IROS 2023 Papers — Page 8
IEEE/RSJ International Conference on Intelligent Robots and Systems · 1195 papers
Memory Maps for Video Object Detection and Tracking on UAVs
Benjamin Kiefer, A. Zell
Object DetectionObject TrackingSimultaneous Localization and MappingVideo
🎯 What it does: Propose a metadata-based memory map method for UAV video object detection and tracking.
MENTOR: Multilingual Text Detection Toward Learning by Analogy
Hsin-Ju Lin, Ching-Chun Huang
Object DetectionData SynthesisConvolutional Neural Network
🎯 What it does: Propose a multi-language text detection framework called MENTOR, which can detect and recognize both seen and unseen language regions in scene images without collecting unseen language supervision data or retraining the model.
Method for Robotic Motion Compensation During PET Imaging of Mobile Subjects
Junxiang Wang, Peter Kazanzides
Robotic IntelligenceBiomedical DataPositron Emission Tomography
🎯 What it does: Designed and evaluated a robotic system for compensating head motion in PET imaging, achieving coarse positioning correction and detailed reconstruction correction.
Microrobot Control Method Based on Movement of Field Free Point in Gradient Magnetic Field
Chutian Wang, Lin Feng
Robotic IntelligencePhysics Related
🎯 What it does: Controlling the motion of cordless microrobots by moving free points (FFP) in a gradient magnetic field, with experimental verification of bidirectional motion in a one-dimensional gradient magnetic field system
MIGHTY: Multi-Functional Suction Cup for Object Gripping and Surface Attachment
E. Papadakis, P. Trahanias
Robotic Intelligence
🎯 What it does: A multifunctional, lightweight sensor-enhanced vacuum suction cup was developed, capable of achieving object grasping, surface adhesion, and force and torque measurement among other functions;
MIMIR-UW: A Multipurpose Synthetic Dataset for Underwater Navigation and Inspection
Olaya Álvarez-Tuñón, Erdal Kayacan
SegmentationData SynthesisDepth EstimationRobotic IntelligenceSimultaneous Localization and MappingImageMultimodalityBenchmark
🎯 What it does: Created a multi-purpose synthetic underwater dataset MIMIR-UW for SLAM, depth estimation, and object segmentation, evaluated in pipeline inspection scenarios.
Minimal Path Violation Problem with Application to Fault Tolerant Motion Planning of Manipulators
Aakriti Upadhyay, Chinwe Ekenna
OptimizationRobotic Intelligence
🎯 What it does: Propose the Minimal Path Violation (MPV) method, which helps robots quickly recover paths when components fail or the configuration space changes.
Minimalistic Collective Perception with Imperfect Sensors
K. Chin, Carlo Pinciroli
Robotic Intelligence
🎯 What it does: The study employs a minimization-based collective perception approach to derive a probabilistic framework in the presence of sensor defects, which is validated in scenarios where robots collectively determine the frequency of environmental features.
Minimally Actuated Tiltrotor for Perching and Normal Force Exertion
Dongjae Lee, H. J. Kim
Robotic Intelligence
🎯 What it does: Proposes a hardware design and control method for a minimum-actuated tilt-rotor unmanned aerial vehicle (UAV) with five control degrees of freedom, and verifies its functionality through experiments.
Model-based Adversarial Imitation Learning from Demonstrations and Human Reward
Jie Huang, Guangliang Li
Reinforcement Learning from Human FeedbackReinforcement LearningGenerative Adversarial Network
🎯 What it does: This paper proposes and verifies a model-based adversarial imitation learning framework called MAILDH, which combines GAIL, interactive RL, and model-based RL to learn control strategies from demonstrations and human rewards.
Model-Based Bending Control of Magnetically-Actuated Robotic Endoscopes for Automatic Retroflexion in Confined Spaces
Yichong Sun, Zheng Li
Robotic IntelligenceBiomedical Data
🎯 What it does: A complete kinematic model of the magnetic control robot endoscope is established based on the cosine stratagem theory, and a magnetic bending control scheme combining error feedback PID with model feedback is proposed; meanwhile, a serial path point strategy is designed for automatic retroflexion in compact spaces, keeping the magnetic tip position as close to the midline as possible;
Model-Based Planning and Control for Terrestrial-Aerial Bimodal Vehicles with Passive Wheels
Ruibin Zhang, Fei Gao
Autonomous DrivingOptimizationRobotic Intelligence
🎯 What it does: Developed a model-based planning and control framework for land-air dual-mode vehicles with passive wheels;
Model-Free Grasping with Multi-Suction Cup Grippers for Robotic Bin Picking
Philipp Schillinger, Ngo Anh Vien
OptimizationRobotic IntelligenceConvolutional Neural NetworkImage
🎯 What it does: Proposed a model-free grasp pose prediction method for multi-suction grippers.
