arXivSub Start free trial

ICRA 2024 Papers — Page 17

IEEE International Conference on Robotics and Automation · 1760 papers

Toward Mass Customization of a Robot’s Morphology Design for Improving Area Coverage

M. V. J. Muthugala, Mohan Rajesh Elara

OptimizationRobotic Intelligence

🎯 What it does: Proposed a system for robot morphology customization to enhance the area coverage of ground cleaning robots, and evaluated the coverage performance of candidate morphologies through simulation

Toward Optimal Tabletop Rearrangement with Multiple Manipulation Primitives

Baichuan Huang, Jingjin Yu

OptimizationRobotic Intelligence

🎯 What it does: Proposed and studied an algorithm that uses multiple manipulation primitives (such as grasp-place and push) to plan high-quality action sequences for desktop rearrangement tasks

Toward Self-Righting and Recovery in the Wild: Challenges and Benchmarks

Rosario Scalise, Chad C. Kessens

Knowledge DistillationRobotic IntelligenceReinforcement LearningBenchmark

🎯 What it does: Presents the challenge of wild self-recovery, trains a set of benchmark self-righting strategies using deep reinforcement learning and student-teacher methods in simulation, evaluates their performance in simulated benchmark environments, and validates the baseline on physical robots.

Toward Wheeled Mobility on Vertically Challenging Terrain: Platforms, Datasets, and Algorithms

A. Datar, Xuesu Xiao

Robotic Intelligence

🎯 What it does: Proposed two lightweight hardware-modified wheeled platforms, collected datasets of their performance on vertical challenging terrains, and presented an algorithm to demonstrate their mobility potential in such environments.

Towards a Novel Soft Magnetic Laparoscope for Single Incision Laparoscopic Surgery

Hui Liu, Jindong Tan

Biomedical Data

🎯 What it does: Designed and verified a flexible magnetic laparoscope probe capable of bending through soft structures while maintaining magnetic coupling during single-port laparoscopic surgery.

Towards a Unified Approach for Continuously-Variable Impedance Control of Powered Prosthetic Legs over Walking Speeds and Inclines

Albert J. Lee, Robert D. Gregg

OptimizationRobotic IntelligenceBiomedical Data

🎯 What it does: Designed and implemented a unified controller for a powered prosthetic foot with continuous damping parameter trajectories, covering the entire gait cycle.

Towards Centimeter-Scale Underwater Mobile Robots: An Architecture for Capable µAUVs

P. Spino, Daniela Rus

Robotic Intelligence

🎯 What it does: Designed a novel micro autonomous underwater vehicle (µAUV)

Towards Enhanced Human Activity Recognition for Real-World Human-Robot Collaboration

Beril Yalçinkaya, Fabio Remondino

RecognitionRobotic IntelligenceRecurrent Neural NetworkTime Series

🎯 What it does: Improved the FS-LSTM architecture to handle uncertain and irregular sensor data in real-world environments

Towards fault-tolerant deployment of mobile robot navigation in the edge: an experimental study

Florian Mirus, Kay-Ulrich Scholl

Robotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Propose and verify a minimal safety fallback mechanism for mobile robot navigation when offloaded to edge computing, ensuring safe and collision-free navigation even when the connection to edge devices fails.

Towards Feasible Dynamic Grasping: Leveraging Gaussian Process Distance Field, SE(3) Equivariance, and Riemannian Mixture Models

Ho Jin Choi, Nadia Figueroa

Robotic Intelligence

🎯 What it does: Propose a method for achieving feasible grasping in dynamic environments, integrating Gaussian Process Distance Field, SE(3) transformation-invariant networks, and Riemannian hybrid models to realize object shape reconstruction, grasp sampling, and implicit grasp pose selection;

Towards Generalizable Zero-Shot Manipulation via Translating Human Interaction Plans

Homanga Bharadhwaj, Shubham Tulsiani

Robotic IntelligenceVision-Language-Action ModelVideo

🎯 What it does: Developed a system that learns 'plans' from human videos and translates them into robot actions, achieving zero-shot generalization capability.

Towards Geometric Motion Planning for High-Dimensional Systems: Gait-Based Coordinate Optimization and Local Metrics

Yanhao Yang, Ross L. Hatton

OptimizationRobotic Intelligence

🎯 What it does: A gait-based coordinate optimization method is proposed to overcome the curse of dimensionality in high-dimensional systems, unify various nonholonomic constraints into local metrics, and combine both to achieve geometric motion planning for high-dimensional systems.

