Proposes the BiT-Align framework, which utilizes depth images and text prompts to perform multimodal affordance map mapping on RGB images, and achieves functional region localization through the Bypass Prompt Module (BPM) and Text Feature Guidance (TFG) attention selection mechanism.

Reusing Attention for One-stage Lane Topology Understanding

Yang Li, Hao Zhao

Autonomous DrivingKnowledge DistillationTransformerImage

🎯 What it does: Propose a single-stage architecture that can simultaneously predict traffic elements, lane centerlines, and their topological relationships, improving accuracy and inference speed.

Revisiting 3D Curve to Surface Registration using Tangent and Normal Vectors for Computer-Assisted Orthopedic Surgery

Zhengyan Zhang, Zhe Min

OptimizationMeshBiomedical Data

🎯 What it does: Proposed a bidirectional hybrid model registration method called BiHMM-DTN, which uses dual-constrained tangent vectors and normal vectors for curve-to-surface registration.

Reward Training Wheels: Adaptive Auxiliary Rewards for Robotics Reinforcement Learning

Linji Wang, Xuesu Xiao

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a Reward Training Wheel (RTW) teacher-student framework to dynamically adjust auxiliary reward weights based on the student's capabilities, improving robot reinforcement learning.

RGB-Thermal Visual Place Recognition via Vision Foundation Model

Minghao Ye, Haoyao Chen

RecognitionTransformerMultimodality

🎯 What it does: Proposed an RGB-thermal multimodal fusion framework for visual localization.

RGBTrack: Fast, Robust Depth-Free 6D Pose Estimation and Tracking

Teng Guo, Jingjin Yu

Object TrackingPose EstimationContrastive LearningImage

🎯 What it does: Developed RGBTrack, a robust framework for real-time 6D pose estimation and tracking using only RGB images.

RH20T-P: A Primitive-Level Robotic Manipulation Dataset towards Composable Generalization Agents in Real-world Scenarios

Zeren Chen, Lu Sheng

Robotic IntelligenceAgentic AIVision Language ModelVideoBenchmark

🎯 What it does: Proposed a primitive-level robot manipulation dataset RH20T-P, containing approximately 38k video segments covering 67 diverse real-world tasks, with manually annotated sets of carefully designed primitive skills; simultaneously implemented the benchmark model RA-P on this dataset using the plan-execute Composable Generalization Agent (CGA) paradigm;

Risk Euclidean Distance-based Model Predictive Path Integral to Safety-Critical Obstacle Avoidance

Zihao Huang, Bo Zhang

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Proposes the RESM-MPPI algorithm for dynamic obstacle avoidance, combining Risk Euclidean Safety Metric (RESM), Risk Control Barrier Function (RCBF), and Control Obstacle Avoidance Annealing (COAA) sampling strategies to achieve safe and smooth path planning for AMR in dynamic environments.

Risk-Aware Autonomous Driving with Linear Temporal Logic Specifications

Shuhao Qi, S. Haesaert

Autonomous DrivingOptimization

🎯 What it does: Extend the human-like driving risk measure that has been validated, applying it to more complex linear temporal logic (LTL) driving scenarios, converting the LTL risk measure into a linear programming (LP) problem via occupation measures, and ultimately synthesizing a driving strategy that balances collision risk and traffic rule violations.

Risk-Aware Reinforcement Learning with Group Opinion for Autonomous Driving

Guanyi Zhao, Jianping Wang

Autonomous DrivingTransformerReinforcement Learning

🎯 What it does: Developed a reinforcement learning method called GORA-RL based on group opinion risk assessment to enhance the safety of autonomous driving decision-making

RMCC: Rigid Multi-joint Coupled Continuum Structure for Bionic Robots

Zi-Ye Zhou, Hui Cheng

Robotic Intelligence

🎯 What it does: Proposed and verified a bio-inspired robot RMCC with a rigid multi-joint coupled continuous structure, demonstrating its flexibility and adaptability in bio-inspired experiments such as lizard crawling, falling cat motion, and bird grasping.

