arXivSub Start free trial

IROS 2023 Papers — Page 6

IEEE/RSJ International Conference on Intelligent Robots and Systems · 1195 papers

Heterogeneous Coalition Formation and Scheduling with Multi-Skilled Robots

Ashay Aswale, Carlo Pinciroli

OptimizationRobotic Intelligence

🎯 What it does: Propose two centralized algorithms for task scheduling and dynamic coalition formation in heterogeneous multi-skill robot systems to meet the multi-skill requirements of tasks.

Heterogeneous Robot-Assisted Services in Isolation Wards: A System Development and Usability Study

Youngsun Kwon, Yoonseob Lim

Robotic Intelligence

🎯 What it does: This study proposes and implements a robot-assisted system for isolation wards, including remote healthcare, emergency alarm services, and medical supply delivery, and provides interactive dashboards and applications for medical staff and patients;

Hierarchical Adaptive Control for Collaborative Manipulation of a Rigid Object by Quadrupedal Robots

M. Sombolestan, Quan Nguyen

OptimizationRobotic Intelligence

🎯 What it does: Studied a hierarchical adaptive control system for quadruped robots to collaboratively transport rigid objects with unknown weight (up to 18 kg) and uncertain shape, while maintaining the ability to track the target trajectory on different terrains.

Hierarchical Attention Network for Planning-Informed Multi-Agent Trajectory Prediction

Wenyi Xiong, Ziheng Qi

Autonomous DrivingTime SeriesSequential

🎯 What it does: Propose a trajectory prediction method based on a hierarchical attention mechanism, integrating the planning information of the host vehicle;

Hierarchical Decision Transformer

André Rosa de Sousa Porfírio Correia, L. Alexandre

Robotic IntelligenceTransformerReinforcement LearningSequentialBenchmark

🎯 What it does: Proposed the Hierarchical Decision Transformer (HDT), which guides low-level controllers through high-level subgoal selection to achieve behavior cloning and imitation learning without task knowledge, significantly enhancing the robustness of Transformer methods in long-horizon and sparse reward tasks.

Hierarchical Imitation Learning for Stochastic Environments

M. Igl, Shimon Whiteson

Autonomous DrivingReinforcement LearningVideoPoint Cloud

🎯 What it does: Proposes the Robust Type Conditioning (RTC) method to address type distribution drift in random environments;

Hierarchical Relaxation of Safety-critical Controllers: Mitigating Contradictory Safety Conditions with Application to Quadruped Robots

Jaemin Lee, A. Ames

OptimizationRobotic Intelligence

🎯 What it does: This paper proposes a hierarchical relaxation method to address multiple conflicting safety constraints in robot safety control, achieving feasible solutions through control barrier functions (CBF) and quadratic programming (QP); the effectiveness of this method is validated in quadruped robot experiments.

Hierarchical Semi-Supervised Learning Framework for Surgical Gesture Segmentation and Recognition Based on Multi-Modality Data

Zhili Yuan, Dandan Zhang

RecognitionSegmentationTransformerMultimodalityBiomedical Data

🎯 What it does: Developed a hierarchical semi-supervised learning framework based on multimodal data (kinematic and visual) for the segmentation and recognition of surgical actions.

Hierarchical Transformer for Visual Affordance Understanding using a Large-scale Dataset

Syed Afaq Ali Shah, Z. Khalifa

SegmentationTransformerImage

🎯 What it does: Constructed a large-scale multi-view RGBD visual affordance learning dataset and proposed a hierarchical Transformer-based affordance segmentation model named VAT.

High-Accuracy Injection Using a Mobile Manipulation Robot for Chemistry Lab Automation

Angelos Angelopoulos, Ron Alterovitz

Robotic IntelligenceImage

🎯 What it does: Developed a method for automatic injection using a mobile manipulator robot, utilizing deep learning for syringe localization and visual servoing to achieve high-precision injection.

