IROS 2023 Papers — Page 4
IEEE/RSJ International Conference on Intelligent Robots and Systems · 1195 papers
Detecting Olives with Synthetic or Real Data? Olive the Above
Yianni Karabatis, Y. Aloimonos
Object DetectionData SynthesisImageAgriculture Related
🎯 What it does: Proposed an olive detection method that does not require manual annotation and constructed the first olive detection dataset containing synthetic and real images.
Determination of the Characteristics of Gears of Robot-Like Systems by Analytical Description of their Structure
S. Landler, Karsten Stahl
Robotic Intelligence
🎯 What it does: Unified symbolic analysis of gear structures in robots and similar systems to solve gear characteristics such as backlash, efficiency, and stiffness.
Development and Evaluation of a Single-arm Robotic System for Autonomous Suturing
Jiawei Liu, Axel Krieger
Robotic IntelligenceBiomedical Data
🎯 What it does: Developed and evaluated a single-arm robot system using a novel suturing management device (SMD) and controller to achieve single-arm suturing management in autonomous suturing with the STAR robot
Development and Evaluation of Exploratory Experiences to Facilitate Reasoning About Robotic Systems
Sogol Balali, Cindy Grimm
Robotic Intelligence
🎯 What it does: Proposed and evaluated an interactive method called Exploratory Experiences to enhance people's reasoning about the capabilities and limitations of robot navigation and object detection.
Development and Online Validation of an Intrinsic Fault Detector for a Powered Robotic Knee Prosthesis
Amir R. Naseri, H. Huang
Anomaly DetectionRobotic IntelligenceTime SeriesBiomedical Data
🎯 What it does: Designed and demonstrated in real-time an active fault detector for a robotic knee prosthesis.
Development of A Dynamic Quadruped with Tunable, Compliant Legs
Fuchen Chen, Daniel M. Aukes
🎯 What it does: Developed a 400g quadruped robot equipped with adjustable stiffness laminated elastic legs and conducted related experiments
Development of a Five-Fingerd Biomimetic Soft Robotic Hand by 3D Printing the Skin and Skeleton as One Unit
K. Miyama, Masayuki Inaba
Robotic Intelligence
🎯 What it does: Proposed an integrated skin-skeleton robotic hand with 15 degrees of freedom, consisting of four parts primarily made of 3D-printed single flexible components, capable of achieving thumb adduction, flexion, opposition, and flexion of the four fingers.
Development of a Whole-Body Work Imitation Learning System by a Biped and Bi-Armed Humanoid
Yutaro Matsuura, M. Inaba
OptimizationRobotic Intelligence
🎯 What it does: Developed a full-body imitation learning system using the biped omnidirectional robot JAXON, combined with the TABLIS full-body control device to achieve bidirectional control of dual arms and legs, and implemented long-term data collection and high-load control through posture optimization methods
Development of an Autonomous Modular Swimming Robot with Disturbance Rejection and Path Tracking
Hankun Deng, Bo Cheng
Robotic Intelligence
🎯 What it does: Developed a self-contained modular swimming robot named µBot 2.0 and experimentally validated the swimming performance of its three-axis electromagnetic drive unit.
Development of the Whole-Body Waterproof Shell Applying and Removing System Using Phase-Change Paraffin and Grease for the Multi-DOF Robot
Tasuku Makabe, Masayuki Inaba
Robotic Intelligence
🎯 What it does: A system was proposed and implemented to form an addable/removable full-body waterproof shell on the surface of a multi-degree-of-freedom robot using phase-change paraffin and oils, with the addition and removal of the shell layer achieved through temperature control; the system uses a small spider-shaped multi-legged robot equipped with temperature sensing as the protected object, cooperating with a full-scale robotic arm to apply and remove the waterproof shell, achieving dual applicability in terrestrial and underwater environments;
DexRepNet: Learning Dexterous Robotic Grasping Network with Geometric and Spatial Hand-Object Representations
Qingtao Liu, Qi Ye
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose DexRep geometry and spatial hand-object interaction representation along with the DexRepNet network for learning high-degree-of-freedom robot grasping strategies
Dexterous Soft Hands Linearize Feedback-Control for In-Hand Manipulation
Adrian Sieler, O. Brock
Robotic Intelligence
🎯 What it does: Proposes a linear feedback control framework based on the deformation state of a soft hand for quickly learning and executing intrinsic grasping skills in real environments.