Towards Large-Scale Incremental Dense Mapping using Robot-centric Implicit Neural Representation

Jianheng Liu, Haoyao Chen

Robotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Proposed the Robot-Centric Implicit Mapping (RIM) technique for large-scale incremental dense mapping;

Towards learning-based planning: The nuPlan benchmark for real-world autonomous driving

Napat Karnchanachari, Holger Caesar

Autonomous DrivingVideoBenchmark

🎯 What it does: Built the first real-world autonomous driving dataset nuPlan and its evaluation benchmark, providing a closed-loop simulation framework to test the safety and efficiency of machine learning planners in diverse driving scenarios.

Towards Motion Forecasting with Real-World Perception Inputs: Are End-to-End Approaches Competitive?

Yihong Xu, P. Pérez

Autonomous DrivingSupervised Fine-TuningBenchmark

🎯 What it does: Propose a unified evaluation pipeline to compare traditional and end-to-end motion prediction methods under real perceptual inputs, and conduct an in-depth analysis of the performance gap between curated data and perceptual data.

Towards Optimal Lane-changing Coordination of CAVs in Multi-lane Mixed Traffic Scenarios

Yan Ding, Pengju Ren

Autonomous DrivingOptimization

🎯 What it does: Designed a collaborative lane-changing method based on Conflict-Based Search (CBS) for lane-changing coordination of connected autonomous vehicles (CAV) in mixed traffic scenarios.

Towards Proactive Safe Human-Robot Collaborations via Data-Efficient Conditional Behavior Prediction

Ravi Pandya, Changliu Liu

Safty and PrivacyRobotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a model-based conditional behavior prediction framework that infers human intent by reasoning about the robot's future behavior, enabling more efficient human-robot collaboration;

Towards Real-World Efficiency: Domain Randomization in Reinforcement Learning for Pre-Capture of Free-Floating Moving Targets by Autonomous Robots

Bahador Beigomi, Zheng H. Zhu

Robotic IntelligenceReinforcement Learning

🎯 What it does: Developed a control method based on deep reinforcement learning for completing the robot's pre-grasping phase in a microgravity environment.

Towards Robo-Coach: Robot Interactive Stiffness/Position Adaptation for Human Strength and Conditioning Training

Chenzui Li, Fei Chen

Robotic Intelligence

🎯 What it does: A robot coach system for adjustable load resistance training through physical human-robot interaction

Towards Robot to Human Skill Coaching: A ML-powered IoT and HRI Platform for Martial Arts Training

Katia Bourahmoune, Marc G. Carmichael

Robotic IntelligenceTime SeriesSequential

🎯 What it does: Built and real-time deployed a machine learning-based physical robot-human interaction skill training platform for Kendo Iaido, capturing movements using a self-developed IoT katana sensor, and evaluated its effectiveness in simulated combat and questionnaires.

Towards Robotic Tree Manipulation: Leveraging Graph Representations

Chung Hee Kim, George Kantor

Robotic IntelligenceGraph Neural NetworkGraphAgriculture Related

🎯 What it does: Proposes a framework for learning the deformation behavior of tree-shaped crops under contact interactions, encoding the state of tree crops using a spring-damper model into a graph and utilizing graph networks to learn forward deformation models and contact control strategies.

Towards Safe Robot Use with Edged or Pointed Objects: A Surrogate Study Assembling a Human Hand Injury Protection Database

R. Kirschner, Sami Haddadin

Safty and PrivacyBiomedical Data

🎯 What it does: Using pig paws and chicken legs as surrogates for human hands in drop tests to construct a human hand injury prevention database, and demonstrating the usability and efficiency of robots in two scenarios using this database.

Towards Solving Cable-Driven Parallel Robot Inaccuracy due to Cable Elasticity

Adolfo Suarez-Roos, St´ephane Caro

Robotic IntelligencePhysics Related

🎯 What it does: Propose a numerical method called SEECR for estimating the behavior of CDPRs with elastic cables and ensuring the static balance of the motion platform.

Towards Standardized Disturbance Rejection Testing of Legged Robot Locomotion with Linear Impactor: A Preliminary Study, Observations, and Implications

Bowen Weng, Ayonga Hereid

Robotic IntelligenceBenchmark

🎯 What it does: Propose using a linear impulse actuator for standardized perturbation resistance testing of legged robot locomotion, conducting experiments on the Digit humanoid robot to compare the performance of three walking controllers.

Towards Unified Interactive Visual Grounding in The Wild

Jie Xu, Tao Kong

RetrievalVision Language ModelMultimodalityBenchmark

🎯 What it does: Propose TiO, an end-to-end interactive visual localization system that unifies visual dialogue and visual localization in human-computer interaction, actively collecting information to resolve linguistic ambiguities.