High-Curvature Consecutive Tip Steering of a Soft Growing Robot for Improved Target Reachability

Dong-Geol Lee, Jee-Hwan Ryu

Robotic Intelligence

🎯 What it does: Proposed a continuous high-curvature turning mechanism for the tip of a soft-growing robot, which was validated through experiments.

Hilbert Space Embedding-Based Trajectory Optimization for Multi-Modal Uncertain Obstacle Trajectory Prediction

Basant Sharma, A. K. Singh

Autonomous DrivingOptimization

🎯 What it does: Proposed a trajectory optimizer that leverages distribution information to enable safe path planning based on the multi-modal distribution of obstacle trajectories

HistoDepth - Novel Depth Perception for Safe Collaborative Robots

Cornelius Buerkle, Kay-Ulrich Scholl

Depth EstimationRobotic IntelligenceImage

🎯 What it does: Proposed a new depth perception method called HistoDepth to enhance the safety and flexibility of collaborative robots.

Holistic Parking Slot Detection with Polygon-Shaped Representations

Lihao Wang, Xavier Perrotton

Object DetectionAutonomous DrivingConvolutional Neural NetworkImage

🎯 What it does: Propose a one-step overall parking spot detection network, HPS-Net, which directly outputs top-view vertex coordinates of parking spots

Hovering Control of Flapping Wings in Tandem with Multi-Rotors

Aniket Dhole, A. Ramezani

Robotic Intelligence

🎯 What it does: Studied the shield design using multiple small thrusters to stabilize the hover flight of Aerobat, and proposed an observer to estimate the shield's unknown states for achieving closed-loop control.

How Do Humans Provide Motion Assistance for a Robotic Shape-Tracing Task?

Taylor M. Higgins, A. M. Fey

Robotic Intelligence

🎯 What it does: This paper proposes a 'flipping' method, allowing humans to assist robots with motion errors in shape tracking tasks, investigating how humans compensate for robot errors.

How the Fingerprint Effect Applies to Digitized Fingerprint-Like Structures

Robert Kovenburg, B. Aksak

Physics Related

🎯 What it does: Experimental verification using micro-pillar array sensors at different angles demonstrated the impact of digitized structures on fingerprint effects, and the existence of four virtual textures was confirmed by measuring vibration frequencies.

How to Achieve Maneuverability and Adaptability in an Underactuated Robotic Fish by using a Bio-inspired Control Approach

G. Manduca, Donato Romano

Robotic Intelligence

🎯 What it does: Proposes a CPG-based control strategy for underactuated bionic fish, utilizing a single DC motor to convert motor oscillations into fish tail movement through magnetic coupling and linear drive systems, and regulating tail torque via proprioceptive feedback to achieve speed and steering control.

How to Make a Robot Grumpy Teaching Social Robots to Stay in Character with Mood Steering

Eric Nichols, Randy Gomez

Robotic IntelligenceLarge Language Model

🎯 What it does: Proposed and implemented an emotion-guided framework to help social robots maintain consistent emotional expressions during interactions.

Human Preferred Augmented Reality Visual Cues for Remote Robot Manipulation Assistance: from Direct to Supervisory Control

Achyuthan Unni Krishnan, Zhi Li

Robotic Intelligence

🎯 What it does: Through user studies, this paper explores people's preferences for augmented reality (AR) visual cues under different levels of robot autonomy and analyzes the impact of learning methods on these preferences.

Human-Aware Navigation in Crowded Environments Using Adaptive Proxemic Area and Group Detection

Carlos Medina Sánchez, P. J. Marrón

Computational EfficiencyRobotic Intelligence

🎯 What it does: Proposes a human-aware navigation method based on adaptive neighborhood areas and group detection, capable of predicting individual and group trajectories, avoiding collisions, and maintaining socially acceptable behavior.

Human-in-the-Loop Optimization of Active Back-Support Exoskeleton Assistance Via Lumbosacral Joint Torque Estimation

Andreas Sochopoulos, C. Natali

OptimizationRobotic IntelligenceBiomedical Data

🎯 What it does: Design of an assistive controller for an active back-support exoskeleton using a single IMU through human-in-the-loop (HIL) optimization, validated via multiple load-lifting experiments with three subjects.