DiffClothAI: Differentiable Cloth Simulation with Intersection-free Frictional Contact and Differentiable Two-Way Coupling with Articulated Rigid Bodies
Siheng Zhao, Lin Shao
Robotic IntelligencePhysics Related
🎯 What it does: Propose DiffClothAI, integrating differentiable cloth simulation with collision-free frictional contact, and achieving bidirectional coupling with joint rigid bodies; evaluate on robot cloth manipulation tasks.
Differentiable Task Assignment and Motion Planning
Jimmy Envall, Stelian Coros
Autonomous DrivingOptimization
🎯 What it does: Propose a fully differentiable task and motion planning method that avoids explicitly enumerating task primitives and supports multi-instance tasks.
DiffuPose: Monocular 3D Human Pose Estimation via Denoising Diffusion Probabilistic Model
Jeongjun Choi, H. Kim
Pose EstimationGraph Neural NetworkDiffusion modelImage
🎯 What it does: This paper proposes a method based on diffusion probability models, generating multiple 3D pose candidates to resolve depth ambiguity when mapping single-frame 2D key points to 3D poses, directly reconstructing 3D human poses from 2D key points detected in a single image.
Directed Real-World Learned Exploration
Matthias Hutsebaut-Buysse, Tom De Schepper
Autonomous DrivingReinforcement LearningWorld ModelImage
🎯 What it does: This paper proposes a learning-based guided exploration method for autonomous vehicles to complete tasks in unseen environments, guided by the uncertainty of the task module; using the warehouse inventory task as an example, it demonstrates how active data collection can improve task performance.
Discovering Adaptable Symbolic Algorithms from Scratch
Stephen Kelly, Esteban Real
Robotic Intelligence
🎯 What it does: Proposed and demonstrated AutoRobotics-Zero (ARZ), a symbol algorithm based on AutoML-Zero that discovers adaptable algorithms from scratch, capable of automatically adjusting parameters and reasoning algorithms in sudden environmental changes to achieve safe robot control.
Discrete-Time Adaptive Control Algorithm for Coordination of Multiagent Systems in the Presence of Coupled Dynamics
I. A. Aly, K. Dogan
Optimization
🎯 What it does: Proposed a discrete-time adaptive control architecture for coupled dynamics in uncertain scalar multi-agent systems.
Disentangled Discriminator for Unsupervised Domain Adaptation on Object Detection
Yangguang Zhu, Xiangbin Wu
Object DetectionDomain AdaptationKnowledge DistillationGenerative Adversarial NetworkImage
🎯 What it does: Proposed a disentangled discriminator for unsupervised domain adaptation (UDA) in object detection, combined with a teacher-student framework to achieve self-training.
DisPlacing Objects: Improving Dynamic Vehicle Detection via Visual Place Recognition under Adverse Conditions
Stephen Hausler, Michael Milford
Object DetectionRetrievalAutonomous Driving
🎯 What it does: Propose a method that utilizes prior maps through visual localization and a binary classification network to refine vehicle detection results.
Distributed Model Predictive Formation Control of Robots with Sampled Trajectory Sharing in Cluttered Environments
Sami Satir, Erol Şahin
OptimizationRobotic Intelligence
🎯 What it does: Proposed a distributed formation control method based on model predictive control (MPC), and verified its feasibility through simulations and experiments on Crazyflie small quadrotors.
Disturbance Preview for Non-Linear Model Predictive Trajectory Tracking of Underwater Vehicles in Wave Dominated Environments
Kyle L. Walker, Francesco Giorgio-Serchi
OptimizationRobotic IntelligenceTime SeriesPhysics Related
🎯 What it does: Propose a nonlinear model predictive control (NMPC) method combined with a deterministic wave predictor for trajectory tracking of underwater vehicles in marine environments significantly affected by waves.
DMCL: Robot Autonomous Navigation via Depth Image Masked Contrastive Learning
Jiahao Jiang, Yuxiang Yang
Robotic IntelligenceReinforcement LearningContrastive LearningImage
🎯 What it does: Proposed an end-to-end robotic visual navigation method called DMCL, which utilizes Depth Image Masked Contrastive Learning to obtain spatial-temporal state representations and combines it with Soft Actor-Critic reinforcement learning for autonomous navigation.
Do Hierarchies in a Robot Team Impact the Service Evaluation by Users?
Soomin Shin, Sonya S. Kwak
Robotic IntelligenceText
🎯 What it does: Studied the impact of robot team hierarchy on user satisfaction by setting up four different relationships using the Korean honorific system and collecting user feedback.
Do More with Less: Single-Model, Multi-Goal Architectures for Resource-Constrained Robots
Zili Wang, Roberto Tron
Computational EfficiencyRobotic IntelligenceConvolutional Neural NetworkSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: Designed and implemented a single multi-objective deep learning architecture that leverages environmental structure to predict different abstractions, achieving multi-task objectives (e.g., object search, topological map building), thereby reducing sensor and computational costs in resource-constrained robots.
