arXivSub Start free trial

IROS 2025 Papers — Page 17

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

Safety-Aware Optimal Scheduling for Autonomous Masonry Construction using Collaborative Heterogeneous Aerial Robots

Marios-Nektarios Stamatopoulos, G. Nikolakopoulos

OptimizationRobotic Intelligence

🎯 What it does: Propose a high-level task planning and optimal collaboration framework for heterogeneous drone teams, used for autonomous wall construction, including brick placement and mortar application, and generate construction plans through automated pipelines.

Safety-Compliant Navigation: Navigation Point-Guided Planning with Primitive Trajectories

Zixuan Deng, Yanping Xiang

Autonomous DrivingSafty and Privacy

🎯 What it does: Proposes a safety-compliant navigation method guided by navigation points, using original trajectories, which integrates the continuity of limited jumps with an improved RRT-connect algorithm.

Safety-Guided RRT*: Hyperoctant Sampling-based Path Planning with SDF-based Robotic Representation

Yangmin Xie, Yusheng Yang

OptimizationRobotic Intelligence

🎯 What it does: Proposed a Safety-Guided RRT* (SG-RRT*) that achieves safer and higher success rate path planning by integrating SDF safety metrics with Hyperoctant sampling strategies.

SAFormer: Spatially Adaptive Transformer for Efficient and Multi-Resolution Occupancy Prediction

Song Tang, Xiaowen Chu

Autonomous DrivingComputational EfficiencyTransformerPoint CloudBenchmark

🎯 What it does: Proposed a Transformer-based spatially adaptive occupancy prediction framework called SAFormer, achieving efficient occupancy prediction through two techniques: Octree-based multi-resolution feature learning and spatially adaptive progress queries.

SAGENet: Binaural Echo-Based 3D Depth Estimation with Sparse Angular Queries and Refined Geometric Cues

Guangyao Liu, Zhi Wang

Depth EstimationTransformerAudio

🎯 What it does: Propose SAGENet, which uses binaural echoes for scene depth estimation, explicitly extracting spatial cues to enhance depth accuracy

Saliency-Guided Domain Adaptation for Left-Hand Driving in Autonomous Steering

Zahra Mehraban, Ronald Schroeter

Domain AdaptationAutonomous DrivingConvolutional Neural NetworkSupervised Fine-TuningImage

🎯 What it does: Using PilotNet and ResNet for domain adaptation training on left-hand driving on Australian highways.

SAMap: Semantic Alignment for HD Map Detection Domain Generalization Under Varying Weather and Lighting

Wenjie Gao, Nanning Zheng

Image TranslationDomain AdaptationAutonomous DrivingTransformerPrompt EngineeringImage

🎯 What it does: Proposed the SAMap framework, which maps images under different weather and lighting conditions to a unified domain using a semantic alignment module, and enhances the domain generalization performance of HD map prediction by ensuring semantic consistency through training with a Vision-Language model.

Sampling-Based Model Predictive Control for Dexterous Manipulation on a Biomimetic Tendon-Driven Hand

Adrian Hess, Robert K. Katzschmann

OptimizationRobotic IntelligenceVision Language ModelVideoText

🎯 What it does: Demonstrated in-hand manipulation on a physical biomimetic tendon-driven robotic hand using sampled MPC, showcasing ball rolling, flipping, and grasping.

Sampling-Based Motion Planning with Discrete Configuration-Space Symmetries

Thomas Cohn, Russ Tedrake

Robotic Intelligence

🎯 What it does: The study investigates how to efficiently implement key primitives of sampling-based motion planning in configuration spaces with discrete symmetry, and validates improvements in sample complexity and performance through theoretical analysis and experiments.

Sampling-Based Path Planning for Tethered Robot Chains

Zeyuan Jin, Sze Zheng Yong

Robotic Intelligence

🎯 What it does: Propose a scalable path planning algorithm for multi-robot systems with finite-length rope chain connections, focusing on achieving feasible and computationally efficient paths.

SAVR: Scooping Adaptation for Variable food properties via Reinforcement Learning

J.-Anne Yow, Wei Tech Ang

Robotic IntelligenceReinforcement LearningImage

🎯 What it does: Developed a reinforcement learning-based variable food attribute sampling method called SAVR, enabling robots to precisely sample specified amounts based on different food properties.

SaWa-ML: Structure-Aware Pose Correction and Weight Adaptation-Based Robust Multi-Robot Localization

Junho Choi, Hyun Myung

Robotic Intelligence

🎯 What it does: Propose the SaWa-ML method to achieve geometry-aware pose correction and weight-adaptive multi-robot localization.

Scalable Learning of High-Dimensional Demonstrations with Composition of Linear Parameter Varying Dynamical Systems

Shreenabh Agrawal, Sami Haddadin

OptimizationRobotic Intelligence

🎯 What it does: Proposed a new composite method for learning stable dynamic systems that satisfy BMI constraints, enabling learning from demonstrations and generalization of robotic tasks.