Human-Robot Collaboration for Unknown Flexible Surface Exploration and Treatment Based on Mesh Iterative Learning Control

Jingkang Xia, Yanan Li

Robotic IntelligenceMesh

🎯 What it does: Propose a human-robot collaboration framework that enables robots to explore unknown flexible surfaces and achieve target contact force through control; during this process, the robot can identify surface geometry and stiffness.

Humanoid Walking System with CNN-Based Uneven Terrain Recognition and Landing Control with Swing-Leg Velocity Constraints

Shimpei Sato, Masayuki Inaba

RecognitionOptimizationRobotic IntelligenceConvolutional Neural NetworkImage

🎯 What it does: Achieving bipedal walking for humanoid robots on uneven terrain by using CNN to recognize terrain and control the landing points of both legs to reduce impact, while satisfying gait stability and swing leg speed constraints;

Hybrid Learning- and Model-Based Planning and Control of In-Hand Manipulation

Rana Soltani-Zarrin, Rianna M. Jitosho

Robotic Intelligence

🎯 What it does: Propose a hybrid learning and model-driven hierarchical framework for fully actuated multi-fingered hands to perform in-hand manipulation with grasping transformations, focusing on scenarios where the grasp is suitable for tool use under target poses.

Hybrid Map-Based Path Planning for Robot Navigation in Unstructured Environments

Jiayang Liu, Huimin Lu

Robotic Intelligence

🎯 What it does: Proposes a new hybrid map representation and develops a path planning method that considers robot posture for traversability assessment, aiming to achieve safe and efficient robot navigation.

Hybrid Object Tracking with Events and Frames

Zhichao Li, C. Bartolozzi

Object TrackingPose EstimationConvolutional Neural NetworkImageMultimodalityPoint Cloud

🎯 What it does: Developed a dual Kalman filter combining event cameras and RGB-D for high-speed and accurate object pose tracking.

Hybrid Tendon and Ball Chain Continuum Robots for Enhanced Dexterity in Medical Interventions

G. Pittiglio, P. Dupont

Robotic Intelligence

🎯 What it does: This paper proposes a hybrid continuous robot design that combines proximal tendon-driven segments and distal permanent magnet ball telescoping segments driven by external magnets, and conducts kinematic modeling, workspace description, and experimental validation of this design.

Hydra-Multi: Collaborative Online Construction of 3D Scene Graphs with Multi-Robot Teams

Yun Chang, L. Carlone

Robotic IntelligenceSimultaneous Localization and MappingWorld ModelPoint CloudGraph

🎯 What it does: Propose the Hydra-Multi system, which can online construct 3D scene graphs from sensor data collected by multi-robot teams. The centralized system processes incremental inputs from multiple robots, computes relative transformations, and fuses loop closure detection to achieve joint 3D scene graph construction.

HyperTraj: Towards Simple and Fast Scene-Compliant Endpoint Conditioned Trajectory Prediction

Renhao Huang, Yang Song

Autonomous DrivingComputational EfficiencyConvolutional Neural NetworkVideo

🎯 What it does: Propose a fully convolutional, endpoint-based trajectory prediction framework called HyperTraj, which generates multiple feasible trajectories in one go using dynamic convolution.

I2mpedance - A Passivity Based Integrative Impedance Controller for Precise and Compliant Manipulation and Interaction

Florian Voigt, Sami Haddadin

Robotic Intelligence

🎯 What it does: Improving position accuracy and torque limits by adding an integral term to standard Cartesian impedance control, and combining a virtual energy box to ensure the overall passivity of the controller, with research conducted on the absolute positioning accuracy of robots within their workspace.