Dodging Like A Bird: An Inverted Dive Maneuver Taking by Lifting-Wing Multicopters
Wenhan Gao, Quan Quan
Robotic IntelligenceOrdinary Differential Equation
🎯 What it does: Proposes a rotation-aware, collision-free trajectory planning method for lift-wing multirotors, supporting large angles and even inverted flight.
Does Unpredictability Influence Driving Behavior?
S. Samavi, Angela P. Schoellig
Autonomous DrivingReinforcement Learning
🎯 What it does: Study the impact of surrounding vehicles' unpredictability on the self-vehicle's lane-changing behavior on highways, using maximum entropy inverse reinforcement learning to construct a reward function and introducing features based on unpredictability
Domain Adaptation on Point Clouds for 6D Pose Estimation in Bin-Picking Scenarios
Liang Zhao, Long Zeng
Pose EstimationDomain AdaptationGenerative Adversarial NetworkPoint Cloud
🎯 What it does: Proposes DAPE-Net, a 6D pose estimation network based on point clouds, and achieves domain adaptation in the Bin-Picking scenario.
Domain Randomization for Robust, Affordable and Effective Closed-Loop Control of Soft Robots
Gabriele Tiboni, Giuseppe Averta
Robotic IntelligenceReinforcement Learning
🎯 What it does: Enhance the robustness of soft robotic reinforcement learning (RL) strategies, reduce training time, and improve environmental exploration through domain randomization; first introduce an algorithm extension for automatic inference of dynamic parameters of deformable objects; conduct extensive evaluations on four tasks and two soft robot designs in simulation.
Domains as Objectives: Multi-Domain Reinforcement Learning with Convex-Coverage Set Learning for Domain Uncertainty Awareness
Wendyam Eric Lionel Ilboudo, Takamitsu Matsubara
Domain AdaptationRobotic IntelligenceReinforcement Learning
🎯 What it does: Investigated a domain uncertainty-aware universal policy based on multi-objective reinforcement learning to address the conservative policies caused by domain randomization, and validated its effectiveness in Mujoco simulations and on the D'Claw robot.
DORMADL - Dataset of Human-Operated Robot Arm Motion in Activities of Daily Living
F. Goldau, Udo Frese
Data SynthesisRobotic IntelligenceSequentialBenchmark
🎯 What it does: Proposed a high-resolution human-operated robotic arm action dataset for assistive robotics, covering tasks in everyday life.
DoubleBee: A Hybrid Aerial-Ground Robot with Two Active Wheels
Muqing Cao, Lihua Xie
Robotic Intelligence
🎯 What it does: Proposed and implemented a dual-mode dual-bee robot called DoubleBee, integrating two-rotor tilting servos and two drive wheels, achieving dynamic modeling and control in both aerial and ground modes.
DR-Pose: A Two-Stage Deformation-and-Registration Pipeline for Category-Level 6D Object Pose Estimation
Lei Zhou, Marcelo H ANG Jr
Pose EstimationPoint CloudBenchmark
🎯 What it does: Proposes a two-stage deformation and registration pipeline, DR-Pose, for category-level 6D object pose estimation.
Driver Distraction Detection for Daytime and Nighttime with Unpaired Visible and Infrared Image Translation
Hong-Ze Shen, Huei-Yung Lin
Image TranslationPose EstimationAutonomous DrivingImage
🎯 What it does: Propose a network model named V2IA-Net that utilizes daytime visible light and nighttime infrared images for driver distraction detection, combining driver action recognition and head pose detection to achieve real-time analysis.
DRKF: Distilled Rotated Kernel Fusion for Efficient Rotation Invariant Descriptors in Local Feature Matching
Ranran Huang, Z. Chai
RetrievalComputational EfficiencyKnowledge DistillationConvolutional Neural NetworkImage
🎯 What it does: Propose rotational convolution kernel fusion (RKF) and multi-directional feature aggregation (MOFA) methods to efficiently learn rotation-invariant local feature descriptors.
DroNeRF: Real-Time Multi-Agent Drone Pose Optimization for Computing Neural Radiance Fields
Dipam Patel, Aniket Bera
Pose EstimationOptimizationNeural Radiance FieldImage
🎯 What it does: Propose the DroNeRF algorithm to achieve autonomous localization of monocular camera drones around objects, enabling real-time 3D reconstruction with only a few images.