Scalable MARL for Cooperative Exploration with Dynamic Robot Populations via Graph-Based Information Aggregation

Xiaoqi Ren, Xiaojian Qiu

Robotic IntelligenceGraph Neural NetworkTransformerReinforcement Learning

🎯 What it does: Proposes a Multi-Robot Information Planner (MIP) that enables collaborative exploration by multiple robots under limited local observations and dynamic group sizes using reinforcement learning, aiming for efficient area coverage.

Scalable Offline Metrics for Autonomous Driving

Animikh Aich, Eshed Ohn-Bar

Autonomous Driving

🎯 What it does: Investigate the relationship between offline evaluation metrics and online performance, conduct extensive experiments, propose an offline metric based on knowledge uncertainty, and verify its effectiveness in simulated and real-world vehicle environments.

Scalable Outdoors Autonomous Drone Flight with Visual-Inertial SLAM and Dense Submaps Built without LiDAR

Sebastián Barbas Laina, Stefan Leutenegger

Autonomous DrivingSimultaneous Localization and MappingImage

🎯 What it does: Developed a miniature drone system entirely based on low-cost passive visual and inertial sensors for large-scale autonomous navigation in outdoor, unstructured, cluttered environments.

Scalable Real2Sim: Physics-Aware Asset Generation Via Robotic Pick-and-Place Setups

Nicholas Pfaff, Russ Tedrake

Data SynthesisRobotic IntelligenceNeural Radiance FieldGaussian SplattingImageMeshPhysics Related

🎯 What it does: Proposed a fully automated Real2Sim pipeline that generates object visual geometry, collision geometry, and inertial parameters for simulation by leveraging robot gripper-grasping and placement, external cameras, and robot joint torque sensors.

Scanning Bot: Efficient Scan Planning using Panoramic Cameras

Euijeong Lee, Young J. Kim

OptimizationComputational EfficiencyRobotic IntelligenceImagePoint Cloud

🎯 What it does: Proposed a fully autonomous scanning planner capable of generating efficient and collision-free cruise paths while ensuring sufficient feature overlap between camera viewpoints;

SCORE: Saturated Consensus Relocalization in Semantic Line Maps

Haodong Jiang, Junfeng Wu

Pose EstimationSimultaneous Localization and Mapping

🎯 What it does: Proposes SCORE, a visual localization system that achieves extremely compact maps by leveraging semantic-tagged 3D line segment maps.

SCORPION: Robust Spatial-Temporal Collaborative Perception Model on Lossy Wireless Network

Ruiyang Zhu, Z. Mao

Autonomous DrivingTransformer

🎯 What it does: Proposed the SCORPION end-to-end collaborative perception framework, which integrates AI to co-design the application layer and system layer, addressing packet loss, localization errors, and synchronization errors, comprising three core components: L-BEV-R, DSCA, and TA.

SDA-LLM: Spatial DisAmbiguation via Multi-turn Vision-Language Dialogues for Robot Navigation

Kuan-Lin Chen, Jen-Jee Chen

Robotic IntelligenceTransformerLarge Language ModelVision Language ModelMultimodality

🎯 What it does: Proposes a two-layer framework that leverages LLM and VLM to enable robots to perform spatial disambiguation through multi-round visual language dialogues, accurately locating navigation targets.

SDF-guided Keyframe Selection: Novel Boost for NeRF SLAM Loop Closure

Hui Ma, Jun Cheng

Robotic IntelligenceNeural Radiance FieldSimultaneous Localization and MappingImagePoint Cloud

🎯 What it does: Propose a keyframe selection algorithm based on SDF to improve loop closure in SLAM

Seamless Transition Control in Spring-Legged Quadrotors: A Hybrid Dynamics Perspective with Guaranteed Feasibility

Hongli Li, Hui Cheng

OptimizationRobotic Intelligence

🎯 What it does: Propose a systematic approach to achieve multi-modal motion for spring leg quadrotors between air and ground, including differential flatness analysis, a unified hybrid trajectory optimization framework, and hybrid nonlinear model predictive control.

SEB-Naver: A SE(2)-based Local Navigation Framework for Car-like Robots on Uneven Terrain

Xiaoying Li (Zhejiang University), Fei Gao (Zhejiang University)

OptimizationRobotic Intelligence

🎯 What it does: Designed and implemented an SE(2)-based local navigation framework called SEB-Naver for real-time terrain assessment and trajectory optimization of wheeled robots on uneven terrain.

Secure Safety Filter: Towards Safe Flight Control under Sensor Attacks

Xiao Tan, A. Ames

Autonomous Driving

🎯 What it does: Propose a safe filter that combines control barrier functions and a safe state estimator to ensure the safe flight control of drones under sensor attacks.

SeGMan: Sequential and Guided Manipulation Planner for Robust Planning in 2D Constrained Environments

Cankut Bora Tuncer, Ozgur S. Oguz

OptimizationRobotic Intelligence

🎯 What it does: Propose the SeGMan hybrid planning framework, combining sampling methods, optimization methods, and guided forward search to address sequential manipulation challenges in 2D constrained environments.