Modeling Action Spatiotemporal Relationships Using Graph-Based Class-Level Attention Network for Long-Term Action Detection
Yuankai Wu, Constantin Patsch
Object DetectionGraph Neural NetworkVideo
🎯 What it does: Proposing an action detection method based on graph neural networks and time modeling
Modeling and Workspace Characterization of Continuously Compliant Robotic Legs
Robin Bendfeld, M. I. C. David Remy
Robotic Intelligence
🎯 What it does: This study proposes a new design paradigm that integrates continuous elasticity into leg structures, develops a prototype leg with continuously elastic shanks, and derives and analyzes a nonlinear beam model to predict contact forces. Subsequently, the validity of the model is verified through two sets of static experiments on separated shanks and complete legs, and the impact of model discretization on accuracy is investigated (approximately 10 nodes suffice). Furthermore, the concept of force workspace is introduced, revealing its influence by nonlinear deformation and kinematic nonlinear coupling, and demonstrating that it is constrained by force singularities, leading to an irreversible relationship between force and joint angles.
Modeling, Characterization, and Control of Bacteria-Inspired Bi-Flagellated Mechanism with Tumbling
Zhuonan Hao, M. Jawed
Robotic IntelligencePhysics Related
🎯 What it does: Constructed and studied a macroscopic biflagellate robot, utilizing two right-handed helical rigid flagella rotating in opposite directions to achieve repeatable, controllable rolling motion, and optimized its thrust and torque through parameter space exploration; proposed a rolling control scheme requiring only two control inputs to achieve the target heading angle.
Modular Neural Network Policies for Learning In-Flight Object Catching with a Robot Hand-Arm System
Wenbin Hu, Zhibin Li
Robotic IntelligenceReinforcement Learning
🎯 What it does: Designed and implemented a modular framework enabling robotic arm systems to learn to capture flying objects.
MoEmo Vision Transformer: Integrating Cross-Attention and Movement Vectors in 3D Pose Estimation for HRI Emotion Detection
David C. Jeong, Christopher A. Kitts
Pose EstimationTransformerMixture of ExpertsVideo
🎯 What it does: Developed a MoEmo cross-attention visual Transformer for emotion detection based on 3D human pose estimation, and proposed a full-body video and emotion label dataset.
MOISST: Multimodal Optimization of Implicit Scene for SpatioTemporal Calibration
Quentin Herau, C. Demonceaux
OptimizationNeural Radiance FieldMultimodality
🎯 What it does: A new optimization scheme for neural radiance fields (NeRF) that jointly optimizes spatial and temporal calibration parameters of multi-modal sensors and implicit voxel scene representations, achieving calibration by leveraging radiometric and geometric measurement information.
MOMA-Force: Visual-Force Imitation for Real-World Mobile Manipulation
Taozheng Yang, Tao Kong
Representation LearningRobotic IntelligenceImage
🎯 What it does: Proposed a visual-force imitation-based mobile manipulator method called MOMA-Force for performing various contact-rich manipulation tasks.
Monolithic Microchannels in Miniature Pneumatic Soft Robots for Sequential Motions
D. Fan, Hongqiang Wang
Robotic Intelligence
🎯 What it does: Designed and fabricated a micro-scale multi-channel pneumatic soft robot embedded in a monolithic structure, achieving continuous motion.
Monte-Carlo Tree Search with Prioritized Node Expansion for Multi-Goal Task Planning
K. Pfeiffer, Quang-Cuong Pham
OptimizationRobotic Intelligence
🎯 What it does: Proposes a multi-objective symbolic task planner based on Monte Carlo Tree Search (MCTS), utilizing priority node expansion and action reduction techniques to achieve multi-subgoal planning in real-time robotic systems;
Motion Control and Planning of a Bio-Inspired Aerial Vehicle with an Actively Controlled Abdomen-Like Appendage
Berrin Güney, M. M. Ankaralı
OptimizationRobotic Intelligence
🎯 What it does: Designed a dual-rotor aircraft equipped with an actively controlled abdominal appendage, established a dynamic model, and employed optimization controllers such as LQR, MPC, and adaptive MPC for control, while combining motion planning with a sampling neighborhood graph method.
Motion Degeneracy in Self-supervised Learning of Elevation Angle Estimation for 2D Forward-Looking Sonar
Yusheng Wang, A. Yamashita
Pose EstimationAudio
🎯 What it does: Studied elevation angle estimation for 2D forward sonar in self-supervised learning, proposing a method that achieves stable training without requiring pre-trained synthetic data.