RMG: Real-Time Expressive Motion Generation with Self-collision Avoidance for 6-DOF Companion Robotic Arms

Jiansheng Li, Yi Cai

OptimizationRobotic IntelligenceVideoSequential

🎯 What it does: Designed a real-time expressive motion generation planner that can generate expressive motions for a 6-DOF robotic arm from any start point to an end point under given time constraints.

RMMI: Reactive Mobile Manipulation using an Implicit Neural Map

Nicolas Marticorena, Niko Suenderhauf

Robotic IntelligenceWorld Model

🎯 What it does: Proposes RMMI, a reactive control framework for mobile manipulator robots implemented using neural signed distance fields (SDF).

ROA-BEV: 2D Region-Oriented Attention for BEV-based 3D Object Detection

Jiwei Chen, Rui Huang

Object DetectionConvolutional Neural NetworkPoint Cloud

🎯 What it does: Proposed a BEV-based 3D object detection network called ROA-BEV, which uses 2D region-oriented attention to make the backbone network focus more on feature learning in regions where targets are located.

Roadside GNSS Aided Multi-Sensor Integrated System for Vehicle Positioning in Urban Areas

Feng Huang, Li-Ta Hsu

Autonomous DrivingOptimizationSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes the RSG-GLIO method, which leverages roadside GNSS and C-V2X collaboration, combined with onboard GNSS/LiDAR/IMU sensors to achieve reliable odometry and map building.

RoadsideSplat: Robust 3D Gaussian Reconstruction from Monocular Roadside Surveillance

Zhaoxiang Liang, Yi Yang

Autonomous DrivingGaussian SplattingVideoPoint Cloud

🎯 What it does: A robust 3D Gaussian reconstruction framework based on a single fixed RGB roadside camera is proposed, capable of reconstructing dynamic road and vehicle scenes from roadside monitoring videos.

RoboCAP: Robotic Classification and Precision Pouring of Diverse Liquids and Granular Media with Capacitive Sensing

Yexin Hu, Zackory Erickson

ClassificationRobotic IntelligenceTabular

🎯 What it does: Integrate capacitive sensors into robotic grippers to enable classification and precise pouring of liquids and granular materials.

RoboDexVLM: Visual Language Model-Enabled Task Planning and Motion Control for Dexterous Robot Manipulation

Haichao Liu, Jun Ma

Robotic IntelligenceVision Language ModelImageText

🎯 What it does: Proposes the RoboDexVLM framework for task planning and grasp detection in natural language instruction-driven collaborative manipulators.

RoboEngine: Plug-and-Play Robot Data Augmentation with Semantic Robot Segmentation and Background Generation

Chengbo Yuan, Yang Gao

SegmentationData SynthesisRobotic IntelligenceImage

🎯 What it does: Proposed RoboEngine, a pluggable robot vision data augmentation toolkit that enables users to generate physics-aware and task-aware robot scenes with minimal code.

RoboEnvision: A Long-Horizon Video Generation Model for Multi-Task Robot Manipulation

Liudi Yang, A. Valada

GenerationRobotic IntelligenceDiffusion modelVideo

🎯 What it does: Proposed the RoboEnvision long-term video generation model for multi-task robot manipulation, which first decomposes high-level goals into atomic tasks to generate keyframes, then interpolates the keyframes into long videos using a second diffusion model, while incorporating a semantic preservation attention module and a lightweight strategy model;

RoboNotonecta: A Backswimmer-inspired Swimming Miniature Robot with Efficient Low-power Propulsion and Agile Aquatic Maneuverability

Chaofeng Wu, Wu Liu

Robotic Intelligence

🎯 What it does: Designed, manufactured, and tested a two-legged micro-swimming robot named RoboNotonecta inspired by backswimmer beetles, with centimeter-level dimensions, a weight of 10.3 micrograms, and a body length of 6.4 centimeters.

RoboNurse-VLA: Robotic Scrub Nurse System based on Vision-Language-Action Model

Shunlei Li, Zheng Li

Robotic IntelligenceLarge Language ModelVision Language ModelVision-Language-Action ModelImageTextMultimodalityAudio

🎯 What it does: Proposes RoboNurse-VLA—a robot surgical room cleaning nurse system based on Vision-Language-Action (VLA) models, capable of real-time precise grasping and delivering surgical instruments according to surgeons' voice commands.