I2P-Rec: Recognizing Images on Large-Scale Point Cloud Maps Through Bird's Eye View Projections

Yixuan Li, Hui Shen

RecognitionDepth EstimationAutonomous DrivingConvolutional Neural NetworkImagePoint Cloud

🎯 What it does: Propose the I2P-Rec method, which converts images into point clouds through a depth estimation network, then projects them into a bird's-eye view (BEV). Global features are extracted using CNN+NetVLAD for cross-modal matching, achieving image localization on large-scale point cloud maps.

I3DOD: Towards Incremental 3D Object Detection via Prompting

Wenqi Liang, Kangru Wang

Object DetectionPrompt Engineering

🎯 What it does: Proposed a prompt-based incremental 3D object detection framework called I3DOD.

IDA: Informed Domain Adaptive Semantic Segmentation

Zheng Chen, Lantao Liu

SegmentationDomain AdaptationImage

🎯 What it does: Proposed the IDA self-training framework guided by category-level segmentation performance for unsupervised domain adaptation in semantic segmentation

IF-Based Trajectory Planning and Cooperative Control for Transportation System of Cable Suspended Payload With Multi UAVs

Yu Zhang, Jiuxiang Dong

OptimizationRobotic Intelligence

🎯 What it does: Studied the control and trajectory planning problem for multi-UAV collaborative transportation systems with suspended payloads, proposing a payload controller considering the dynamic coupling between UAVs and payloads to achieve active swing suppression and track complex trajectories; meanwhile, three Insetting Formation (IF) algorithms were designed to handle complete obstacle shapes, generating collision-free waypoints and integrating them into an IF strategy to improve obstacle avoidance success rates and reduce algorithmic complexity during high-speed flight.

Image Restoration via UAVFormer for Under-Display Camera of UAV

Zhuoran Zheng, X. Jia

RestorationTransformerImage

🎯 What it does: This paper proposes a deep network called UAVFormer specifically designed to address image degradation caused by transparent film covering UAV-mounted cameras, achieving image restoration.

Image-based Regularization for Action Smoothness in Autonomous Miniature Racing Car with Deep Reinforcement Learning

Hoang-Giang Cao, I-Chen Wu

Autonomous DrivingReinforcement LearningImage

🎯 What it does: Designed a image-based action smoothing regularization method named 1-RAS, and an impact ratio-based adaptive regularization control (IR), applied to deep reinforcement learning control for autonomous small racing cars;

Image-Based Visual Servo Control for Aerial Manipulation Using a Fully-Actuated UAV

Guanqi He, S. Scherer

Object TrackingRobotic IntelligenceImage

🎯 What it does: Developed an image-based visual servo control strategy, using fully actuated drone platforms to perform high-altitude operations in bridge maintenance;

Imitation Is Not Enough: Robustifying Imitation with Reinforcement Learning for Challenging Driving Scenarios

Yiren Lu, S. Levine

Autonomous DrivingReinforcement LearningSequential

🎯 What it does: Train a driving strategy combining imitation learning and reinforcement learning, and evaluate its robustness in test scenarios with different levels of collision risk

Imitation-Guided Multimodal Policy Generation from Behaviourally Diverse Demonstrations

Shibei Zhu, Ville Kyrki

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes an imitation-guided reinforcement learning framework to address the problem of learning multimodal policies from limited state-only demonstrations, and introduces the LfBD (Learning from Behaviourally diverse Demonstration) algorithm.

Impact of Imperfect Exoskeleton Algorithms on Step Characteristics, Task Performance, and Perception of Exoskeleton Performance

Man I Wu, Leia Stirling

Robotic IntelligenceTextTabular

🎯 What it does: Evaluated five different bipedal ankle exoskeleton control algorithms, with participants completing directional stepping tasks under error rates up to 10%, analyzing step length, step width, task error, and user perception.