Motion Magnification in Robotic Sonography: Enabling Pulsation-Aware Artery Segmentation
Dianye Huang, Zhongliang Jiang
SegmentationImageBiomedical DataUltrasound
🎯 What it does: Proposed a neural network (PAS-NN) that utilizes cardiac-induced pulse motion to assist in segmentation, and employs motion amplification technology to extract real-time pulse information to enhance the accuracy and stability of arterial segmentation.
Motion Orchestration in Dual-Stage Wafer Scanners
Y. Al-Rawashdeh, M. Janaideh
OptimizationPhysics Related
🎯 What it does: Study motion scheduling in a two-stage wafer scanner, focusing on the interaction between measurement cycles for additional wafer chains and stepping scan cycle chains, and propose two reference trajectory schedules to suppress vibrations caused by inertial forces.
Motion Planning Diffusion: Learning and Planning of Robot Motions with Diffusion Models
João Carvalho, Jan Peters
Robotic IntelligenceDiffusion model
🎯 What it does: Use diffusion models to learn trajectory distributions as priors and directly sample posterior trajectories through an inverse denoising process under given task objectives, thereby accelerating robot motion planning.
MOTLEE: Distributed Mobile Multi-Object Tracking with Localization Error Elimination
Mason B. Peterson, J. How
Object TrackingRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose a distributed mobile multi-target tracking algorithm that enables a team of robots to collaboratively track moving objects in the presence of localization errors.
Motor Unit Action Potential Based Classification of Hand and Arm Motions
Michael D Twardowski, John P. Chiodini
ClassificationTime SeriesBiomedical Data
🎯 What it does: A lightweight, real-time architecture for classifying arm and hand movements was designed, utilizing motor unit action potential features from surface electromyography (sEMG) signals.
MPC-Based Human-Accompanying Control Strategy for Improving the Motion Coordination Between the Target Person and the Robot
Jianwei Peng, Houde Dai
OptimizationRobotic Intelligence
🎯 What it does: Proposes a human companion control strategy based on Model Predictive Control (MPC) to improve motion coordination between mobile robots and target individuals, enabling adaptation to changes in the target's movement and obstacle avoidance.
MUFeat: Multi-Level CNN and Unsupervised Learning for Local Feature Detection and Description
Sheng-Hung Kuo, Kuan-Wen Chen
RetrievalConvolutional Neural NetworkImage
🎯 What it does: Proposed MUFeat, an unsupervised learning framework that jointly learns local feature detectors and descriptors without requiring ground truth correspondences
Multi-Agent Collective Construction Using 3D Decomposition
A. Srinivasan, Bhaskar Vundurthy
OptimizationRobotic Intelligence
🎯 What it does: A 3D structural decomposition algorithm is proposed to address the problem of multi-robot collective construction, and mixed integer linear programming (MILP) is used to generate construction plans for each substructure, ultimately aggregating the plans of individual substructures into a complete structure.
Multi-Agent Multi-Objective Ergodic Search Using Branch and Bound
A. Srinivasan, H. Choset
Optimization
🎯 What it does: Propose a multi-agent multi-objective equipotential search task allocation algorithm based on branch and bound.
Multi-Arm Robot Task Planning for Fruit Harvesting Using Multi-Agent Reinforcement Learning
Tao Li, Qingchun Feng
Robotic IntelligenceReinforcement LearningAgriculture Related
🎯 What it does: Proposed a task planning strategy for a four-armed fruit-picking robot, employing a Markov game framework and multi-agent reinforcement learning (MARL) for task allocation and coordination.
Multi-Dimensional Deformable Object Manipulation Using Equivariant Models
Tianyu Fu, Chaoqiang Zhao
Robotic Intelligence
🎯 What it does: Proposed a Transporter Network with equivariant encoding and decoding, along with an equivariant goal-conditioned model, for manipulating multi-dimensional deformable objects (e.g., ropes, fabrics, bags).
Multi-Gait Locomotion Planning and Tracking for Tendon-Actuated Terrestrial Soft Robot (TerreSoRo)
A. Mahendran, V. Vikas
OptimizationData-Centric LearningRobotic IntelligenceSimultaneous Localization and MappingImage
🎯 What it does: Proposed a gait-based closed-loop path planning framework applicable to TerreSoRo, achieving real-time path planning and tracking.