RoboSwap: A GAN-driven Video Diffusion Framework For Unsupervised Robot Arm Swapping

Yang Bai, Gitta Kutyniok

Robotic IntelligenceDiffusion modelGenerative Adversarial NetworkVideo

🎯 What it does: Proposed the RoboSwap framework, achieving an unsupervised video editing method that replaces one robot arm with another in videos.

Robot Behavior Adaptation in Physical Human-Robot Interactions Based on Learned Safety Preferences

K. Majd, Rana Soltani-Zarrin

Robotic IntelligenceReinforcement Learning from Human Feedback

🎯 What it does: Propose a framework based on constrained partially observable Markov decision processes (constrained POMDP) for learning human safety preferences and adapting robot behavior in real-time.

Robot Teleoperation Design Requirements from End Users in Nuclear Facilities

Alperen Kenan, Manuel Giuliani

Robotic IntelligenceText

🎯 What it does: Conducted interviews to study the needs of robot operators in nuclear facilities, proposing 10 robot-specific requirements and 10 general user requirements, and providing recommendations for design and operation.

Robot-Mediated gesture-based memory game for older adult psychophysical stimulation

Luca Pozzi, M. Gandolla

ClassificationRecognitionPose EstimationRobotic IntelligenceConvolutional Neural NetworkImage

🎯 What it does: Implemented an application using the service robot TIAGo for a gesture memory game, where players must recognize and imitate letter sequences displayed by the robot, with sequence lengths increasing to train memory.

RobotFingerPrint: Unified Gripper Coordinate Space for Multi-Gripper Grasp Synthesis and Transfer

Ninad Khargonkar, Yu Xiang

GenerationData SynthesisOptimizationRobotic IntelligenceAuto EncoderPoint Cloud

🎯 What it does: Propose a Unified Gripper Coordinate Space (UGCS) to achieve multi-gripper grasp synthesis and transfer.

Robotic Grasping for Automated Sorting of Complex, Highly Contaminated Industrial Food Waste: A Benchmark Study

M. Thilakarathna, D. Herath

Pose EstimationRobotic IntelligenceBenchmark

🎯 What it does: Investigate the feasibility and limitations of traditional robotic grasping systems in high-pollution industrial food waste environments, and construct a complete automated grasping production line.

Robotic Hand Tool Use with Contact-Based Demonstration: The Case of Cucumber Peeling

Lingzi Xie, Chenguang Yang

Robotic Intelligence

🎯 What it does: Conducted an experiment on tool use with a three-fingered robotic hand for cucumber peeling through tactile feedback control

Robotic In Situ Measurement of Multiple Intracellular Physical Parameters Based on Three-micropipettes System

Mengya Liu, Qili Zhao

Robotic IntelligenceBiomedical Data

🎯 What it does: A method based on a self-developed three-microtubule system for robotic in-situ measurement of multiple intracellular physical parameters is proposed, achieving automated sequential measurement of cell mass density, elasticity, and intracellular pressure in sheep oocytes.

Robotic Manipulation of a Rotating Chain with Bottom End Fixed

Qi Jing Chen, Quang-Cuong Pham

Robotic Intelligence

🎯 What it does: Proposed a manipulation strategy using a robotic arm to achieve stable and consistent shape transformation of a uniform rotating chain.

Robotic Programmer: Video Instructed Policy Code Generation for Robotic Manipulation

Senwei Xie, Xilin Chen

GenerationRobotic IntelligenceLarge Language ModelVideoTextMultimodality

🎯 What it does: Proposed and implemented RoboPro, a robot foundation model capable of generating executable strategy code for robotic manipulation through visual perception and free-text instructions in zero-shot scenarios, and designed the Video2Code method to automatically synthesize executable code from large-scale online videos.

Robotic Task Ambiguity Resolution via Natural Language Interaction

Eugenio Chisari (University of Freiburg), A. Valada (University of Freiburg)

Robotic IntelligenceVision Language ModelVision-Language-Action ModelText

🎯 What it does: Propose the AmbResVLM method to address ambiguity in robot task descriptions, verify it in simulation and real-world environments, and improve the success rate of downstream robot strategies.