Towards Unifying Human Likeness: Evaluating Metrics for Human-Like Motion Retargeting on Bimanual Manipulation Tasks

Andre Meixner, Tamim Asfour

Robotic Intelligence

🎯 What it does: Propose a unified human similarity metric and evaluate it in motion retargeting for dual-arm collaboration tasks.

Towards Visibility Estimation and Noise-Distribution-Based Defogging for LiDAR in Autonomous Driving

Jie Zhan, Jie Ma

Autonomous DrivingPoint Cloud

🎯 What it does: Propose a LiDAR point cloud de-fogging method based on noise distribution, which estimates the fog attenuation coefficient using road prior, fuses the fog-induced noise distribution with the spatial non-uniform distribution caused by LiDAR structure, and then achieves de-fogging through statistical filtering based on noise relative sparsity.

TP3M: Transformer-based Pseudo 3D Image Matching with Reference Image

Liming Han, Shiguo Lian

RetrievalTransformerImage

🎯 What it does: Propose a pseudo-3D image matching method based on the Transformer framework, which upgrades the 2D features of the source image to 3D features using a reference image, and performs two-level (coarse-to-fine) 3D matching with the 2D features of the target image.

TPGP: Temporal-Parametric Optimization with Deep Grasp Prior for Dexterous Motion Planning

Haoming Li, Jiming Chen

OptimizationRobotic Intelligence

🎯 What it does: Propose a temporal parameterized grasp prior optimization method to simplify dexterous hand grasp trajectory planning while maintaining smooth and natural trajectories

Tracking Snake-Like Robots in the Wild Using Only a Single Camera

Jingpei Lu, Michael C. Yip

Robotic IntelligenceImage

🎯 What it does: Tracking a snake-like robot using a single camera combined with differentiable rendering and Kalman filtering

Tractable Joint Prediction and Planning over Discrete Behavior Modes for Urban Driving

Adam R. Villaflor, Jeff Schneider

Autonomous DrivingTransformerSequential

🎯 What it does: By using anchor embeddings from a multi-modal trajectory prediction model as discrete behavior mode parameters, a fully reactive autoregressive closed-loop planning framework is constructed, capable of real-time processing of dynamic interactions and avoiding the 'frozen robot' problem common in traditional planners.

Traffic Flow Learning Enhanced Large-Scale Multi-Robot Cooperative Path Planning Under Uncertainties

Xingyao Han, Zhe Liu

Autonomous DrivingOptimizationRobotic IntelligenceGraph Neural Network

🎯 What it does: Proposes a hierarchical framework to achieve large-scale multi-robot collaborative path planning, including traffic flow prediction, lane-level planning, and road-level coordination, while considering motion/communication uncertainty.

Trajectory Optimization for Cooperatively Localizing Quadrotor UAVs

H. Go, Hugh H. T. Liu

OptimizationSimultaneous Localization and Mapping

🎯 What it does: Developed an active cooperative localization system and reduced the localization uncertainty of multirotor drones through trajectory optimization methods.

Trajectory Optimization Strategy That Considers Body Tip-Over Stability, Limb Dynamics, and Motion Continuity in Legged Robots

Kuan-Lun Lu, Pei-Chun Lin

OptimizationRobotic Intelligence

🎯 What it does: Propose a quadruped robot arm trajectory planning method considering body and limb dynamics, optimized using a genetic algorithm;

Trajectory Prediction for Robot Navigation using Flow-Guided Markov Neural Operator

Rashmi Bhaskara, Aniket Bera

Robotic IntelligenceOptical FlowVideoTime Series

🎯 What it does: Proposes the FlowMNO model for pedestrian trajectory prediction in robot navigation.

Trajectory Tracking Runtime Assurance for Systems with Partially Unknown Dynamics

M. Cao, Samuel Coogan

🎯 What it does: This paper proposes a method for tracking a reference trajectory under state-dependent unknown disturbances.

Trajectory-prediction-based Dynamic Tracking of a UGV to a Moving Target under Multi-disturbed Conditions

Jinge Si, Junzheng Wang

Autonomous DrivingOptimization

🎯 What it does: Proposes a UGV dynamic tracking scheme based on trajectory prediction, including target localization, trajectory prediction, and UGV control, capable of tracking moving targets under multi-disturbance conditions.