Multi-Goal Audio-Visual Navigation Using Sound Direction Map
Haruo Kondoh, Asako Kanezaki
Autonomous DrivingReinforcement LearningMultimodality
🎯 What it does: Proposed and defined the multi-object audio-visual navigation task, analyzed its difficulty, and introduced the Sound Vector Map (SDM) method for dynamic sound source localization;
Multi-IMU Proprioceptive Odometry for Legged Robots
Shuo Yang, Zachary Manchester
Robotic IntelligenceSimultaneous Localization and MappingTime Series
🎯 What it does: Propose a low-cost self-perception odometry scheme by adding an IMU on the foot of the legged robot, and using an extended Kalman filter (EKF) to fuse the data from the body IMU, joint encoders, and foot IMU, achieving accurate estimation of the robot's body and foot positions, and reliably determining the foot contact mode and sliding without tactile or pressure sensors;
Multi-IMU Proprioceptive State Estimator for Humanoid Robots
Fabio Elnecave Xavier, Franccois Goulette
Robotic IntelligenceTime SeriesSequential
🎯 What it does: Proposed a multi-IMU attitude estimator based on the Extended Kalman Filter (EKF), capable of handling arbitrary contact configurations and tracking multiple body parts.
Multi-Instance Task in Swarm Robotics: Sorting Groups of Robots or Objects into Clusters with Minimalist Controllers
Adilson Krischanski, Roberto Silvio Ubertino Rosso
OptimizationRobotic Intelligence
🎯 What it does: Using a multi-robot swarm based solely on a simple response controller, enabling robots or objects to autonomously aggregate and group into different clusters without communication.
Multi-Modal Planning on Regrasping for Stable Manipulation
Jiaming Hu, Henrik I. Christensen
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposes a multi-modal planner based on Markov Decision Process (MDP) that can rearrange target objects to suitable positions for stable manipulation through continuous actions such as sliding, regrasping, and transferring, and verifies its performance improvement in grasping and manipulation tasks in both simulation and real environments.
Multi-Modal Upper Limbs Human Motion Estimation from a Reduced Set of Affordable Sensors
Mohamed Adjel, Vincent Bonnet
Pose EstimationMultimodalityBiomedical Data
🎯 What it does: Developed a human upper limb motion capture system based on a few affordable visual inertial measurement units and a markerless skeleton tracking algorithm
Multi-Objective Sparse Sensing with Ergodic Optimization
Ananya Rao, H. Choset
OptimizationRobotic Intelligence
🎯 What it does: To address the multi-objective sparse sensing problem, the paper proposes the MO-SS-E metric based on the ergodic metric, which is used to plan robot trajectories to optimize sensing timing and positions under various sensor constraints.
Multi-Robot Planning on Dynamic Topological Graphs Using Mixed- Integer Programming
Cora A. Dimmig, Joseph L. Moore
OptimizationRobotic IntelligenceGraph
🎯 What it does: Propose a new method for multi-robot planning using mixed integer programming (MIP), with the core being based on a dynamic topological graph where edge weights dynamically change based on the robots' positions in the graph.
Multi-Scale Point Octree Encoding Network for Point Cloud Based Place Recognition
Zhilong Tang, Hong Zhang
RetrievalTransformerPoint Cloud
🎯 What it does: Propose a multi-scale point octree encoding network (MPOE-Net), which generates high-discrimination global descriptors through a point octree encoding module, a multi-Transformer with grouped offset-attention mechanism, and multi-layer NetVLAD, achieving efficient retrieval for point cloud location recognition.
Multi-Session, Localization-Oriented and Lightweight LiDAR Mapping Using Semantic Lines and Planes
Zehuan Yu, S. Shen
Autonomous DrivingOptimizationComputational EfficiencySimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes a centralized multi-session LiDAR map building framework using lightweight line and plane representations, applicable to urban environments, and achieves consistent maps through a coarse-to-fine approach.
Multi-Source Fusion for Voxel-Based 7-DoF Grasping Pose Estimation
Junning Qiu, Zheng Dang
Pose EstimationRobotic IntelligenceConvolutional Neural NetworkPoint CloudBenchmark
🎯 What it does: Solves the problem of 7-DoF hand pose estimation from point cloud data, proposing a multi-source fused voxel network framework
Multi-Source Soft Pseudo-Label Learning with Domain Similarity-based Weighting for Semantic Segmentation
Shigemichi Matsuzaki, Jun Miura
SegmentationDomain Adaptation
🎯 What it does: Perform domain adaptation training for multi-source semantic segmentation, generating soft pseudo-labels using the probabilities predicted by multi-source models, weighting them based on the similarity between the source and target domains, and finally employing an entropy-based training method to enhance the quality of pseudo-labels.
Multi-UAV Adaptive Path Planning Using Deep Reinforcement Learning
Jonas Westheider, Marija Popovic
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose a multi-robot information path planning method based on deep reinforcement learning for adaptive terrain monitoring scenarios.