DS-MPEPC: Safe and Deadlock-Avoiding Robot Navigation in Cluttered Dynamic Scenes
Senthil Hariharan Arul, Dinesh Manocha
OptimizationRobotic Intelligence
🎯 What it does: Proposed the DS-MPEPC algorithm for safe and deadlock-free robot navigation in complex dynamic environments, employing a variant of model predictive equilibrium point control and achieving smooth navigation through optimization of the trajectory cost function at each time step
Dual Variable Actor-Critic for Adaptive Safe Reinforcement Learning
Junseo Lee, Songhwai Oh
Reinforcement Learning
🎯 What it does: Propose a dual-variable actor-critic framework that trains general policies and general Q-functions capable of adapting to different safety levels, and ensures convergence to the Pareto optimal strategy set by extending the soft actor-critic method.
DualCross: Cross-Modality Cross-Domain Adaptation for Monocular BEV Perception
Yunze Man, Yu-Xiong Wang
Domain AdaptationAutonomous DrivingImagePoint Cloud
🎯 What it does: Proposed the DualCross framework to achieve cross-modal and cross-domain training of monocular bird's-eye-view perception models, transferring LiDAR knowledge to different domains with only camera test scenarios.
DVL-Based Odometry for Autonomous Underwater Gliders
G. Billings, Richard Camilli
Pose EstimationSimultaneous Localization and MappingTime Series
🎯 What it does: Proposes a method that utilizes a Doppler Velocity Log (DVL) to estimate the speed of an autonomous underwater vehicle (AUV) in real-time and dynamically profile the current velocity in the water column, thereby improving positioning accuracy.
Dynamic Decision Frequency with Continuous Options
Amir-Hossein Karimi, Samuele Tosatto
Reinforcement Learning
🎯 What it does: Proposed the Continuous-Time Continuous Options (CTCO) framework to dynamically adjust the time frequency of decision-making in reinforcement learning.
Dynamic Finger Gaits via Pivoting and Adapting Contact Forces
Yuechuan Xue, Yan-Bin Jia
Robotic Intelligence
🎯 What it does: Proposes a single-finger gait method that uses the tool tip as a fulcrum and dynamically adjusts contact force in three stages, demonstrated on tools and screwdrivers.
Dynamic Hand Proprioception via a Wearable Glove with Fabric Sensors
Lily Behnke, Rebecca Kramer‐Bottiglio
Pose EstimationTime Series
🎯 What it does: Designed, manufactured, and characterized a thin, breathable fabric glove for reconstructing joint angles from fine hand movements;
Dynamic Heart Simulator for Ultrasound-Guided Pericardiocentesis
Kim Yan, S. Cheng
ImageUltrasound
🎯 What it does: Developed a Dynamic Heart Simulator (DHS) capable of simulating cardiac pulsations, pericardial effusion, and generating realistic ultrasound images under ultrasound imaging for training in pericardial puncture.
Dynamic Hybrid Locomotion and Jumping for Wheeled-Legged Quadrupeds
M. Hosseini, Sven Behnke
OptimizationRobotic Intelligence
🎯 What it does: Propose a motion optimization framework that enables four-legged robots with non-steerable wheels to perform dynamic jumps, achieve hybrid wheel-leg locomotion, and overcome obstacles without deceleration.
Dynamic Modeling and Analysis of Impact-Resilient MAVs Undergoing High-Speed and Large-Angle Collisions with the Environment
Zhichao Liu, Konstantinos Karydis
Physics Related
🎯 What it does: Studied the impact recovery capability of deformable micro aerial vehicles (MAVs) equipped with passive springs during high-speed and large-angle collisions, and analyzed the effects of flexible structures through dynamic modeling.
Dynamic Multi-Query Motion Planning with Differential Constraints and Moving Goals
Michael Gentner, Eckehard G. Steinbach
OptimizationRobotic IntelligenceOrdinary Differential Equation
🎯 What it does: Propose a robot motion planning method based on third-order differential constraints, applicable to mobile targets in dynamic environments;
Dynamic Multi-Target Tracking Using Heterogeneous Coverage Control
Ruoyu Lin, Magnus Egerstedt
Object TrackingRobotic Intelligence
🎯 What it does: A coverage-based collaborative control strategy is developed for multi-robot systems to simultaneously estimate and track the states of multiple targets governed by stochastic dynamics within heterogeneous effective perception ranges and safe operational areas.
Dynamic Object Tracking for Quadruped Manipulator with Spherical Image-Based Approach
Tianlin Zhang, Y. Lou
Object TrackingRobotic IntelligenceImage
🎯 What it does: Proposes an image-based visual servoing (IBVS) method for quadrupedal robotic arms to accurately estimate and track the motion of surrounding dynamic objects using only an RGB camera; the method's effectiveness is verified through hardware experiments.