Transformer-Based Prediction of Human Motions and Contact Forces for Physical Human-Robot Interaction

A. Fusco, Marco Cognetti

Robotic IntelligenceTransformerTime SeriesSequential

🎯 What it does: Proposes a Transformer-based network for predicting contact forces and human motion in physical human-robot interaction

Transformer-CNN Cohort: Semi-supervised Semantic Segmentation by the Best of Both Students

Xueye Zheng, Lin Wang

SegmentationKnowledge DistillationConvolutional Neural NetworkTransformerImage

🎯 What it does: Proposed the Transformer-CNN Cohort (TCC) framework, which achieves semi-supervised semantic segmentation using two students (a Vision Transformer and a CNN) through multi-layer consistency regularization and pseudo-labeling;

Translating Universal Scene Descriptions into Knowledge Graphs for Robotic Environment

G. Nguyen, Michael Beetz

Robotic IntelligenceGraph

🎯 What it does: The study uses virtual reality technology to achieve robot environment modeling and proposes a method to convert scene graphs in Universal Scene Description (USD) format into knowledge graphs, facilitating semantic queries and integration with knowledge sources.

Transparency Control of a 1-DoF Knee Exoskeleton via Human-in-the-Loop Velocity Optimisation

Lukas Cha, Ravi Vaidyanathan

OptimizationRobotic IntelligenceTime Series

🎯 What it does: Achieve transparent control of a 1-DoF knee exoskeleton through a human-in-the-loop velocity optimization framework, combining adaptive frequency oscillators and force sensors for position control.

Tree Instance Segmentation and Traits Estimation for Forestry Environments Exploiting LiDAR Data Collected by Mobile Robots

Meher V. R. Malladi, C. Stachniss

RecognitionSegmentationPoint CloudAgriculture Related

🎯 What it does: Automatically extract tree instances and estimate the diameter at breast height and position of each tree using LiDAR point clouds collected by a mobile platform to generate forest inventory tables.

Tree-based Representation of Locally Shortest Paths for 2D k-Shortest Non-homotopic Path Planning

Tong Yang, Rong Xiong

Optimization

🎯 What it does: Proposing a tree-based 2D k-shortest non-homotopic path planning algorithm

TreeScope: An Agricultural Robotics Dataset for LiDAR-Based Mapping of Trees in Forests and Orchards

Derek Cheng, Vijay Kumar

SegmentationDepth EstimationRobotic IntelligencePoint CloudBenchmarkAgriculture Related

🎯 What it does: Collected and released the TreeScope v1.0 dataset, which includes LiDAR data acquired from UAV and mobile robot platforms, 1,800+ trunk semantic labels, and measured tree diameters, along with benchmark scripts and baseline results of open-source algorithms.

TRTM: Template-based Reconstruction and Target-oriented Manipulation of Crumpled Cloths

Wenbo Wang, Stelian Coros

RestorationImageMesh

🎯 What it does: Propose the TRTM system to achieve single-view depth map template-based reconstruction and goal-oriented manipulation of wrinkled cloth.

Trust Recognition in Human-Robot Cooperation Using EEG

Caiyue Xu, Bin He

RecognitionRobotic IntelligenceTransformerBiomedical Data

🎯 What it does: Developed a trust recognition method based on electroencephalogram (EEG) for human-machine collaboration scenarios.

Trust-Aware Motion Planning for Human-Robot Collaboration under Distribution Temporal Logic Specifications

Pian Yu, Marta Z. Kwiatkowska

Autonomous DrivingOptimizationReinforcement Learning

🎯 What it does: A trust-based motion planning method is proposed, utilizing partially observable Markov decision processes (POMDP) and syntactically safe linear dynamic temporal logic (scLDTL). By constructing the product of belief MDP and automaton, an improved point-based value iteration algorithm is used to solve for the optimal strategy, with human-robot collaboration experiments conducted in a driving simulator to verify its effectiveness.

Trust-Region Neural Moving Horizon Estimation for Robots

Bingheng Wang, Lin Zhao

OptimizationRobotic IntelligenceTime SeriesSequential

🎯 What it does: Proposed and trained a NeuroMHE based on Trust-Region Policy Optimization, and validated its performance on quadrotor flight data.

TSCM: A Teacher-Student Model for Vision Place Recognition Using Cross-Metric Knowledge Distillation

Yehui Shen, Xieyuanli Chen

RecognitionKnowledge DistillationImage

🎯 What it does: This paper proposes a teacher-student model TSCM, which achieves the visual place recognition (VPR) task by utilizing span-based knowledge distillation;

TVFusionGAN: Thermal-Visible Image Fusion Based on Multi-level Adversarial Network Strategy

Guoyu Lu

GenerationGenerative Adversarial NetworkImageMultimodality

🎯 What it does: Proposed an end-to-end thermal-visible image fusion network based on generative adversarial networks (GANs), which utilizes a generator and two discriminators to fuse salient features from both modalities.