Impact-Friendly Object Catching at Non-Zero Velocity Based on Combined Optimization and Learning

Jianzhu Zhao, Arash Ajoudani

OptimizationRobotic Intelligence

🎯 What it does: Proposing a non-grasping high-speed object capture method that combines optimization and learning

Implementation of a Cosserat Rod-Based Configuration Tracking Controller on a Multi-Segment Soft Robotic Arm

Azadeh Doroudchi, S. Berman

Robotic IntelligencePhysics Related

🎯 What it does: Implement closed-loop configuration tracking control for a multi-segment soft robotic arm with independently controllable segments using the Cosserat soft rod model, and execute the control in real-time on an actual robot.

Implications of Personality on Cognitive Workload, Affect, and Task Performance in Remote Robot Control

Go-Eum Cha, Byung-Cheol Min

Robotic IntelligenceMultimodality

🎯 What it does: Studied the impact of personality traits (such as extraversion, conscientiousness, agreeableness) of robot operators on the performance of remote robot control tasks, and assessed the relationships between personality traits, cognitive workload, emotions, and task performance through correlation analysis.

Implicit Projection: Improving Team Situation Awareness for Tacit Human-Robot Interaction via Virtual Shadows

Andrew Boateng, Yu Zhang

Robotic Intelligence

🎯 What it does: Enhancing team situational awareness through virtual shadows via implicit projection to achieve silent human-computer interaction

Improved Inference of Human Intent by Combining Plan Recognition and Language Feedback

Ifrah Idrees, Stefanie Tellex

RecognitionRobotic Intelligence

🎯 What it does: Proposed a dialogue-based goal recognition framework (D4GR) that improves human intent inference by asking users questions to clarify noisy sensor data and suboptimal behaviors.

Improving Amputee Endurance over Activities of Daily Living with a Robotic Knee-Ankle Prosthesis: A Case Study

T. Best, Robert D. Gregg

Robotic IntelligenceBiomedical Data

🎯 What it does: Conduct a case study on an above-knee amputee, using active robotic knee-ankle prosthetics and passive prosthetics for rapid cycling activities to evaluate walking endurance.

Improving Deep Dynamics Models for Autonomous Vehicles with Multimodal Latent Mapping of Surfaces

Johan Vertens, Wolfram Burgard

Autonomous DrivingMultimodality

🎯 What it does: Designed and implemented a surface-perception dynamics model based on multi-modal latent mapping, which was evaluated on a real mini electric vehicle.

Improving Human-Robot Interaction Effectiveness in Human-Robot Collaborative Object Transportation Using Force Prediction

J. E. Domínguez-Vidal, Alberto Sanfeliu

Robotic IntelligencePoint CloudTime Series

🎯 What it does: Studied the use of predicted human-applied forces in medium-distance collaborative transportation tasks to enhance human-robot interaction, proposed and validated a deep learning-based force prediction model, and explored its applications in intent detection and trajectory estimation.

Improving Reliable Navigation Under Uncertainty via Predictions Informed by Non-Local Information

R. I. Arnob, Gregory J. Stein

Autonomous DrivingGraph Neural Network

🎯 What it does: In partially mapped environments, utilizing globally available non-local information to predict the quality of time-expanded actions entering unseen spaces, thereby improving reliable long-term goal-oriented navigation.

Improving Surgical Situational Awareness with Signed Distance Field: A Pilot Study in Virtual Reality

H. Ishida, R. Taylor

MultimodalityBiomedical Data

🎯 What it does: Developed an open-source library based on the Signed Distance Field (SDF) for implementing a multi-modal surgical navigation system in virtual reality. The system calculates the minimum distance between surgical instruments and anatomical structures, and provides real-time guidance to surgeons through visual, auditory, and tactile modalities; its effectiveness was validated in a mastoidectomy simulation experiment.

Improving the Performance of Backward Chained Behavior Trees that use Reinforcement Learning

Mart Kartašev, Petter Ögren

Robotic IntelligenceReinforcement Learning

🎯 What it does: Improved the performance of backward chaining behavior trees (BT) using reinforcement learning by leveraging active constraints (ACC) identified in theoretical convergence proofs

In-Hand Cube Reconfiguration: Simplified

Sumit Patidar, O. Brock

Robotic Intelligence

🎯 What it does: Proposed a simplified method for in-hand cube reconstruction.