Robotic Ultrasound-Guided Femoral Artery Reconstruction of Anatomically-Representative Phantoms

Lidia Al-Zogbi, Axel Krieger

SegmentationRobotic IntelligenceConvolutional Neural NetworkImageVideoBiomedical DataComputed TomographyUltrasound

🎯 What it does: Developed and validated a method for autonomous robot-based femoral artery ultrasound scanning and 3D vascular reconstruction

Robotics Virtual Laboratory Featuring Serial, Parallel, Wheeled Robots, and Autonomous Off-road Vehicles, and Covering Analysis, Control and Sensors

D. Nasrallah, Angelo A. Chrabieh

Robotic Intelligence

🎯 What it does: Developed and introduced a real-time robot virtual laboratory covering serial-parallel manipulators, wheeled mobile robots, and autonomous off-road vehicles, providing functions such as motion analysis, control design, and sensor perception;

Robots Calling the Shots: Using Multiple Ground Robots for Autonomous Tracking in Cluttered Environments

Weijian Zhang, Masoumeh Mansouri

OptimizationRobotic IntelligenceImage

🎯 What it does: Proposed a framework for multi-robot cooperative autonomous tracking using differential drive robots equipped with omnidirectional cameras in complex, unstructured environments.

Robots that Suggest Safe Alternatives

Hyundoo Jeong, Andrea V. Bajcsy

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes a framework (SALT) that enables robots to suggest safe alternative goals within the goal space when they cannot safely execute user-specified objectives.

Robust and Efficient Embedded Convex Optimization through First-Order Adaptive Caching

Ishaan Mahajan, B. Plancher

Autonomous DrivingOptimizationComputational Efficiency

🎯 What it does: Proposed a first-order adaptive cache-based model predictive control method that can online update hyperparameters without recalculating the cache.

Robust and Expressive Humanoid Motion Retargeting via Optimization-Based Rig Unification

Taemoon Jeong, Sungjoon Choi

OptimizationRobotic Intelligence

🎯 What it does: Propose a robust and automated motion retargeting pipeline capable of generating natural, stable, and expressive motions for various humanoid robots.

Robust and High-Fidelity 3D Gaussian Splatting: Fusing Pose Priors and Geometry Constraints for Texture-Deficient Outdoor Scenes

Meijun Guo, Bin Liang

GenerationPose EstimationOptimizationGaussian SplattingSimultaneous Localization and MappingImagePoint Cloud

🎯 What it does: Improved pose estimation in large-scale outdoor scenes by combining prior poses from LiDAR-IMU Odometry with COLMAP triangulation, and enhanced directional and shape consistency in scene representation by incorporating normal vector constraints and effective rank regularization into 3D Gaussian Splatting (3DGS), achieving higher-quality digital asset rendering.

Robust and Modular Multi-Limb Synchronization in Motion Stack for Space Robots with Trajectory Clamping via Hypersphere

Elian Neppel, Kazuya Yoshida

Robotic Intelligence

🎯 What it does: Propose a robust and modular multi-arm synchronization method that achieves trajectory synchronization for multi-dimensional states in space robots through hyper-sphere constraints and adapts to system changes

Robust and Real-Time Perception and Planning for UGVs in Complex Outdoor Environments

Dongjie Huo, Zhengcai Cao

Autonomous DrivingOptimizationRobotic IntelligenceMixture of ExpertsPoint Cloud

🎯 What it does: Proposes a Dynamic and Low-Height Obstacle Avoidance Navigation (DLAN) system, integrating perception, path planning, and target point correction to achieve safe navigation for UGVs in complex outdoor environments.

Robust Deep Reinforcement Learning in Robotics via Adaptive Gradient-Masked Adversarial Attacks

Zongyuan Zhang, Yue Gao

Adversarial AttackRobotic IntelligenceReinforcement Learning

🎯 What it does: Propose AGMR Attack, a white-box adversarial attack method combining gradient soft masking mechanism with reinforcement learning, which identifies critical state dimensions and dynamically allocates perturbations to enhance the robustness of DRL robot control.