TWIST: Teacher-Student World Model Distillation for Efficient Sim-to-Real Transfer

Jun Yamada, Ingmar Posner

Domain AdaptationComputational EfficiencyKnowledge DistillationRobotic IntelligenceWorld ModelImage

🎯 What it does: Propose the TWIST method, achieving efficient sim-to-real transfer in visual foundation model learning through teacher-student world model distillation.

Two-Stage Learning of Highly Dynamic Motions with Rigid and Articulated Soft Quadrupeds

Francecso Vezzi, C. D. Santina

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes a two-stage learning framework that uses gradient-free evolutionary strategies and deep reinforcement learning to generate dynamic quadruped robot motions.

UAV-Sim: NeRF-based Synthetic Data Generation for UAV-based Perception

Christopher Maxey, Heesung Kwon

Object DetectionData SynthesisNeural Radiance FieldImage

🎯 What it does: Generate high-altitude drone images using neural rendering technology, enhance static and dynamic novel view synthesis, and train object detection models with a mixture of real and synthetic data.

UIVNAV: Underwater Information-driven Vision-based Navigation via Imitation Learning

Xiao-sheng Lin, Y. Aloimonos

Robotic IntelligenceReinforcement Learning from Human FeedbackImage

🎯 What it does: Proposed a vision-based underwater navigation system called UIVNAV, which guides a robot to patrol and avoid obstacles in underwater target areas (OOI) using imitation learning without relying on localization.

UKF-Based Sensor Fusion for Joint-Torque Sensorless Humanoid Robots

Ines Sorrentino (Istituto Italiano di Tecnologia), Daniele Pucci (Istituto Italiano di Tecnologia)

Robotic IntelligenceMultimodality

🎯 What it does: Propose a multi-modal sensor fusion method based on the unscented Kalman filter (UKF) for online estimation of joint torque in humanoid robots without joint torque sensors, and integrate it into a two-layer torque control architecture;

Ultrafast capturing in-flight objects with reprogrammable working speed ranges

Yongkang Jiang, Yingtian Li

Robotic Intelligence

🎯 What it does: Developed an ultra-fast gripper with a reprogrammable speed range

Ultrafast Square-Root Filter-based VINS

Yuxiang Peng, Guoquan Huang

Pose EstimationSimultaneous Localization and Mapping

🎯 What it does: Developed a square root filtering-based visual-inertial navigation system (SR-VINS) that achieves efficient, numerically stable real-time localization and mapping on edge devices

Uncertainty-aware 3D Object-Level Mapping with Deep Shape Priors

Ziwei Liao, Steven L. Waslander

Pose EstimationOptimizationSimultaneous Localization and MappingImageMultimodality

🎯 What it does: Construct a high-quality object-level 3D map of unknown objects using multi-frame RGB-D images, outputting dense 3D shapes and a 9-degree-of-freedom (DOF) pose with three scale parameters, while modeling uncertainties in both shape and pose.

Uncertainty-Aware Contextual Visualization for Human Supervision of OCT-Guided Autonomous Robotic Subretinal Injection

Michael Sommersperger, Nassir Navab

Robotic IntelligenceConvolutional Neural NetworkBiomedical Data

🎯 What it does: Proposes a focus and context visualization method that integrates uncertainty information into intraoperative optical coherence tomography (iOCT) data during robotic autonomous injection, to enhance the effectiveness of manual supervision.

Uncertainty-aware hybrid paradigm of nonlinear MPC and model-based RL for offroad navigation: Exploration of transformers in the predictive model

F. Lotfi, Gregory Dudek

Autonomous DrivingTransformerReinforcement Learning

🎯 What it does: Proposes a hybrid scheme combining nonlinear model predictive control (MPC) with model-based reinforcement learning (RL) for off-road vehicle navigation planning in predefined maps.

Uncertainty-aware Reinforcement Learning for Autonomous Driving with Multimodal Digital Driver Guidance

Wenhui Huang, Chen Lv

Autonomous DrivingReinforcement LearningMultimodality

🎯 What it does: Propose an improved Learning from Intervention (LfI) method that learns optimal RL strategies through collaborative interventions of multimodal human behavior and N digital drivers, embedding these digital drivers into the RL training loop while enhancing the RL architecture and optimization objectives to form the uncertainty-aware reinforcement learning (UnaRL) algorithm.