SEI3D: CPU-only 3D Object Tracking Fusing Sparse-flow-filtered Edge and Interior Alignment

Jixiang Chen, Leshan Wang

Object TrackingOptical FlowVideo

🎯 What it does: Propose a monocular 3D object tracking method that achieves real-time performance (60Hz) using only CPU. The method filters redundant edges through sparse flow and aligns edges with the object interior by leveraging internal correspondences.

Selective Motion Control of Cell Microrobots in Three-Dimensional Space

Haoyu Zhang, Qianqian Wang

Robotic Intelligence

🎯 What it does: A selective control strategy based on a movable electromagnetic coil system was proposed, combined with a mass-spring-damper model and visual feedback to achieve precise control of magnetic microrobots in three-dimensional space. The control capability was verified through experiments such as staircase climbing and circular navigation. Furthermore, real-time sorting and selective manipulation of multiple robots were achieved by utilizing the differences in magnetic response among microrobots.

Self-Assembly Planning for Modular Robots via Multi-Agent Path Finding on Time-Expanded Networks

Zhen Huang, Minghe Shan

Robotic Intelligence

🎯 What it does: Propose a multi-agent path planning framework based on a time-expanded network for self-assembly planning of modular robots

Self-Decoupling and Hysteresis Compensation in a Soft Multi-Axis Force Sensor for Improved Performance

Cong Peng, Yantao Shen

🎯 What it does: A multi-axis piezoresistive soft force sensor was developed, achieving self-decoupling through Wheatstone bridge resistance compensation, while the Preisach model was employed for hysteresis compensation to enhance measurement accuracy and reliability.

Self-Distilled Stereo Matching: Real-Time Domain Generalization for Robotic Depth Perception

Xuxin Zhang, Yulan Guo

Depth EstimationDomain AdaptationAutonomous DrivingKnowledge DistillationRobotic IntelligenceConvolutional Neural NetworkSupervised Fine-TuningImageBenchmark

🎯 What it does: Proposes a learning framework called LMC for challenging regions, combining pre-training on a high-frequency dataset, self-distillation training on the left non-overlapping regions, and adaptive difficult region masking to enhance domain generalization in stereo matching.

Self-localization on a 3D map by fusing global and local features from a monocular camera

Satoshi Kikuch, Tsuyoshi Tasaki

Pose EstimationConvolutional Neural NetworkTransformerSimultaneous Localization and MappingImage

🎯 What it does: Proposed a self-localization method combining CNN and Vision Transformer, achieving localization on a 3D map using a monocular camera.

Self-Sensing Liquid Crystal Elastomer Actuator with Magnetic-Thermal Synergy

Shen Gao, Yue Wang

Robotic IntelligencePhysics Related

🎯 What it does: Designed and implemented a self-sensing liquid crystal elastomer (SS-LCE) actuator capable of achieving complex motions under thermal actuation and enabling real-time feedback through magnetic field variations

Self-supervised 3D Reconstruction of Tibia and Fibula from Biplanar X-rays

Kai Pan, Shoudong Huang

GenerationGraph Neural NetworkImageBiomedical Data

🎯 What it does: Reconstruct patient-specific 3D models of the tibia and fibula using two X-ray images from the coronal and sagittal planes combined with a generic template, integrating point-based deformation and deep learning techniques.

Self-Supervised Complementary Learning between Vision and Tactility by Probing Action into an Open-Mouth Container

Daiki Takamori, Dotaro Usui

Robotic IntelligenceVision-Language-Action ModelImageMultimodality

🎯 What it does: Propose a learning method based on visual and tactile perception, using exploration actions to enable the robot to insert its hand into containers or plastic bag-shaped objects.

Self-Supervised Enhancement for Depth from a Lightweight ToF Sensor with Monocular Images

Laiyan Ding, Rui Huang

Depth EstimationConvolutional Neural NetworkImage

🎯 What it does: Proposed a self-supervised learning framework named SelfToF, which generates scale-aware and detail-rich depth maps by fusing high-resolution RGB images with low-resolution ToF depth maps, and further improved to SelfToF* to adapt to ToF signals with varying sparsity.

Self-Supervised Geometry-Guided Initialization for Robust Monocular Visual Odometry

Takayuki Kanai, Kazuhiro Shintani

Autonomous DrivingSimultaneous Localization and MappingImageVideoBenchmark

🎯 What it does: This study diagnoses the main failure cases of the learning-based dense SLAM model DROID-SLAM on outdoor benchmarks, and proposes utilizing a frozen pre-trained monocular depth estimator as a self-supervised prior to initialize the dense bundle adjustment process, thereby achieving more robust monocular visual odometry;

Self-supervised Monocular Depth Estimation for Dynamic Objects with Ground Propagation

Huan Li, Stefano Mattoccia

Depth EstimationImage

🎯 What it does: Proposes a self-supervised monocular depth estimation method that leverages the relationship between ground contact points and the depth of dynamic objects, recalibrating the depth of dynamic objects by iteratively propagating ground features to moving targets within the perception layer, without adding extra networks or complex training.