In-Hand Manipulation of Unknown Objects with Tactile Sensing for Insertion

Chaoyi Pan, Jeannette Bohg

OptimizationRobotic Intelligence

🎯 What it does: Propose an incremental method utilizing tactile sensing for reorienting unknown objects within the hand.

In-Situ Measurement of Extrusion Width for Fused Filament Fabrication Process Using Vision and Machine Learning Models

Arya Shabani, Uriel Martinez-Hernandez

Object DetectionSegmentationComputational EfficiencyConvolutional Neural NetworkImage

🎯 What it does: A visual on-site monitoring system is proposed for real-time measurement of extrusion width during the Fused Filament Fabrication process, achieving 2D detection of print path width through a printable housing embedded with a camera.

Incipient Slip Detection with a Biomimetic Skin Morphology

David Córdova Bulens, Benjamin Ward-Cherrier

ClassificationRobotic IntelligenceTime Series

🎯 What it does: This paper designs and implements a biomimetic skin structure based on human fingerprints, and applies it to the TacTip optical tactile sensor for detecting initial slips in tactile contact.

Incorporating Stochastic Human Driving States in Cooperative Driving Between a Human-Driven Vehicle and an Autonomous Vehicle

Sanzida Hossain, Weihua Sheng

Autonomous DrivingOptimizationStochastic Differential Equation

🎯 What it does: Developed a cooperative driving framework that incorporates human drivers' attention levels and tendency to follow recommended commands, and proposed a stochastic model predictive controller (sMPC) for human-vehicle lane merging to optimize autonomous vehicle inputs and influence human driving behavior.

Inertial Propulsion Robot Using the Shape Characteristics of a Streamlined Body Frame

Masatsugu Nishihara, Fumihiko Asano

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: Studied an inertially propelled robot with a streamlined body shape and an internal pendulum, designed equations of motion and control inputs, constructed constraint conditions for contact with smooth ground, and verified its ability to achieve efficient gliding on smooth surfaces through numerical simulation and experiments; simultaneously, Bayesian optimization was employed to search for local optima of motion efficiency, achieving significant energy efficiency improvements in both simulation and experiments.

INF: Implicit Neural Fusion for LiDAR and Camera

Shuyi Zhou, Takeshi Oishi

Pose EstimationAutonomous DrivingNeural Radiance FieldImagePoint Cloud

🎯 What it does: Proposed and implemented the Implicit Neural Fusion (INF) method, which first trains a neural density field using LiDAR frames, then uses camera images and the trained density field to train a neural color field, while simultaneously estimating radar pose and optimizing extrinsic parameters during training.

Influence of Nanoparticle Coating on the Differential Magnetometry and Wireless Actuation of Biohybrid Microrobots

V. Magdanz, Islam S. M. Khalil

Robotic IntelligencePhysics Related

🎯 What it does: Studied the impact of biohybrid microrobots with nanoparticle coatings in nonlinear differential magnetometers and wireless magnetic actuation, utilizing electrostatic attraction between positively charged nanoparticles and negatively charged sperm cells to assemble nanoparticles around sperm cells, and activating the particles through high-frequency alternating current fields and pulsed static fields to measure their nonlinear response.

Informative Path Planning for Scalar Dynamic Reconstruction Using Coregionalized Gaussian Processes and a Spatiotemporal Kernel

Lorenzo Booth, Stefano Carpin

OptimizationTime Series

🎯 What it does: Using Coregionalized Gaussian Processes to estimate dynamic scalar fields varying in space and time to address information path planning problems.