Uncertainty-Aware Shape Estimation of a Surgical Continuum Manipulator in Constrained Environments using Fiber Bragg Grating Sensors

Alexander Schwarz, Mehran Armand

Robotic IntelligenceBiomedical Data

🎯 What it does: A deep neural network is used to directly estimate the shape of a continuum manipulator from the wavelength of a fiber Bragg grating (FBG) sensor, with uncertainty estimation integrated to quantify model confidence.

Uncertainty-bounded Active Monitoring of Unknown Dynamic Targets in Road-networks with Minimum Fleet

Shuaikang Wang, Meng Guo

Autonomous DrivingOptimization

🎯 What it does: This paper proposes an online task and motion coordination algorithm that can perform uncertainty-constrained active monitoring of unknown dynamic targets in road networks while minimizing the average number of active robots.

Uncertainty-driven Exploration Strategies for Online Grasp Learning

Yitian Shi, Ngo Anh Vien

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes an online grasp learning method based on uncertainty, using exploration strategies to enhance the adaptability of robotic bin picking.

UncertaintyTrack: Exploiting Detection and Localization Uncertainty in Multi-Object Tracking

Chang Won Lee, Steven L. Waslander

Object DetectionObject TrackingAutonomous DrivingVideo

🎯 What it does: Propose the UncertaintyTrack extension, which leverages the localization uncertainty from probabilistic object detectors to enhance multi-object tracking

Unconstrained Model Predictive Control for Robot Navigation under Uncertainty

Senthil Hariharan Arul, Dinesh Manocha

OptimizationRobotic Intelligence

🎯 What it does: Propose a probabilistic unconstrained model predictive control framework for robot navigation under uncertainty.

Under pressure: learning-based analog gauge reading in the wild

Maurits Reitsma, Roland Siegwart

RecognitionExplainability and InterpretabilityRobotic IntelligenceImage

🎯 What it does: Proposes an interpretable framework for reading simulated gauges, deployable on real robotic systems, decomposing the reading task into multiple steps and detecting potential failures at each step; does not require prior knowledge of gauge types or scale ranges, and can extract the units used.

Underwater Dome-Port Camera Calibration: Modeling of Refraction and Offset through N-Sphere Camera Model

Monika Roznere, Alberto Quattrini Li

ImagePhysics Related

🎯 What it does: Designed and verified N-Sphere and Shifted N-Sphere camera models for underwater waterproof housing top plate cameras, addressing optical effects such as lens distortion, housing refraction, and camera offset.

Underwater Volumetric Mapping using Imaging Sonar and Free-Space Modeling Approach

António J. Oliveira, N. Cruz

Simultaneous Localization and Mapping

🎯 What it does: Propose an underwater volume mapping technique based on imaging sonar and free space modeling, which utilizes a free space model for each sonar pulse. A volume method is employed to generate a grid subgraph of unoccupied water, and the occupied space representation is obtained by exploring the free space frontier.

Unifying Foundation Models with Quadrotor Control for Visual Tracking Beyond Object Categories

Alessandro Saviolo, Giuseppe Loianno

Object TrackingRobotic IntelligenceLarge Language ModelImageVideo

🎯 What it does: Propose a visual tracking framework based on a foundational model, integrating multi-layer trackers and model-agnostic controllers to enable visual tracking of drones in different scenarios.

Unifying Local and Global Multimodal Features for Place Recognition in Aliased and Low-Texture Environments

Alberto García-Hernández, Rudolph Triebel

RetrievalConvolutional Neural NetworkTransformerImagePoint Cloud

🎯 What it does: Proposed a model called UMF for place recognition in environments with perceptual confusion and low texture.

UniGen: Unified Modeling of Initial Agent States and Trajectories for Generating Autonomous Driving Scenarios

R. Mahjourian, Shimon Whiteson

Autonomous DrivingVideo

🎯 What it does: Proposed UniGen, a method that generates new traffic scenarios through simulation to evaluate and improve autonomous driving software.

Universal Visual Decomposer: Long-Horizon Manipulation Made Easy

Zichen Zhang, Luca Weihs

Robotic IntelligenceReinforcement LearningVideo

🎯 What it does: Propose a Universal Visual Decomposer (UVD) that leverages pre-trained visual representations to detect phase changes in the embedding space, automatically discovers subgoals from visual demonstrations, and trains goal-conditioned policies and performs reward shaping using these subgoals without incurring additional training costs.

Unknown Object Grasping for Assistive Robotics

Elle Miller, Jorn Vogel

Robotic IntelligenceImage

🎯 What it does: Propose a pipeline for unknown object grasping in shared robot autonomous control, using stereo images for instance segmentation and 3D reconstruction, and achieving grasping through a physics-driven grasping planner combined with user-shared control.