Self-Supervised Monocular Visual Drone Model Identification through Improved Occlusion Handling

Stavrow Bahnam, G. D. de Croon

Pose EstimationAutonomous DrivingContrastive LearningSimultaneous Localization and MappingVideo

🎯 What it does: Proposed a self-supervised learning scheme to train a neural network drone model using only monocular video and flight controller data, and improved occlusion handling to enhance relative pose estimation in high-speed, obstacle-proximate environments.

SEM-RRT*: Fast Risk Assessment and Path Planning in Uneven Terrain using Statistical Elevation Map

Xudong Dong, Wenzhe Wang

OptimizationRobotic Intelligence

🎯 What it does: Propose a path planning method called SEM-RRT* for uneven terrain, combining a lightweight statistical elevation map (SEM) representation, an omnidirectional multi-scale terrain risk filter, and integrating multi-objective cost assessment, backward search, and rolling optimization strategies into the Informed RRT* framework.

Semantic Enhancement for Object SLAM with Heterogeneous Multimodal Large Language Model Agents

Jungseok Hong, John J. Leonard

Large Language ModelAgentic AISimultaneous Localization and MappingMultimodality

🎯 What it does: Propose the SEO-SLAM framework, which enhances semantic mapping in object SLAM by leveraging heterogeneous multimodal large language model agents, and improves efficiency and accuracy through asynchronous inference and multi-data association strategies.

Semantic Risk Assessment in Visual Scenes for AUV-Assisted Marine Debris Removal

Sakshi Singh, Junaed Sattar

ClassificationObject DetectionDepth EstimationTransformerPrompt EngineeringVision Language ModelImageMultimodality

🎯 What it does: Propose a method combining a vision-language model with monocular depth estimation for identifying risks in underwater scenarios related to fragment search and removal tasks, and effectively classifying and localizing underwater fragments and surrounding sensitive entities.

Semantic-Geometric-Physical-Driven Robot Manipulation Skill Transfer via Skill Library and Tactile Representation

Mingchao Qi, Panfeng Huang

Robotic IntelligenceTransformerLarge Language ModelMultimodalityChain-of-Thought

🎯 What it does: Construct a skill library based on a knowledge graph and propose a hierarchical skill transfer framework to achieve collaboration between task-level thinking and action-level precise execution;

sEMG-Based Continues Motion Prediction of Shoulder exoskeleton Control Using the VGANet Model*

Tongxin Jiang, Yili Fu

Graph Neural NetworkBiomedical Data

🎯 What it does: Proposed a continuous shoulder joint motion prediction model based on VGANet, which predicts future joint angles using surface electromyography (sEMG) signals and applies them to exoskeleton control systems.

SemGauss-SLAM: Dense Semantic Gaussian Splatting SLAM

Siting Zhu, Hesheng Wang

Pose EstimationOptimizationGaussian SplattingSimultaneous Localization and MappingImagePoint Cloud

🎯 What it does: Proposed the SemGauss-SLAM system, achieving dense semantic SLAM using 3D Gaussian representations, supporting precise semantic mapping, robust camera tracking, and high-quality rendering.

Semi-distributed Cross-modal Air-Ground Relative Localization

Weining Lu, Bin Liang

Autonomous DrivingOptimizationSimultaneous Localization and MappingImageMultimodalityPoint Cloud

🎯 What it does: Implemented a semi-distributed cross-modal aerial-ground relative localization framework, enabling unmanned ground vehicles (UGV) and unmanned aerial vehicles (UAV) to independently complete SLAM and use deep learning to extract keypoints and global descriptors for relative positioning

SemP-NBV: Semantic-Aware Predictive Next-Best-View for Autonomous Plant 3D Reconstruction

Xingjian Li, Lirong Xiang

Robotic IntelligenceImageAgriculture Related

🎯 What it does: Proposed a semantic-aware prediction of the next best view (SemP-NBV) method for efficient 3D plant reconstruction.

SemSegGrasp: Plug-and-play Task-oriented Grasping via Semantic Segmentation

Cheng He, Xuebo Zhang

SegmentationRobotic IntelligenceTransformerLarge Language ModelVision Language ModelTextPoint Cloud

🎯 What it does: Reconstruct task-oriented grasping (TOG) as a semantic segmentation problem based on point cloud and text matching. First, generate local geometric descriptions of target objects using a vision-language model (VLM), then obtain operational instructions by combining user commands with a large language model (LLM). Subsequently, encode features of point clouds and operational instructions, match them using multi-head cross-attention, predict the probability of each point becoming a TOG grasp point, and finally fuse the segmentation results with candidate poses generated by existing grasp detection algorithms to obtain the grasp pose.

SENIOR: Efficient Query Selection and Preference-Guided Exploration in Preference-based Reinforcement Learning

Hexian Ni, Shuo Wang

Computational EfficiencyReinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: Proposes an efficient query selection and preference-guided exploration method named SENIOR, combining two mechanisms: Motion-Distinction-based Selection (MDS) and Preference-Guided Exploration (PGE).