DynGMP: Graph Neural Network-Based Motion Planning in Unpredictable Dynamic Environments
Wenjin Zhang, Bo Yuan
Autonomous DrivingGraph Neural Network
🎯 What it does: Proposes DynGMP, a motion planner based on graph neural networks (GNN), designed for unpredictable dynamic environments;
EARL: Eye-on-Hand Reinforcement Learner for Dynamic Grasping with Active Pose Estimation
Baichuan Huang, Siddarth Jain
Pose EstimationRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a system called EARL that achieves pose tracking and dynamic grasping of moving objects under real-time vision;
Early or Late Fusion Matters: Efficient RGB-D Fusion in Vision Transformers for 3D Object Recognition
Georgios Tziafas, H. Kasaei
RecognitionDomain AdaptationTransformerImageMultimodalityBenchmark
🎯 What it does: Propose a simple and effective approach to transfer pre-trained ViT to the RGB-D domain for 3D object recognition, focusing on joint encoding and fusion of RGB and depth representations.
EasyGaze3D: Towards Effective and Flexible 3D Gaze Estimation from a Single RGB Camera
Jinkai Li, Yao Guo
Pose EstimationImage
🎯 What it does: Proposes the EasyGaze3D framework and the Easy-Cali calibration module, achieving 3D gaze estimation using a single RGB camera
ECTLO: Effective Continuous-Time Odometry Using Range Image for LiDAR with Small FoV
Xin Zheng, Jianke Zhu
Pose EstimationAutonomous DrivingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes an effective continuous-time LiDAR odometry (ECTLO) method for Risley prism-based LiDAR.
EDI: ESKF-based Disjoint Initialization for Visual-Inertial SLAM Systems
Weihan Wang, Philippos Mordohai
Simultaneous Localization and MappingMultimodality
🎯 What it does: Propose a split visual-inertial SLAM initialization method called EDI based on the error-state Kalman filter (ESKF) to achieve fast, accurate, and robust visual-inertial initialization.
EELS: Towards Autonomous Mobility in Extreme Terrain with a Versatile Snake Robot with Resilience to Exteroception Failures
Rohan Thakker, M. Ono
Autonomous DrivingRobotic Intelligence
🎯 What it does: Proposed a 4-meter-long snake robot EELS with a peak torque of 400Nm, and designed a scalable adaptive gait generation architecture NEO with multiple degrees of freedom, completing hardware evaluation and field deployment in indoor icy environments and extreme sandy/icy scenarios.
Effective Traffic Signal Control with Offline-to-Online Reinforcement Learning
Jinming Ma, Feng Wu
Autonomous DrivingOptimizationReinforcement Learning
🎯 What it does: Proposes an offline-to-online reinforcement learning framework, pre-training traffic signal control models using enhanced pre-collected data in the offline phase, and rapidly adapting to new traffic scenarios in the online phase through a difference metric.
Effectively Rearranging Heterogeneous Objects on Cluttered Tabletops
Kai Gao, Jingjin Yu
OptimizationRobotic Intelligence
🎯 What it does: Studied long-term sequence rearrangement of heterogeneous objects in a desktop environment and proposed a solver capable of generating near-optimal plans.
Effects of Personalization on Gait-State Tracking Performance Using Extended Kalman Filters
José A. Montes-Pérez, Robert D. Gregg
Pose EstimationTime SeriesBiomedical Data
🎯 What it does: Quantified the error reduction in EKF gait state tracking using a personalized measurement model compared to a unified measurement model.
Efficiency Estimation and Optimization of Multistage Compound Planetary Gearboxes and Application to the Design of the Active Skin Propulsion of EELS
Nikola-Zlatkov Georgiev
OptimizationPhysics Related
🎯 What it does: Propose a geometric method based on virtual tangent point modeling to estimate the efficiency of multi-stage compound planetary gearboxes, and apply this method to the design of a three-stage compound planetary gearbox to support the active skin thruster of EELS.
Efficient and Feasible Robotic Assembly Sequence Planning via Graph Representation Learning
Matan Atad, Rudolph Triebel
Representation LearningRobotic IntelligenceGraph Neural NetworkGraph
🎯 What it does: Proposed a holistic graph method that includes Assembly Graph representation and Graph Assembly Processing Network (GRACE) for robot assembly sequence planning.
Efficient Constrained Multi-Agent Trajectory Optimization Using Dynamic Potential Games
Maulik Bhatt, Negar Mehr
Optimization
🎯 What it does: Developed an efficient and fast multi-agent constrained trajectory optimizer, utilizing a dynamic potential game framework to address constrained interactive path planning problems.