Robust Instant Policy: Leveraging Student’s t-Regression Model for Robust In-context Imitation Learning of Robot Manipulation

Hanbit Oh, Y. Domae

Robotic IntelligenceLarge Language ModelSequential

🎯 What it does: Proposed a robust online imitation learning algorithm (RIP), which aggregates multiple candidate trajectories generated by LLM using a Student's t regression model to suppress hallucination phenomena and generate reliable robot trajectories.

Robust Ladder Climbing with a Quadrupedal Robot

D. Vogel, Marco Hutter

Robotic IntelligenceReinforcement Learning

🎯 What it does: By combining reinforcement learning control strategies with a hook claw end-effector, enabling a quadruped robot to climb stairs.

Robust Maritime Object Detection under Adverse Conditions via Joint Semantic Learning without Extra Computational Overhead

Junseok Lee, Kyoobin Lee

Object DetectionSegmentationConvolutional Neural NetworkImage

🎯 What it does: Proposed and implemented the Joint Semantic Learning (JSL) framework, integrating the marine scene segmentation module with the object detection network during training, and removing the segmentation module during inference to achieve robust object detection without additional computational overhead.

Robust Model Predictive Control for Quadruped Locomotion Under Model Uncertainties and External Disturbances

Weipeng Xia, Yun-Hui Liu

OptimizationRobotic Intelligence

🎯 What it does: Proposed a tubular robust model predictive control (TR-MPC) framework for quadruped robots under model uncertainties and external disturbances

Robust Model-Free Path Tracking Algorithm for Hydraulic Center-Articulated Scooptrams

Zihan Zhang, Zheng Fang

Autonomous Driving

🎯 What it does: Proposes a model-free adaptive steering control method for path tracking of hydraulic central hinge excavators in narrow underground mine environments.

Robust Offline Imitation Learning Through State-level Trajectory Stitching

Shuze Wang, Jie Chen

Robotic IntelligenceReinforcement LearningSequential

🎯 What it does: Propose a state-based search framework that concatenates state-action pairs from incomplete demonstrations into richer training trajectories to enhance policy learning.

Robust Online Calibration for UWB-Aided Visual-Inertial Navigation with Bias Correction

Yizhi Zhou, Xuan Wang

Autonomous DrivingOptimizationSimultaneous Localization and Mapping

🎯 What it does: Proposed a robust online UWB anchor calibration framework for UWB-assisted visual-inertial navigation systems.

Robust Reinforcement Learning based on Momentum Adversarial Training

Li He, Zhiyan Dong

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Propose a robust reinforcement learning framework based on momentum adversarial training, modeling action space perturbations and improving optimization methods.

Robust Robotic Assembly of Reusable, Rectangular Blocks

Zhongming Huang, Nils Napp

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: Propose an automatic solver to optimize the overlap of rectangular blocks in user-specified structures, design a new robot capable of manipulating, fixing, and climbing blocks of comparable width to the robot, conduct a detailed analysis of robot primitives, demonstrate linear, curved, cantilever, and semi-circular arch structures, and validate the enhancement of structural integrity through physical simulations.

Robust Stabilization of an Autonomous Underwater Vehicle in Specified Finite-time with Disturbance Rejection

Hongjiao Niu, Xiayang Li

Robotic Intelligence

🎯 What it does: Propose a robust finite-time stable control method to achieve stability of autonomous underwater vehicles within a predetermined time and suppress unknown external disturbances.

Robust Wrench-Feasible Control for Multiple UAVs Aerial Transportation System with Adaptive Cable Configuration

Yu Yang, Liaoni Wu

OptimizationRobotic Intelligence

🎯 What it does: This paper proposes a robust torque-feasible control framework. For the rope-suspended load in multi-robot aerial transportation systems, a load controller based on an extended state observer is designed. A segmented torque adjustment strategy and an adaptive rope configuration strategy are proposed. The effectiveness of the method is verified through experiments.

RoCaRS: Robust Camera-Radar BEV Segmentation for Sensor Failure Scenarios

Byounghun Park, Soonmin Hwang

SegmentationAutonomous DrivingImagePoint Cloud

🎯 What it does: Proposed the RoCaRS model for bird's-eye view (BEV) segmentation in scenarios where camera and radar sensors fail.