Multi-View Robust Collaborative Localization in High Outlier Ratio Scenes Based on Semantic Features
Yujie Tang, Yufeng Yue
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose a robust collaborative localization algorithm for outdoor environments with high outlier rates, HORCL, which utilizes a Mixture Probability Model to compute inlier probabilities and combines a hierarchical EM algorithm to perform two-layer outlier filtering on loop closure constraints and point pairs, thereby enhancing the accuracy and robustness of multi-robot localization.
Multi-View Stereo with Learnable Cost Metric
Guidong Yang, Ben M. Chen
Depth EstimationImagePoint CloudBenchmark
🎯 What it does: Proposes an LCM-MVSNet, a multi-view stereo (MVS) network based on a learnable cost metric (LCM), for more accurate and complete depth estimation and dense point cloud reconstruction.
Multimodal Diffusion Segmentation Model for Object Segmentation from Manipulation Instructions
Yui Iioka, Komei Sugiura
SegmentationVision-Language-Action ModelDiffusion modelMultimodality
🎯 What it does: Proposed a model that can understand natural language instructions and generate pixel-level segmentation masks for target daily objects.
Multiplanar Self-Calibration for Mobile Cobot 3D Object Manipulation Using 2D Detectors and Depth Estimation
T. Dang, M. Huber
Depth EstimationRobotic IntelligenceImagePoint Cloud
🎯 What it does: Propose a multi-plane self-calibration method to register the camera system with the robot end-effector for 3D object manipulation.
Multiple-Contact Estimation for Tendon-Driven Continuum Robots with Proprioceptive Sensor Information by Contact Particle Filter and Kinetostatic Models
Tim-David Job, Moritz Schappler
Robotic Intelligence
🎯 What it does: A method for estimating single-point and multi-point contact forces on tendon-driven continuum robots using proprioceptive sensors (tendon force and length sensors) is proposed.
Navigation Among Movable Obstacles Using Machine Learning Based Total Time Cost Optimization
Kai Zhang, David Filliat
Autonomous DrivingOptimizationReinforcement Learning
🎯 What it does: Proposed a pipeline to solve the NAMO problem by optimizing total time, which includes a supervised learning model for predicting planning and obstacle movement time, and a reinforcement learning-based pose generator.
NaviSTAR: Socially Aware Robot Navigation with Hybrid Spatio-Temporal Graph Transformer and Preference Learning
Weizheng Wang, Byung-Cheol Min
Robotic IntelligenceGraph Neural NetworkTransformerReinforcement LearningGraphBenchmark
🎯 What it does: Proposes a social robot navigation benchmark named NaviSTAR, employing a hybrid spatiotemporal graph transformer, offline reinforcement learning, and preference learning, along with the design of a social score function for evaluation.
navlie: A Python Package for State Estimation on Lie Groups
C. Cossette, J. R. Forbes
Optimization
🎯 What it does: Developed a Python package called navlie for fast prototyping of state estimation on Lie groups.
NBV-SC: Next Best View Planning Based on Shape Completion for Fruit Mapping and Reconstruction
Rohit Menon, Maren Bennewitz
Robotic IntelligencePoint CloudAgriculture Related
🎯 What it does: Proposes a next-best-view planning method based on shape completion for fruit mapping and reconstruction, validated through simulation and real-world experiments.
Need for Speed: Fast Correspondence-Free Lidar-Inertial Odometry Using Doppler Velocity
David J. Yoon, Timothy D. Barfoot
Autonomous DrivingComputational EfficiencySimultaneous Localization and MappingPoint Cloud
🎯 What it does: Propose a fast correspondence-free LiDAR-IMU odometry method using Doppler velocity measurements from FMCW LiDAR.
Nematode-Inspired Cable Routing Method for Cable Driven Redundant Manipulator*
Hoyoung Kim, Jungwon Yoon
Robotic Intelligence
🎯 What it does: Proposed and verified a cable routing method based on the alternating muscle structure of nematodes, combined with a pulley amplification structure, using an 8-DOF cable-driven redundant robotic arm prototype constructed with quaternion-based joints for kinematic and stiffness experiments.
NeRF-SLAM: Real-Time Dense Monocular SLAM with Neural Radiance Fields
Antoni Rosinol, L. Carlone
Pose EstimationDepth EstimationNeural Radiance FieldSimultaneous Localization and MappingImage
🎯 What it does: Proposed a new geometric and photometric 3D mapping pipeline capable of achieving precise and real-time scene reconstruction from arbitrarily captured monocular images.