Efficient Deep Learning of Robust, Adaptive Policies using Tube MPC-Guided Data Augmentation
Tongyu Zhao, J. How
OptimizationComputational EfficiencyRepresentation LearningRobotic IntelligenceReinforcement Learning
🎯 What it does: An IL algorithm based on MPC that learns robust adaptive strategies to adapt to model/environment uncertainties
Efficient Domain Coverage for Vehicles with Second-Order Dynamics via Multi-Agent Reinforcement Learning
Xinyu Zhao, Mo Chen
Autonomous DrivingOptimizationRecurrent Neural NetworkTransformerReinforcement Learning
🎯 What it does: This paper studies the multi-agent coverage problem combining reinforcement learning with control methods, considering vehicles with second-order dynamics.
Efficient Exploration Using Extra Safety Budget in Constrained Policy Optimization
Haotian Xu, T. Zhang
OptimizationReinforcement LearningBenchmark
🎯 What it does: Propose an algorithm named ESB-CPO that achieves efficient exploration in constrained policy optimization by utilizing an additional safety budget;
Efficient Heuristics for Multi-Robot Path Planning in Crowded Environments
Teng Guo, Jingjin Yu
OptimizationRobotic Intelligence
🎯 What it does: Proposed and implemented two hybrid algorithms, DCBS and SCBS, for multi-robot path planning in dense scenarios.
Efficient Object Manipulation Planning with Monte Carlo Tree Search
Huaijiang Zhu, L. Righetti
OptimizationRobotic Intelligence
🎯 What it does: Propose a method for efficient object manipulation planning using Monte Carlo tree search (MCTS) and ADMM trajectory optimization
Efficient Path Planning In Manipulation Planning Problems by Actively Reusing Validation Effort
V. Hartmann, Marc Toussaint
OptimizationRobotic Intelligence
🎯 What it does: In repetitive manipulation planning problems, decompose collision detection into reusable and non-reusable parts, and treat multi-path planning problems as multi-query path planning to achieve active reuse of prior knowledge.
Efficient Q-Learning over Visit Frequency Maps for Multi-Agent Exploration of Unknown Environments
Xuyang Chen, A. H. Qureshi
Computational EfficiencyRepresentation LearningReinforcement LearningAgentic AI
🎯 What it does: Proposed the Integrated Visit Frequency Map, as well as an access frequency benchmark scheme supporting multi-agent information exchange and control;
Efficient Symbolic Approaches for Quantitative Reactive Synthesis with Finite Tasks
Karan Muvvala, Morteza Lahijanian
Computational EfficiencyRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposed an efficient symbolic algorithm for quantitative reaction synthesis in limited tasks, applied to complex tasks involving human-robot interaction in resource-constrained robotic manipulators.
Efficient Visual Perception of Human-Robot Walking Environments using Semi-Supervised Learning
Dmytro Kuzmenko, Brokoslaw Laschowski
ClassificationRobotic IntelligenceTransformerImage
🎯 What it does: Proposed and implemented a semi-supervised learning system ExoNet-SSL based on a mobile vision transformer to enhance visual perception in human-robot walking environments.
Efficient Visuo-Haptic Object Shape Completion for Robot Manipulation
Lukas Rustler, M. Hoffmann
Robotic IntelligenceVision-Language-Action ModelImageMultimodality
🎯 What it does: Achieving robotic object shape completion through closed-loop visual and tactile reconstruction.
Ego-Noise Reduction of a Mobile Robot Using Noise Spatial Covariance Matrix Learning and Minimum Variance Distortionless Response
Pierre-Olivier Lagacé, François Grondin
Robotic IntelligenceAudio
🎯 What it does: An adaptive robot self-noise suppression method is proposed using a microphone array and less than two minutes of noise recordings, leveraging PCA to select the optimal covariance matrix and combining it with the MVDR beamformer.
ElC-OIS: Ellipsoidal Clustering for Open-World Instance Segmentation on LiDAR Data
Wenbang Deng, Xieyuanli Chen
SegmentationDiffusion modelPoint Cloud
🎯 What it does: Propose an open-world instance segmentation framework suitable for LiDAR point clouds that can accurately segment both known and unknown instances.
Emergent Cooperative Behavior in Distributed Target Tracking with Unknown Occlusions
Tianqi Li, Swaminathan Gopalswamy
Object TrackingRobotic Intelligence
🎯 What it does: Extended the active perception framework for multi-robot systems in distributed target tracking, explicitly detecting, maintaining, and sharing occlusion information to achieve occlusion-aware planning without prior semantic occlusion knowledge, and analyzed three collaborative behavior patterns emerging from the same algorithm.