ROD-VLM: A Framework of Real-time Robotic Perception, Reasoning and Manipulation

Yinkai Zhu, Yizhuo Sun

Object DetectionRobotic IntelligenceConvolutional Neural NetworkVision Language Model

🎯 What it does: Designed and implemented a real-time robotic perception, reasoning, and control framework named ROD-VLM that integrates YOLO-v5x and Vision-Language Models (VLM).

Rotation-Equivariant Robot Vision: A Perspective via Correspondence-Matching and Pre-training

Shuai Su, Qi Chen

Robotic IntelligenceTransformerImage

🎯 What it does: Proposed a correspondence matching method based on pre-trained group equivariant neural networks

RRT*former: Environment-Aware Sampling-Based Motion Planning using Transformer

Mingyang Feng, Xiang Yin

OptimizationRobotic IntelligenceTransformer

🎯 What it does: Proposed a sampling-based optimal path planning algorithm called RRT*former, which combines RRT* and Transformer for robot path planning in complex dynamic environments

RSSS: Robust Structural Semantic Segmentation for Autonomous Drone Delivery to Door

Shengqing Xia, Chunyi Peng

SegmentationAutonomous DrivingImage

🎯 What it does: Propose a robust structural semantic segmentation method called RSSS, which can enhance segmentation performance in different environments without retraining.

RT-HCP: Dealing with Inference Delays and Sample Efficiency to Learn Directly on Robotic Platforms

Zakaria El Asri, Nicolas Thome

Robotic IntelligenceReinforcement Learning

🎯 What it does: A general framework for handling inference latency is proposed, and under this framework, the RT-HCP algorithm is designed to directly learn controllers on a robot platform.

rt-RISeg: Real-Time Model-Free Robot Interactive Segmentation for Active Instance-Level Object Understanding

Howard H. Qian, Kaiyu Hang

SegmentationRobotic Intelligence

🎯 What it does: Proposes rt-RISeg, a real-time unsupervised instance-level object segmentation framework based on robot interaction, which can generate and update segmentation masks in real-time by analyzing body-frame invariant features produced from robot motion, without relying on pre-trained models.

Runtime Energy-Efficient Control Policy for Mobile Robots with Computing Workload and Battery Awareness

Chen Wu, Juha Plosila

OptimizationComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposes a coordinated optimization strategy for mechanical and computational units based on battery SOC, utilizing configuration space exploration to adjust motor speed and CPU frequency to enhance the energy efficiency of mobile robots.

RwoR: Generating Robot Demonstrations from Human Hand Collection for Policy Learning without Robot

Liang Heng, Hao Dong

GenerationData SynthesisRobotic IntelligenceVideoSequential

🎯 What it does: A demonstration collection system based on human hands was built, and a hand-to-gripper generative model was used to convert human hand demonstrations into robot gripper demonstrations for robot learning without prior knowledge.

S3D: A Spatial Steerable Surgical Drilling Framework for Robotic Spinal Fixation Procedures

Daniyal Maroufi, F. Alambeigi

Robotic IntelligenceImageBiomedical DataComputed Tomography

🎯 What it does: Proposed the S3D framework, achieving cross-all-vertebral-level guided surgical drilling, and completed four-stage calibration, registration, and navigation by integrating enhanced CT-SDR with a seven-degree-of-freedom robotic arm, validated through plane and oblique drilling experiments.

SA-MVSNet: Spatial-aware Multi-view Stereo Network with Attention Cost Volume

Haoran Kong, Hongbo Jiang

Depth EstimationConvolutional Neural NetworkImage

🎯 What it does: Proposed SA-MVSNet, combining the Pixel-Driven Spatial Interaction (PDSI) module and attention cost volume to achieve more accurate multi-view stereo reconstruction.

SAC(λ): Efficient Reinforcement Learning for Sparse-Reward Autonomous Car Racing using Imperfect Demonstrations

Heeseong Lee, Dongjun Lee

Autonomous DrivingReinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: Propose an RLfD algorithm named SAC(λ) for autonomous racing tasks with sparse rewards, achieving efficient learning through imperfect examples.