NeU-NBV: Next Best View Planning Using Uncertainty Estimation in Image-Based Neural Rendering
Liren Jin, Marija Popovic
OptimizationRobotic IntelligenceNeural Radiance FieldImage
🎯 What it does: Designed a mapless next best view planning framework to collect the most informative RGB images for unknown scenes under a limited measurement budget.
Neural Field Movement Primitives for Joint Modelling of Scenes and Motions
Ahmet E. Tekden, Yasemin Bekiroglu
GenerationRepresentation LearningNeural Radiance FieldImageMesh
🎯 What it does: Proposes a learning from demonstration (LfD) method using neural fields, which employs shared embeddings to simultaneously learn generative representations of scenes and motions, enabling efficient and accurate learning of new skills.
Neural Implicit Vision-Language Feature Fields
Kenneth Blomqvist, R. Siegwart
SegmentationNeural Radiance FieldImagePoint Cloud
🎯 What it does: Proposes a zero-shot volumetric open-vocabulary semantic scene segmentation method by integrating visual-language model image features into neural implicit representations to create a feature field that can be segmented based on text prompts.
Next-Best-View Selection from Observation Viewpoint Statistics
S. Aravecchia, Cédric Pradalier
Robotic IntelligenceSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes a complete method that utilizes viewpoint statistics as prior information to estimate and enhance 3D reconstruction quality, integrating this information into the Next Best View (NBV) selection strategy;
No Contact Needed: Humans Adapt Their Gait to Suit Legged Robot Companions
Paul M. Riek, Amy R. Wu
Robotic IntelligenceTime SeriesBiomedical Data
🎯 What it does: An experiment was conducted on 14 healthy subjects, recording their gait when walking alone and when walking alongside a small quadruped robot; spatiotemporal and stability parameters were calculated to assess the robot's impact on gait.
Non-Gaussian Uncertainty Minimization Based Control of Stochastic Nonlinear Robotic Systems
Weiqiao Han, B. Williams
OptimizationRobotic Intelligence
🎯 What it does: Designed a closed-loop controller that minimizes non-Gaussian uncertainty to reduce system state deviations caused by probabilistic uncertainties and disturbances.
Non-Linear Heterogeneous Bayesian Decentralized Data Fusion
O. Dagan, N. Ahmed
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Developed a factor graph-based distributed data fusion (FG-DDF) framework for heterogeneous Bayesian distributed fusion problems, enabling robots to efficiently update and fuse probability density functions on overlapping random state subsets.
Non-Parametric Self-Identification and Model Predictive Control of Dexterous In-Hand Manipulation
Podshara Chanrungmaneekul, Kaiyu Hang
OptimizationRobotic IntelligenceImage
🎯 What it does: Proposes utilizing non-parametric learning under a small number of exploration actions to enable online self-identification of hand-object local models, which are then embedded into a model predictive control (MPC) framework to achieve high-precision fingertip grasping manipulation.
Nonlinear Deterministic Observer for Inertial Navigation Using Ultra-Wideband and IMU Sensor Fusion
Hashim A. Hashim, M. Abouheaf
Autonomous DrivingTime SeriesOrdinary Differential Equation
🎯 What it does: This paper proposes a nonlinear deterministic navigation observer that utilizes the fusion of UWB and IMU.
Nonlinear Model Predictive Control for Cooperative Transportation and Manipulation of Cable Suspended Payloads with Multiple Quadrotors
Guanrui Li, Giuseppe Loianno
OptimizationRobotic Intelligence
🎯 What it does: Proposed a method for multirotor cooperative suspended load transportation and manipulation based on nonlinear model predictive control (NMPC).
Nonprehensile Planar Manipulation through Reinforcement Learning with Multimodal Categorical Exploration
Juan Del Aguila Ferrandis, S. Vijayakumar
Robotic IntelligenceReinforcement LearningMultimodality
🎯 What it does: Propose a multi-modal category distribution exploration method, using reinforcement learning to achieve planar non-grasping pushing control, enabling the robot to complete pushing tasks under any initial and target object poses with higher precision.
Novel Gripper with Rotatable Distal Joints for Home Robots: Picking and Placing Tableware
Sung-Woo Kim, Sungchul Kang
Robotic Intelligence
🎯 What it does: Designed and experimentally validated a novel gripper suitable for home robots to pick up and place various tableware in narrow cluttered environments.
Null-Space Compliance Variation for Safe Human-Robot Collaboration in Redundant Manipulators using Safety Control Barrier Functions
Julian M. Salt Ducaju, Rolf Johansson
Robotic Intelligence
🎯 What it does: Adjust the compliant behavior of redundant robots in void spaces using Safe Control Barrier Functions (SCBF) to enhance the safety of human-robot collaboration while maintaining the primary Cartesian task unchanged.