Unknown Object Retrieval in Confined Space through Reinforcement Learning with Tactile Exploration

Xinyuan Zhao, Yan Wu

RetrievalRobotic IntelligenceReinforcement Learning

🎯 What it does: Investigated the technology of using a tactile-perceiving tool rod to retrieve unknown objects in confined spaces.

Unlocking Versatile Locomotion: A Novel Quadrupedal Robot with 4-DoFs Legs for Roller Skating

Jiawei Chen, Xilun Ding

OptimizationRobotic Intelligence

🎯 What it does: Designed and verified a quadruped robot with 4 degrees of freedom per leg, supporting multiple wheeled sliding gaits (Swizzling, Stroking, trot-like) and capable of achieving small-radius turning motion.

Unraveling the Single Tangent Space Fallacy: An Analysis and Clarification for Applying Riemannian Geometry in Robot Learning

Noémie Jaquier, Tamim Asfour

Robotic Intelligence

🎯 What it does: Analyze and clarify the fallacy of the single tangent space, provide experimental evidence, and propose best practice recommendations.

UNRealNet: Learning Uncertainty-Aware Navigation Features from High-Fidelity Scans of Real Environments

S. Triest, A. Agha-mohammadi

Autonomous DrivingSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Propose UNRealNet, an unlabeled deep learning model that predicts high-fidelity dense density features from low-quality single-frame LiDAR, enabling robot-agnostic traversability estimation.

Unsupervised Learning of Neuro-symbolic Rules for Generalizable Context-aware Planning in Object Arrangement Tasks

Siddhant Sharma, Rohan Paul

Reinforcement Learning

🎯 What it does: Developed a method called RLAP that can learn general, sparse, context-aware action primitive rules from human demonstrations without explicit rule label supervision, and generate goal-achieving plans in complex Sokoban-style object arrangement tasks.

Unsupervised Spike Depth Estimation via Cross-modality Cross-domain Knowledge Transfer

Jiaming Liu, Shanghang Zhang

Depth EstimationDomain AdaptationKnowledge DistillationImageMultimodalityBiomedical Data

🎯 What it does: Proposed an unsupervised spike depth estimation framework called BiCross, which achieves spike depth estimation using open-source RGB data through cross-modal cross-domain knowledge transfer.

Untethered Bimodal Robotic Fish with Tunable Bistability

Xu Chao, Xingjian Jing

Robotic Intelligence

🎯 What it does: Developed an untethered robotic fish with adjustable bistability, capable of switching between monostable and bistable motion modes, and achieving different movement behaviors by adjusting the energy barrier.

Untethered Underwater Soft Robot with Thrust Vectoring

Robin Hall, C. Onal

Robotic Intelligence

🎯 What it does: Introduced and designed the DRAGON flexible deformable thrust vectorization underwater soft robot, constructed and conducted untethered tests in multiple environments

Uplifting Range-View-based 3D Semantic Segmentation in Real-Time with Multi-Sensor Fusion

Shiqi Tan, Bingbing Liu

SegmentationAutonomous DrivingKnowledge DistillationImageMultimodalityPoint Cloud

🎯 What it does: Proposes a new LiDAR and Camera Range-view-based 3D point cloud semantic segmentation method called LaCRange

Usability Evaluation Framework for Close-Proximity Collaboration With Large Industrial Manipulators

Kasper Hald, Matthias Rehm

Robotic Intelligence

🎯 What it does: Designed and implemented multiple standardized questionnaires to evaluate close collaboration systems between industrial robots and humans in the DrapeBot project.

Using Large Language Models to Generate and Apply Contingency Handling Procedures in Collaborative Assembly Applications

J. Kang, Satyandra K. Gupta

Robotic IntelligenceLarge Language Model

🎯 What it does: Using LLM to generate and update tasks in HTN to handle uncertainties in collaborative assembly.

Using Specularities to Boost Non-Rigid Structure-from-Motion

A. Sengupta, Adrien Bartoli

OptimizationImageBiomedical Data

🎯 What it does: Improving the reconstruction accuracy of non-rigid structure from motion (NRSfM) by detecting specular highlights and utilizing sparse surface normals

Utilizing a Malfunctioning 3D Printer by Modeling Its Dynamics with Machine Learning

Renzo Caballero, Jürgen Schmidhuber

RestorationRobotic IntelligenceWorld Model

🎯 What it does: Developed a method utilizing machine learning models to model the dynamics of failed 3D printers, enabling them to operate reliably even after damage and achieving self-repair.