Sensing Differently: Unifying Vision, Language, Posture and Tactile in Robotic Perception

Yanmin Zhou, Bin He

Robotic IntelligenceVision Language ModelMultimodality

🎯 What it does: Proposed a multi-modal grasping dataset VLaPT that aligns visual-language with pose-tactile, and trained a lightweight multi-modal alignment framework CLIP-ME based on this dataset.

Sensor-Free Self-Calibration for Collaborative Robots Using Tri-Sphere End-Effector Toward High Orientation Accuracy

Jianhui He, Wenjun Shen

Pose EstimationRobotic Intelligence

🎯 What it does: A sensorless self-calibration method for collaborative robots based on a three-ball end-effector is proposed, achieving high-precision TCP positioning and pose calibration using position and distance constraints.

Sequen-Sync Contact Force/Torque Control Using Nested Fast Terminal Sliding Mode Control Approach

Yilan Xu, Tong Heng Lee

🎯 What it does: Proposed a nested fast terminal sliding mode control method to achieve sequential and synchronous convergence in force/torque control

Sequential Multi-Object Grasping with One Dexterous Hand

Sicheng He, Daniel Seita

Robotic IntelligenceDiffusion modelPoint Cloud

🎯 What it does: Propose the SeqMultiGrasp system, which uses a four-fingered Allegro hand to sequentially grasp two objects, first fully enclosing and lifting the first object, then grasping the second object without releasing the first.

Service Discovery-Based Hybrid Network Middleware for Efficient Communication in Distributed Robotic Systems

Shiyao Sang, Yinggang Ling

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Designed and implemented the RI-MAOS2C middleware to improve communication efficiency and scheduling stability between Orin computing units in distributed robotic systems, and verified it on L4 vehicles and Jetson Orin.

Servo-Driven Flapping Robot that Uses Its Tail for Self-Standing Takeoff

Kaspul Anuar, N. Takesue

Robotic Intelligence

🎯 What it does: Developed an autonomous takeoff method without using additional mechanisms or external platforms, utilizing the tail for takeoff support, and validated the method through static experiments, parameter tuning, and indoor flight tests.

Set Phasers to Stun: Beaming Power and Control to Mobile Robots with Laser Light

Charles J. Carver, Xia Zhou

Robotic Intelligence

🎯 What it does: Propose the Phaser system, which utilizes a narrow laser beam to provide concurrent wireless power transfer and communication for mobile robots

SF-TIM: A Simple Framework for Enhancing Quadrupedal Robot Jumping Agility by Combining Terrain Imagination and Measurement

Ze Wang, Kaiwei Wang

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed the SF-TIM framework, combining terrain imagination and measurement to enhance the jumping flexibility of quadruped robots while maintaining blind walking capability.

SFExplorer: A Surface-Frontier-based Efficient UAV Exploration Method for Large-Scale Unknown Environments

Peiming Duan, Hui Cheng

Autonomous DrivingComputational Efficiency

🎯 What it does: Proposed an efficient UAV exploration method based on surface frontiers, supporting rapid exploration of large-scale unknown environments.

SGLoc: Semantic Localization System for Camera Pose Estimation from 3D Gaussian Splatting Representation

Beining Xu, Hesheng Wang

Pose EstimationGaussian SplattingImage

🎯 What it does: Propose the SGLoc system, which directly utilizes 3D Gaussian Splatting representation and semantic information to regress the camera's 6DoF pose from a query image without prior pose information.

Shaken, Not Stirred: A Novel Dataset for Visual Understanding of Glasses in Human-Robot Bartending Tasks

Lukás Gajdosech, Stefan Wermter

Object DetectionRobotic IntelligenceImage

🎯 What it does: This paper proposes an automatic annotation pipeline based on RGB-D sensors, which generates labels for each frame of image using depth information. This method is used to construct a new glass object dataset (Glass Object Dataset 3). Subsequently, a baseline model is trained on this dataset, and the model is deployed on the humanoid robot NICOL for robot bartending tasks.

Shape Completion and Real-Time Visualization in Robotic Ultrasound Spine Acquisitions

Miruna Gafencu, Nassir Navab

Robotic IntelligenceBiomedical DataUltrasound

🎯 What it does: Developed a system integrating robotic ultrasound with real-time shape completion for automatically acquiring lumbar ultrasound scans, extracting vertebral surfaces, reconstructing complete spinal structures, and providing real-time interactive visualization.

Shape-Adaptive Planning and Control for a Deformable Quadrotor

Yuze Wu, Fei Gao

OptimizationRobotic Intelligence

🎯 What it does: Proposes a deformation-adaptive UAV trajectory planning and control method, including integrating deformation dynamics into path generation, scalable dynamic A* search, spatiotemporal optimization backend, and enhanced control strategies for external force and torque disturbances.