OA-Bug: An Olfactory-Auditory Augmented Bug Algorithm for Swarm Robots in a Denied Environment
Siqi Tan, Q. Quan
OptimizationRobotic IntelligenceMultimodalityAudio
🎯 What it does: Proposed the OA-Bug algorithm based on olfactory and auditory signals for swarm robots' search tasks in rejected environments, and validated it in simulations and real robots.
Object Detection Based on Raw Bayer Images
Guoyu Lu
Object DetectionConvolutional Neural NetworkImage
🎯 What it does: Proposes the BayerDetect network, an end-to-end deep object detection framework for raw Bayer images, integrating frequency domain attention blocks and multi-scale deformable convolution spatial attention to enhance spectral context and boundary recognition.
Object Goal Navigation with Recursive Implicit Maps
Shizhe Chen, C. Schmid
Autonomous DrivingTransformerVision-Language-Action ModelSimultaneous Localization and MappingPoint CloudBenchmark
🎯 What it does: Propose a recursively updated implicit spatial map and use transformers and auxiliary tasks to achieve object-target navigation.
Object Manipulation Through Contact Configuration Regulation: Multiple and Intermittent Contacts
Orion Taylor, Alberto Rodriguez
Robotic IntelligenceImage
🎯 What it does: Through a factor graph estimation framework that integrates limited visual feedback, force/torque sensing, and robot proprioception, the estimation and control of all contact points, geometry, and patterns between the robot, object, and environment are achieved, enabling manipulation tasks for unknown objects.
Object Rearrangement Planning for Target Retrieval in a Confined Space with Lateral View
Minjae Kang, Songhwai Oh
OptimizationRobotic Intelligence
🎯 What it does: Object rearrangement task for target retrieval in environments with confined spaces and limited observation directions.
Object-Level Unknown Obstacle Detection
Chuane Huang, Kuan-Wen Chen
Object DetectionDepth EstimationAnomaly Detection
🎯 What it does: Proposes an object-level unknown obstacle detection method that integrates anomaly detection, depth estimation, and object detection techniques, capable of generating bounding box-level anomaly detection results and reducing false positives.
Object-Oriented Option Framework for Robotics Manipulation in Clutter
Jing-Cheng Pang, Bill Huang
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed an object-oriented option framework (O3F) for robotic manipulation in cluttered environments.
Off the Radar: Uncertainty-Aware Radar Place Recognition with Introspective Querying and Map Maintenance
Jianhao Yuan, Matthew Gadd
Autonomous DrivingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposed a multi-session map management system that constructs the optimal map by utilizing the variance attributes of learned embedding spaces, and self-excludes localization queries based on the same variance attributes;
Offline Reinforcement Learning for Quadrotor Control: Overcoming the Ground Effect
Luca Sacchetto, K. Diepold
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose the use of offline reinforcement learning to overcome the sim-to-real transfer problem for small quadrotors under the ground effect, and demonstrate the feasibility of the method.
On Collaborative Robot Teams for Environmental Monitoring: A Macroscopic Ensemble Approach
Victoria Edwards, M. Hsieh
Robotic Intelligence
🎯 What it does: Proposed a macro-integrated approach based on spatial proximity collaboration for task switching in collaborative robot teams for urban river environment monitoring.
On Cyber-Attacks Mitigation for Distributed Trajectory Generators
Y. Al-Rawashdeh, M. Janaideh
Anomaly DetectionSafty and Privacy
🎯 What it does: Proposes a distributed trajectory generator based on immune average consensus behavior for network attack detection and mitigation in multi-agent systems, achieving distributed detection and active defense through the decomposition of consensus values;
On Designing a Learning Robot: Improving Morphology for Enhanced Task Performance and Learning
Maks Sorokin, Mohi Khansari
Robotic Intelligence
🎯 What it does: Propose a comprehensive method that optimizes the robot's morphology as a whole to enhance learning and task execution capabilities.
On Intuitive Control of Ankle-Foot Prostheses: A Sensor Fusion-Based Algorithm for Real-Time Prediction of Transitions to Compliant Surfaces
Charikleia Angelidou, Panagiotis K. Artemiadis
ClassificationMultimodalityBiomedical Data
🎯 What it does: A real-time algorithm based on sensor fusion was studied to predict whether ankle-foot prosthesis users would step on rigid or flexible ground, enabling intuitive control.
On Semi-Autonomous Robotic Telemanipulation Employing Electromyography Based Motion Decoding and Potential Fields
Bonnie Guan, Minas V. Liarokapis
Robotic IntelligenceBiomedical Data
🎯 What it does: Proposed a semi-autonomous teleoperation framework based on EMG motion decoding and potential fields for controlling a dexterous robotic arm-hand system in complex cube and cylinder stacking tasks.