Safe and Efficient Navigation for Differential-Drive Robots in Dynamic Pedestrian Environments

Wenhao Liu, Yunjiang Lou

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Proposes a navigation framework that integrates pedestrian risk maps (modeled using asymmetric Gaussian distributions) into B-spline trajectory optimization, achieving safe and efficient navigation in dynamic pedestrian environments.

Safe and Efficient Target Singulation with Multi-Fingered Gripper using Collision-Free Push-Stack Synergy

Hyojeong Kim, ChangHwan Kim

OptimizationSafty and PrivacyRobotic Intelligence

🎯 What it does: Propose a collision-avoiding push-stacking coordination and an improved LOBS algorithm to achieve goal unification in constrained workspaces.

Safe Corridor-Based MPC for Follow-Ahead and Obstacle Avoidance of Mobile Robot in Cluttered Environments

Yikun Zhang, Jian Huang

OptimizationRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud

🎯 What it does: An integrated follow-ahead framework is proposed, including the Leg Motion Model-based Extended Kalman Filter (LMM-EKF) for predicting the detour behavior of the followed human, an iterative human path search algorithm, and Safe Corridor-based Model Predictive Control (SCMPC) to obtain optimal control solutions.

Safe Lattice Planning for Motion Planning with Dynamic Obstacles

Emil Wiman, Mattias Tiger

Autonomous DrivingBenchmark

🎯 What it does: Proposed a safe lattice planning framework, Safe Lattice Planner (SLP), for dynamic uncertain environments, and extended the replanning and survivability capabilities of the Receding-Horizon Lattice Planner (RHLP)

Safe Motion Planning and Control Using Predictive and Adaptive Barrier Methods for Autonomous Surface Vessels

Alejandro Gonzalez-Garcia, Daniela Rus

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Propose a safety motion planning strategy combining Model Predictive Control (MPC) and Control Barrier Functions (CBF), using time-varying inflated elliptical obstacles to achieve real-time safe navigation

Safe Navigation in Uncertain Crowded Environments Using Risk Adaptive CVaR Barrier Functions

Xinyi Wang, Dimitra Panagou

OptimizationRobotic Intelligence

🎯 What it does: Proposes a risk-adaptive method based on the Conditional Value at Risk (CVaR) obstacle function, introducing a dynamic interval obstacle function to evaluate the robot's relative state with obstacles. It automatically adjusts the risk level to achieve minimal necessary risk, enhancing safety and optimizability in uncertain environments.

Safe Probabilistic Planning for Human-Robot Interaction using Conformal Risk Control

Jake Gonzales, Lillian J. Ratliff

Safty and PrivacyRobotic Intelligence

🎯 What it does: Proposes a probabilistic safety control framework that combines control barrier functions with compliance risk control for human-robot interaction, dynamically adjusting safety margins.

Safe Reinforcement Learning with a Predictive Safety Filter for Motion Planning and Control: A Drifting Vehicle Example

Bei Zhou, Johannes Betz

Autonomous DrivingReinforcement Learning

🎯 What it does: Proposes a safe reinforcement learning motion planner integrated with a predictive safety filter for automatic drift.

Safe, Task-Consistent Manipulation with Operational Space Control Barrier Functions

Daniel Morton (Stanford University), Marco Pavone (Stanford University)

Robotic Intelligence

🎯 What it does: Proposes an operational space control barrier function (OSCBF) framework for achieving safe and task-consistent robotic manipulation control in unstructured environments.

Safety Aware Task Planning via Large Language Models in Robotics

A. Khan, Ali Anwar

Robotic IntelligenceLarge Language ModelAgentic AI

🎯 What it does: Propose the multi-LLM framework SAFER, which achieves safe feedback through a Safety Agent and LLM-as-a-Judge, and integrates control barrier functions (CBF) to ensure the safety of robot task planning.

Safety-Aware Geometric Force-Impedance Control for Manipulators

Danping Zeng, Hui Zhang

Robotic Intelligence

🎯 What it does: Proposed a safety-aware geometric force damping controller that achieves precise force tracking and is compatible with damping behaviors.