Emergent Sequential Motion Through Compliant Auxetic Shells
Audrey Sedal, S. Kota
Physics Related
🎯 What it does: Proposes a sequential motion design method for soft fluidic robots based on Bessel curve beams and representative auxetic elements (RAE), validated by two prototypes.
Emotionally Specific Backchanneling in Social Human-Robot Interaction and Human-Human Interaction
Pourya Shahverdi, W. Louie
Robotic Intelligence
🎯 What it does: An exploratory study on behind-channel behaviors of humans and social robots under different emotional contexts, comparing emotion-specific behind-channel differences in human-human and human-robot interactions
Employing Multi-Layer, Sensorised Kirigami Grippers for Single-Grasp Based Identification of Objects and Force Exertion Estimation
Junbang Liang, Minas V. Liarokapis
Classification
🎯 What it does: Using a multi-layer sensorized Kirigami gripper to perform single-grasp object classification and grasp force estimation.
End-to-End Learning of Behavioural Inputs for Autonomous Driving in Dense Traffic
Jatan Shrestha, A. K. Singh
Autonomous DrivingOptimization
🎯 What it does: Propose an end-to-end framework for learning the distribution of behavioral inputs, used for trajectory planning in autonomous driving within congested traffic.
End-to-End Learning of Deep Visuomotor Policy for Needle Picking
Hongbin Lin, K. W. S. Au
Robotic IntelligenceReinforcement LearningAuto EncoderWorld ModelImage
🎯 What it does: Developed an end-to-end deep visual-motor policy learning framework called DreamerfD for needle grasping tasks in robotic surgery.
End-to-End Point Cloud Registration via Rotation Equivariant Descriptors
Yue Cao, Hongdong Li
Pose EstimationPoint Cloud
🎯 What it does: Proposed an end-to-end point cloud registration method that utilizes rotation-invariant and rotation-equivariant descriptors to achieve keypoint matching and transformation recovery.
End-to-End Reinforcement Learning for Torque Based Variable Height Hopping
Raghav Soni, F. Kirchner
Domain AdaptationRobotic IntelligenceReinforcement Learning
🎯 What it does: This paper proposes an end-to-end reinforcement learning-driven torque controller that can implicitly detect the relevant phases of jumping, eliminating the need for manual state detection, and successfully migrates it to real robots without parameter tuning; meanwhile, it extends the sim-to-real transfer method from simulation to reality to handle contact-rich dynamic tasks.
Energy Constrained Multi-Agent Reinforcement Learning for Coverage Path Planning
Chenyang Zhao, Zhentong Zhang
OptimizationRobotic IntelligenceReinforcement Learning
🎯 What it does: Decompose the multi-agent area coverage path planning problem into two subproblems: multi-agent path planning and sub-area coverage path planning, and propose an energy-constrained multi-agent reinforcement learning algorithm (ECMARL) for online dynamic strategy adjustment.
Energy-Aware Planning of Heterogeneous Multi-Agent Systems for Serving Cooperative Tasks with Temporal Logic Specifications
Ali Tevfik Buyukkocak, Yasin Yazıcıoğlu
Optimization
🎯 What it does: For energy-constrained heterogeneous multi-agent teams, a high-level path is designed using sampling-based graph abstraction and mixed-integer programming, achieving collaborative tasks that satisfy Signal Temporal Logic (STL) spatiotemporal constraints through a stochastic energy model and resupply strategy.
Energy-Efficient Team Orienteering Problem in the Presence of Time-Varying Ocean Currents
Ariella Mansfield, M. Hsieh
OptimizationBenchmarkPhysics Related
🎯 What it does: Proposes a multi-objective formula to balance visiting multiple task locations and energy consumption under time budget constraints, specifically for autonomous underwater vehicles in time-varying current environments;
Enhance Local Feature Consistency with Structure Similarity Loss for 3D Semantic Segmentation
Cheng-Wei Lin, Kuan-Wen Chen
SegmentationPoint Cloud
🎯 What it does: Propose a new loss function called Linearity and Planarity to enhance local feature consistency in regions with similar structures
Enhanced Performance of Human-Robot Collaboration Using Braking Surfaces and Trajectory Scaling
Bakir Lacevic, Paolo Rocco
Safty and PrivacyRobotic Intelligence
🎯 What it does: Designed a control method based on braking surfaces and trajectory scaling to enhance production cycle speed in human-robot collaboration scenarios while satisfying the speed and separation monitoring principles of the latest safety standards.