Utilizing Inpainting for Training Keypoint Detection Algorithms Towards Markerless Visual Servoing

Sreejani Chatterjee, B. Çalli

RestorationPose EstimationRobotic IntelligenceImage

🎯 What it does: Proposed a method utilizing image inpainting technology to remove ArUco markers, generating unmarked keypoint detection data during the training phase, and applying it to real-time detection of natural features in robots and visual servo control.

V-STRONG: Visual Self-Supervised Traversability Learning for Off-road Navigation

Sanghun Jung, Alexander Lambert

ClassificationSegmentationAutonomous DrivingTransformerContrastive LearningImage

🎯 What it does: Proposes a vision-based self-supervised traversability learning method to improve terrain traversability prediction for outdoor navigation.

V2CE: Video to Continuous Events Simulator

Zhongyang Zhang, Tauhidur Rahman

GenerationData SynthesisVideo

🎯 What it does: Proposed a video-to-event stream conversion method that considers DVS characteristics from multiple perspectives, designed a series of loss functions to enhance the quality of generated event voxels, and introduced a local dynamic perception timestamp inference strategy to achieve continuous timestamp recovery and eliminate temporal layering issues.

VAPOR: Legged Robot Navigation in Unstructured Outdoor Environments using Offline Reinforcement Learning

Kasun Weerakoon, Dinesh Manocha

Robotic IntelligenceReinforcement LearningPoint Cloud

🎯 What it does: Proposed a navigation method for legged robots in unstructured outdoor vegetation environments called VAPOR, based on offline reinforcement learning;

Vascular Centerline-Guided Autonomous Navigation Methods for Robot-Lead Endovascular Interventions

Naner Li, Han Ding

Robotic IntelligenceBiomedical Data

🎯 What it does: Studied autonomous catheter guidance methods for robots, proposing a trial-and-error approach and a centerline guidance strategy, and integrating Dynamic Time Warping (DTW) technology in dynamic environments.

VBR: A Vision Benchmark in Rome

Leonardo Brizi, Giorgio Grisetti

Simultaneous Localization and MappingMultimodalityPoint CloudBenchmark

🎯 What it does: Collected a multimodal dataset containing RGB, 3D point clouds, IMU, and GPS data, and proposed a benchmark for visual odometry and SLAM based on this dataset.

VDNA-PR: Using General Dataset Representations for Robust Sequential Visual Place Recognition

Benjamin Ramtoula, Paul Newman

RecognitionComputational EfficiencyRepresentation LearningImageSequential

🎯 What it does: Adopt the general dataset representation technique VDNA to learn a lightweight encoder that generates robust descriptors for visual pose recognition.

Vehicle Behavior Prediction by Episodic-Memory Implanted NDT

Peining Shen, Jianru Xue

Autonomous DrivingExplainability and Interpretability

🎯 What it does: By constructing a neural decision tree eMem-NDT based on text embedding hierarchical clustering, replacing the softmax layer of a pre-trained deep learning model, and utilizing memory prototype matching and leaf node link aggregation to achieve interpretable prediction of vehicle behavior.

Vehicle Intention Classification Using Visual Clues

Marvin Klemp, Christoph Stiller

ClassificationAutonomous DrivingTransformerImage

🎯 What it does: Proposed VISUAL INTENTION FORMER for visual intention classification of image sequences tracking traffic participants.

VeloVox: A Low-Cost and Accurate 4D Object Detector with Single-Frame Point Cloud of Livox LiDAR

Tao Ma, Hongsheng Li

Object DetectionPoint Cloud

🎯 What it does: Proposed a 4D object detector named VeloVox based on single-frame point clouds from Livox LiDAR, achieving accurate object detection and speed estimation.

Verifiable Learned Behaviors via Motion Primitive Composition: Applications to Scooping of Granular Media

A. Benton, Prithvi Akella

Robotic IntelligenceGraph Neural NetworkText

🎯 What it does: A framework is constructed to enable real-time verification of robot behaviors generated through natural language input, generating behaviors by synthesizing motion primitives into a directed graph; verified in simulation exploration tasks and hardware granular material grasping tasks.

VERSE: Virtual-Gradient Aware Streaming Lifelong Learning with Anytime Inference

S. Banerjee, Piyush Rai

🎯 What it does: Proposes a streaming lifelong learning method based on virtual gradients, supporting single-pass training, class-incremental learning, and on-the-fly inference.

Vertical Vibratory Transport of Grasped Parts Using Impacts

C. Yako, John Kenneth Salisbury

Robotic IntelligencePhysics Related

🎯 What it does: Utilize impact-induced acceleration and periodic stick-slip to achieve vertical upward transportation of the gripped component, along with corresponding theoretical analysis and design guidelines.