SheepDA-YOLO: Cross-Domain Adaptive Mean Teacher with Dual-Path Decoupling for Sheep Behavior Recognition

Xinjie Chen, Meili Wang

RecognitionImage TranslationObject DetectionDomain AdaptationConvolutional Neural NetworkContrastive LearningImageAgriculture Related

🎯 What it does: Proposed a sheep behavior recognition framework based on SheepDA-YOLO, aiming to achieve generalization across different sheep sheds with a single annotation.

SHIELD: Safety on Humanoids via CBFs In Expectation on Learned Dynamics

Lizhi Yang, Aaron D. Ames

Robotic IntelligenceReinforcement LearningStochastic Differential Equation

🎯 What it does: Proposes the SHIELD framework, which adds a safety layer based on stochastic discrete-time control barrier functions to existing RL controllers, and achieves probabilistic safety constraints through training a generative stochastic dynamics residual model.

Side Scan Sonar-based SLAM for Autonomous Algae Farm Monitoring

Julian Valdez, Ivan Stenius

Simultaneous Localization and MappingAgriculture RelatedAudio

🎯 What it does: Proposes a SLAM framework based on side-scan sonar for autonomous AUV navigation and structural inspection in seaweed farms.

Signal Temporal Logic Compliant Co-design of Planning and Control

Manas Sashank Juvvi, Pushpak Jagtap

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Proposes a framework that co-designs trajectory planning and control to satisfy autonomous robot tasks based on signal temporal logic (STL).

SILM: A Subjective Intent Based Low-Latency Framework for Multiple Traffic Participants Joint Trajectory Prediction

Weiming Qu, D. Luo

Autonomous DrivingTime Series

🎯 What it does: Proposed a low-latency framework (SILM) based on subjective intent for multi-agent joint trajectory prediction without relying on maps.

SIME: Enhancing Policy Self-Improvement with Modal-level Exploration

Yang Jin, Cewu Lu

Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement LearningMultimodality

🎯 What it does: By introducing a modal-level exploration mechanism and performing data screening during the execution of robot policies, the robot can start from human-provided data, autonomously improve its capabilities through interaction with the environment, and validate its effectiveness in simulation benchmarks and real-world experiments.

SimLauncher: Launching Sample-Efficient Real-World Robotic Reinforcement Learning via Simulation Pre-Training

Mingdong Wu, Hao Dong

Robotic IntelligenceReinforcement Learning

🎯 What it does: Improving sampling efficiency and exploration by pre-training visual motion policies in a simulated environment and applying them to real-world reinforcement learning

Simpler Is Better: Revisiting Doppler Velocity for Enhanced Moving Object Tracking with FMCW LiDAR

Yubin Zeng, Youjin Yu

Object TrackingPoint Cloud

🎯 What it does: Proposes a learning-free tracking method called DopplerTrack based on FMCW LiDAR Doppler velocity, utilizing Doppler information for point cloud preprocessing, target detection, and velocity vector reconstruction to achieve efficient tracking of moving objects.

Simulating Automotive Radar with Lidar and Camera Inputs

Peili Song, Jingtai Liu

Object DetectionGenerationData SynthesisAutonomous DrivingImageMultimodalityPoint Cloud

🎯 What it does: Using camera images, LiDAR point clouds, and vehicle speed data, two novel neural networks, DIS-Net and RSS-Net, are employed to simulate 4D millimeter-wave radar signals (pitch, yaw, distance, Doppler velocity, and RSS)

Simultaneous 6-DOF localization and scanning angle detection of magnetic ultrasound capsule endoscope (MUSCE) with internal sensors

Zhengxin Yang, Yaoyao Cui

Pose EstimationSimultaneous Localization and MappingBiomedical DataUltrasound

🎯 What it does: A composite sensing method is proposed, combining internal magnetic sensor arrays with external permanent magnet sources to achieve 6-degree-of-freedom (6-DOF) positioning and real-time ultrasound beam scanning angle detection for magnetic ultrasound capsule endoscopy (MUSCE), with experimental validation of its performance.

Simultaneous Locomotion Mode Classification and Continuous Gait Phase Estimation for Transtibial Prostheses

Ryan R. Posh, Patrick M. Wensing

ClassificationExplainability and InterpretabilityComputational EfficiencyTime SeriesSequentialBiomedical Data

🎯 What it does: Developed an interpretable and computationally efficient algorithm that can simultaneously perform gait pattern classification and continuous gait phase estimation for trans-tibial amputees.

Simultaneous Pick and Place Detection by Combining SE(3) Diffusion Models with Differential Kinematics

Tianyi Ko, Koichi Nishiwaki

Robotic IntelligenceDiffusion model

🎯 What it does: Propose a method that directly considers placement and reachability constraints during the grasping detection phase, combining SE(3) diffusion networks with multi-objective differential inverse kinematics to generate executable grasping poses.