Enhanced Robot Navigation with Human Geometric Instruction
Hideki Deguchi, Satoshi Koide
Robotic IntelligenceVision-Language-Action Model
🎯 What it does: Proposes a system that utilizes human geometric instructions (e.g., rough target positions indicated by hand gestures) to assist robot navigation. The system adaptively estimates the reliability of geometric instructions and switches between exploration mode and instruction-following mode based on reliability values.
Enhancing 5G-Enabled Robots Autonomy by Radio-Aware Semantic Maps
A. Ibanez, Xavier Pérez Costa
Robotic Intelligence
🎯 What it does: Proposes a framework for constructing a semantic map based on wireless signal quality, enabling mobile robots to obtain real-time wireless environment information and improve connection reliability and operational efficiency.
Enhancing Fine-Grained 3D Object Recognition Using Hybrid Multi-Modal Vision Transformer-CNN Models
Songsong Xiong, H. Kasaei
RecognitionData SynthesisConvolutional Neural NetworkTransformer
🎯 What it does: Proposed a hybrid multimodal approach combining Vision Transformer and Convolutional Neural Networks to enhance fine-grained 3D object recognition performance
Enhancing Robustness of Line Tracking Through Semi-Dense Epipolar Search in Line-Based SLAM
Dong-Uk Seo, Hyun Myung
Depth EstimationSimultaneous Localization and MappingImageVideo
🎯 What it does: Proposes a robust line tracking method for line-based monocular visual-inertial odometry, utilizing semi-dense depth and sparse grids for line matching and reducing false detections;
Enhancing Sample Efficiency and Uncertainty Compensation in Learning-Based Model Predictive Control for Aerial Robots
K. Y. Chee, George Pappas
OptimizationRobotic Intelligence
🎯 What it does: Propose a framework that combines C1 adaptive control with learning-enhanced MPC, providing online dynamic model synthesis and validated on a quadrotor.
Enhancing State Estimation in Robots: A Data-Driven Approach with Differentiable Ensemble Kalman Filters
Xinyu Liu, H. B. Amor
Robotic Intelligence
🎯 What it does: Proposes a novel robotic state estimation framework based on the differentiable ensemble Kalman filter (DEnKF), which implicitly models process noise using stochastic neural networks.
Enhancing Teleoperated Robot Customer Service through Speech Monitoring and Filtering
K. Yamada, Takayuki Kanda
Robotic IntelligenceAudio
🎯 What it does: Proposed a voice monitoring and filtering system supporting remote operation of robotic customer service agents, helping operators provide high-quality service;
Enhancing Value Estimation Policies by Post-Hoc Symmetry Exploitation in Motion Planning Tasks
Yazied A. Hasan, Lydia Tapia
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposes a new method that leverages symmetry during the execution phase of reinforcement learning policies by providing the policy with geometrically transformed observations to generate multiple possible actions and selecting the one with the highest value.
Epistemic Planning for Heterogeneous Robotic Systems
Lauren Bramblett, N. Bezzo
OptimizationRobotic Intelligence
🎯 What it does: Proposes a framework based on cognitive planning, enabling heterogeneous multi-robot systems to achieve information dissemination and task allocation in unreliable communication environments through dynamic cognitive logic and mixed integer programming.
Error-State Kalman Filter Based External Wrench Estimation for MAVs Under a Cascaded Architecture
Yuhan Yin, Hao Fang
Robotic IntelligenceTime Series
🎯 What it does: A two-stage error state Kalman filter-based algorithm for external force estimation in multirotor drones is proposed, which utilizes rotor tachometers, inertial measurement units (IMUs), and motion capture systems to achieve real-time external force estimation.
Estimating 4D Data Associations Towards Spatial-Temporal Mapping of Growing Plants for Agricultural Robots
Luca Lobefaro, C. Stachniss
Robotic IntelligenceSimultaneous Localization and MappingPoint CloudAgriculture Related
🎯 What it does: The study investigates the spatiotemporal correlation of 3D points during plant growth and proposes an association method based on RGB-D SLAM, visual place recognition, and 2D/3D matching.
Estimating Human Comfort Levels in Autonomous Vehicles Based on Vehicular Behaviors and Physiological Signals
Haotian Su, Yunyi Jia
Autonomous Driving
🎯 What it does: Proposed a dynamic model based on vehicle behavior and a comfort estimation method using a Kalman filter to quantify human comfort levels in autonomous vehicles.
Estimating Properties of Solid Particles Inside Container Using Touch Sensing
Xiaofeng Guo, Wenzhen Yuan
Robotic Intelligence
🎯 What it does: Using tactile sensing to estimate the mass, volume, particle size, and shape of solid particles within a container