On the Design of Region-Avoiding Metrics for Collision-Safe Motion Generation on Riemannian Manifolds
Holger Klein, T. Asfour
Robotic Intelligence
🎯 What it does: Propose a Riemannian geometry-based regional avoidance metric, which modifies the configuration space metric using a barrier function, enabling robot motion to be generated as geodesics while avoiding restricted areas.
On the Potentials of Surface Tactile Imaging and Dilated Residual Networks for Early Detection of Colorectal Cancer Polyps
Nethra Venkatayogi, F. Alambeigi
Anomaly DetectionConvolutional Neural NetworkImageBiomedical Data
🎯 What it does: Proposes a new diagnostic framework based on the hyper-sensitive visual tactile sensor (HySenSe) and deep residual neural networks to reduce the miss rate in the early detection of colorectal polyps.
On-Robot Bayesian Reinforcement Learning for POMDPs
Hai V. Nguyen, Chris Amato
Robotic IntelligenceReinforcement Learning
🎯 What it does: A Bayesian reinforcement learning framework specifically designed for physical systems is proposed, implemented on a robot with online sampling solutions combining Monte-Carlo tree search and particle filtering; the framework utilizes factorized representations to capture expert knowledge and learns in unknown dynamic environments.
Onboard Predictive Flocking of Quadcopter Swarm in the Presence of Obstacles and Faulty Robots
Giray Önür, Erol Şahin
Robotic Intelligence
🎯 What it does: Extend the predictive swarm model to achieve safe UAV swarms in environments with obstacles and faulty robots.
One-4-All: Neural Potential Fields for Embodied Navigation
Sacha Morin, L. Paull
Representation LearningRobotic IntelligenceImageSequential
🎯 What it does: Propose the One-4-All (O4A) method, which constructs an end-to-end navigation pipeline without maps by combining self-supervised learning and manifold learning. It achieves greedy minimization on potential fields over image embeddings, trains based on offline RGB sampling and control sequences, requires no depth or pose information, and validates long-range navigation capabilities in 8 simulated Gibson indoor environments and the real Jackal UGV platform.
One-Shot Affordance Learning (OSAL): Learning to Manipulate Articulated Objects by Observing Once
Ruomeng Fan, Yuji Yamakawa
Representation LearningRobotic IntelligenceImage
🎯 What it does: Through a single observation of human demonstration, learning the manipulation of articulated objects and transferring this manipulation to robot execution.
Online Adaptive Disparity Estimation for Dynamic Scenes in Structured Light Systems
Rukun Qiao, Hong-yan Zha
Depth Estimation
🎯 What it does: Proposes a framework based on self-supervised online adaptation, leveraging unsupervised loss functions from long-sequence inputs and sparse trajectory and confidence masks from multi-frame pattern streams, significantly improving online adaptation speed and performance on unseen data
Online Continual Learning for Robust Indoor Object Recognition
Umberto Michieli, Mete Ozay
RecognitionImage
🎯 What it does: Proposes the RobOCLe framework for few-shot online continual learning scenarios, which constructs an enhanced feature space incorporating higher-order statistical moments and performs category prediction by computing similarity within this space.
Online Estimation of 2D Human Arm Stiffness for Peg-in-Hole Tasks with Variable Impedance Control
Huayang Wu, Yanan Li
Robotic Intelligence
🎯 What it does: Proposed a two-dimensional online estimation model for human arm stiffness based on physiological findings, validated the model parameters and accuracy through perturbation experiments under different grip forces; subsequently applied the model to the design of a variable stiffness controller for a hole-pin task
Online Human Capability Estimation Through Reinforcement Learning and Interaction
Chengke Sun, Matteo Leonetti
Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement Learning
🎯 What it does: Propose an online adaptation method based on reinforcement learning and Bayesian inference, enabling robots to automatically assess human capabilities and determine the optimal level of support in collaborative tasks
Online Monocular Lane Mapping Using Catmull-Rom Spline
Zhijian Qiao, S. Shen
Autonomous DrivingOptimizationSimultaneous Localization and MappingImage
🎯 What it does: Proposed an online lane mapping method based on a monocular camera and odometry, generating a lane map based on Catmull-Rom splines.
Online Self-Supervised Thermal Water Segmentation for Aerial Vehicles
Connor T. Lee, Soon-Jo Chung
SegmentationDomain AdaptationContrastive LearningImage
🎯 What it does: Through online self-supervision, utilizing texture and motion cues to transfer an RGB-trained water segmentation network to the aerial thermal imaging domain.