SimWorld: A Unified Benchmark for Simulator-Conditioned Scene Generation via World Model

Xinqing Li, Yunfeng Ai

GenerationData SynthesisWorld ModelBenchmark

🎯 What it does: Proposed a simulator-conditioned scene generation engine based on a world model, and constructed a corresponding benchmark for scaling virtual and real data ratios.

Single-Microphone-Based Sound Source Localization for Mobile Robots in Reverberant Environments

Jiang Wang, Kazuhiro Nakadai

Robotic IntelligenceAudio

🎯 What it does: Developed an online sound source localization method based on a single microphone, applicable for mobile robots in reverberant environments.

Six-DoF Hand-Based Teleoperation for Omnidirectional Aerial Robots

Jinjie Li, Moju Zhao

Robotic Intelligence

🎯 What it does: Designed and implemented a six-degree-of-freedom teleoperation system based on hand movements for omnidirectional drones, with four interaction modes, validated its effectiveness in a vertical valve rotation task.

SkB-Hand: A Skeleton Bionic Hand with Dual-Tendon for General Purpose Robotic Grasping Tasks

Duan-Hong Yang, Chung-Hsien Kuo

Robotic Intelligence

🎯 What it does: Developed a lightweight, low-cost Skeleton Bionic Hand (SkB-Hand) that uses a dual elastic tendon mechanism to control extensor muscles and palmar plates, suitable for various grasping tasks.

Skeleton-Guided Rolling-Contact Kinematics for Arbitrary Point Clouds via Locally Controllable Parameterized Curve Fitting

Qingmeng Wen, Seyed Amir Tafrishi

Point CloudPhysics Related

🎯 What it does: Propose a framework based on differential geometry that uses local parameterization to model continuous rolling contact on discrete point clouds. The method defines a rotating reference structure using skeletonization, extracts controllable local geometric structures via Fourier curve fitting, introduces a novel 2D manifold coordinate system for local parameterization of arbitrary surfaces, derives kinematic equations for rolling contact, and validates the approach through multi-body simulations.

SkinGrip: An Adaptive Soft Robotic Manipulator with Capacitive Sensing for Whole-Limb Bed Bathing Assistance

Fukang Liu, Zackory Erickson

Safty and PrivacyRobotic Intelligence

🎯 What it does: Proposed a flexible, scalable soft robotic manipulator equipped with a capacitive proximity sensing array for safe and efficient bed bath assistance.

SKT: Integrating State-Aware Keypoint Trajectories with Vision-Language Models for Robotic Garment Manipulation

Xin Li, Cewu Lu

Data SynthesisPose EstimationRobotic IntelligenceTransformerVision Language ModelImage

🎯 What it does: Using a vision-language model to uniformly predict key points across different clothing categories, enabling robots to operate in various clothing states.

SkyVLN: Vision-and-Language Navigation and NMPC Control for UAVs in Urban Environments

Tianshun Li, Xinhu Zheng

Autonomous DrivingOptimizationRobotic IntelligenceLarge Language ModelVision-Language-Action ModelMultimodality

🎯 What it does: Proposed the SkyVLN framework, integrating vision-and-language navigation (VLN) with nonlinear model predictive control (NMPC) to enhance the autonomous navigation capabilities of drones in urban environments.

SLAM-Based Performance Evaluation of Industrial Robotic Arms*

Chieh-Yu Liao, Han-Pang Huang

Robotic IntelligenceSimultaneous Localization and MappingPoint CloudBenchmark

🎯 What it does: Proposed and evaluated a LiDAR-based SLAM method for performance assessment of industrial robot arms.

SLOOP: Aligned Coordinate System-aided LiDAR LOOP Closure Detection based on Semantic Node Graph Matching

Yujie Tang, Yufeng Yue

Autonomous DrivingGraph Neural NetworkSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes SLOOP, a LiDAR loop closure detection method that first aligns semantic maps and then performs efficient similarity comparison.

SLTNet: Efficient Event-based Semantic Segmentation with Spike-driven Lightweight Transformer-based Networks

Xianlei Long, Fuqiang Gu

SegmentationSpiking Neural NetworkTransformerVideo

🎯 What it does: Proposed an event-driven semantic segmentation method based on a spike-driven lightweight transformer network (SLTNet)

SLU-DQN: A Model for Anticipatory Steam Detection for Steamer-Filling in Baijiu Intelligent Distillation Systems

Jia Yu, Yunquan Sun

OptimizationConvolutional Neural NetworkTransformerReinforcement Learning

🎯 What it does: Propose the SLU-DQN model to predict steam generation areas and implement a pre-arrangement plan in the distillation system, addressing the ASDSF problem in the smart distillation system for sake brewing.

SMA-TENG Actuator with Tactile Sensing Capability

Yiping Zhang, Ziyu Ren

Robotic Intelligence

🎯 What it does: Developed an integrated tactile sensing and actuation SMA-TENG actuator that can achieve tactile detection while maintaining driving force.

SmartWay: Enhanced Waypoint Prediction and Backtracking for Zero-Shot Vision-and-Language Navigation

Xiangyu Shi, Qi Wu

Autonomous DrivingTransformerLarge Language ModelVision Language ModelMultimodalityBenchmark

🎯 What it does: Propose a zero-shot visual language navigation (VLN-CE) framework that improves the path point predictor and navigator, achieving better spatial awareness and historical reasoning.

SN-LiDAR: Semantic Neural Fields for Novel Space-time View LiDAR Synthesis

Yi Chen, Jingchuan Wang

SegmentationData SynthesisConvolutional Neural NetworkPoint Cloud

🎯 What it does: Jointly accomplish precise semantic segmentation, high-quality geometric reconstruction, and realistic LiDAR synthesis.

Snuggle-Pack: Speeding Up Multi-Heuristic Packing Planning of Complex Objects

Tim Nickel, Kai O. Arras

OptimizationBenchmark

🎯 What it does: Proposed the Snuggle-Pack algorithm, achieving efficient 3D object packing and positioning.

Social Robot Haru Assisting Dynamic Group Discussion with Autonomous Eye Gaze Behavior

Fei Tang, Guangliang Li

Robotic IntelligenceTransformerLarge Language ModelReinforcement Learning

🎯 What it does: Designed and implemented a social robot system named Haru, consisting of three modules, to assist in dynamic multi-party discussions, including dialogue assistance, balanced engagement, and autonomous gaze behavior.

Social-LLaVA: Enhancing Social Robot Navigation through Human-Language Reasoning

Amirreza Payandeh, Xuesu Xiao

Robotic IntelligenceLarge Language ModelSupervised Fine-TuningVision Language ModelVision-Language-Action ModelMultimodality

🎯 What it does: Constructed a human-annotated visual question answering dataset named SNEI, and fine-tuned the Vision-Language model Social-LLaVA on it to achieve social robot navigation based on language reasoning.

Socially-Aware Robot Navigation Enhanced by Bidirectional Natural Language Conversations Using Large Language Models

Congcong Wen, Yi Fang

Robotic IntelligenceLarge Language ModelReinforcement LearningText

🎯 What it does: Proposes the HSAC-LLM framework, combining deep reinforcement learning with large language models (LLMs) to enable social perception navigation for robots in shared spaces, and actively resolves conflicts with pedestrians through bidirectional natural language dialogue.

Soft Actuators with Integrated Electrohydrodynamic Pump and Intrinsic Electroadhesion

Yuki Sato, Jun Shintake

Robotic Intelligence

🎯 What it does: Integrated EHD pump, actuator, and reservoir on flexible materials to form a single soft actuator, achieving electrical-driven bending and electro-adhesion functions.

Soft Electrohydraulic Actuators with Intrinsic Electroadhesion

Takumi Shibuya, Jun Shintake

Robotic IntelligencePhysics Related

🎯 What it does: Proposed an integrated soft electrohydraulic actuator with electro-controlled adhesion functionality, and experimentally verified its driving and adhesion performance.

Soft Tactile Sensors for Robot Grippers Using Acoustic Sensing

Kevin Xu, Justin Chan

Robotic IntelligenceAudio

🎯 What it does: Designed and implemented a low-cost, flexible tactile sensor that utilizes active acoustics technology to detect deformation during robotic grasping through embedded speakers and microphones, thereby achieving tactile perception.

Soft Wearable Robotic Kit for Forearm Rotation and Grasping Motion Tracking Based on Embedded End-Effector-Level Sensor System

Huimin Su, L. Masia

Robotic IntelligenceTime Series

🎯 What it does: Implemented a flexible wearable robotic suit based on end-effector level sensors for tracking forearm rotation and grasping movements, and proposed a wearable robotic suit composed of sensor-only exoskeletons and motorized exoskeletons, utilizing a leader-follower control mode to achieve motion capture and rapid response.

Soft-Rigid Coupled Blade Leg Achieves Spatio-temporal Terrain Classification with Minimal Sensor Configuration

Chapa Sirithunge, Fumiya Iida

Classification

🎯 What it does: Designed and tested a blade leg with soft-rigid coupling and minimal pressure sensors for spatiotemporal terrain classification under low-sensor configurations;

SOLO-SMap: Semantic-Aided Online LiDAR Odometry and 3D Static Mapping for Dynamic Scenes

Ruyi Li, Youwei Wang

Object TrackingAutonomous DrivingSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Developed SOLO-SMap, a real-time localization and static map building framework based on LiDAR point clouds, which utilizes semantic reasoning to identify potential dynamic points and removes real dynamic points during the pre-alignment phase through geometric rules and multi-object tracking, maintaining static constraints to improve localization accuracy.

SORT3D: Spatial Object-centric Reasoning Toolbox for Zero-Shot 3D Grounding Using Large Language Models

Nader Zantout, Wenshan Wang

Object DetectionData-Centric LearningTransformerLarge Language ModelImageTextChain-of-Thought

🎯 What it does: Propose the SORT3D method, which combines 2D object attributes with a heuristic spatial reasoning toolbox and LLM's sequential reasoning to achieve zero-shot 3D object localization.