Infrastructure to support robots: a practical, scalable model for comparative evaluation of design choices

Grace McFassel, Dylan A. Shell

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a flexible mathematical model based on the Markov Decision Process (MDP) framework for evaluating design choices of robot external infrastructure, and demonstrated the model's instantiation through four case studies

Initial Task Allocation for Multi-Human Multi-Robot Teams with Attention-Based Deep Reinforcement Learning

Ruiqi Wang, Byung-Cheol Min

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposes a context-aware multi-attribute decision model for the initial task allocation problem in multi-robot teams, and solves it using an attention mechanism-based deep reinforcement learning method.

Insertion, Retrieval and Performance Study of Miniature Magnetic Rotating Swimmers for the Treatment of Thrombi

Yitong Lu, Julien Leclerc

Robotic IntelligenceBiomedical Data

🎯 What it does: Designed and experimentally verified two improved mini magnetic rotating swimmers (MMRS) for thrombus removal and recovery in a bifurcated artery model, while demonstrating tools for inserting, recovering, and switching MMRS in experiments, and proving that both designs can achieve precise 3D path tracking.

InsMOS: Instance-Aware Moving Object Segmentation in LiDAR Data

Neng Wang, Xieyuanli Chen

SegmentationAutonomous DrivingConvolutional Neural NetworkPoint CloudBenchmark

🎯 What it does: Proposed a new network that can predict point-level moving labels and detect instance information of major traffic participants, achieving instance-aware LiDAR moving object segmentation.

Insole-Type Walking Assist Device Capable of Inducing Inversion-Eversion of the Ankle Angle to the Neutral Position

Taku Itami, Takaaki Aoki

🎯 What it does: Developed an insole-based assistive device that induces ankle inversion/eversion upon heel contact, returning the ankle angle to a neutral position.

InstaGrasp: An Entirely 3D Printed Adaptive Gripper with TPU Soft Elements and Minimal Assembly Time

Xin Zhou, A. Spiers

Robotic Intelligence

🎯 What it does: Developed a fully 3D printed adaptive gripper that can be assembled within 10 minutes.

Integrable Whole-Body Orientation Coordinates for Legged Robots

Yu-Ming Chen, J. Pratt

Robotic Intelligence

🎯 What it does: Proposes a full-body pose coordinate based on system configuration, enabling the full-body pose to depend only on the current configuration without requiring motion history; this coordinate can represent the entire system's pose without being bound to any specific link; its rate of change approximates the system's total angular momentum.

Integrated Design of a Robotic Bio-Inspired Trunk

Tanguy Chevillon, R. Merzouki

Robotic Intelligence

🎯 What it does: Developed an integrated body-perception soft robotic biomimetic exoskeleton using Longlin 30 material, and proposed a fast and simple design method.

Interaction-Aware and Hierarchically-Explainable Heterogeneous Graph-based Imitation Learning for Autonomous Driving Simulation

Mahan Tabatabaie, Kang G. Shin

Autonomous DrivingExplainability and InterpretabilityGraph Neural NetworkGenerative Adversarial NetworkGraph

🎯 What it does: Propose an interaction-aware and hierarchically explainable imitation learning framework based on heterogeneous graphs for autonomous driving simulation.

InteractionNet: Joint Planning and Prediction for Autonomous Driving with Transformers

Jiawei Fu, Nanning Zheng

Autonomous DrivingTransformer

🎯 What it does: Proposed InteractionNet, which uses Transformer to achieve joint reasoning of planning and prediction, and captures interactions among traffic participants through global context sharing; additionally, a specialized Transformer module focusing on critical or unseen vehicles was added.

Interactive Spatiotemporal Token Attention Network for Skeleton-Based General Interactive Action Recognition

Yuhang Wen, Mengyuan Liu

RecognitionConvolutional Neural NetworkTransformerGraph

🎯 What it does: Proposes an ISTA-Net network for skeletal-based interactive action recognition, capable of simultaneously modeling spatial, temporal, and interaction relationships.

Interactive Task Learning for Social Robots: A Pilot Study

Alexander Tyshka, W. Louie

Robotic IntelligenceLarge Language ModelText

🎯 What it does: Proposed an interactive task learning (ITL) system based on natural language interaction to enable social assistive robots to learn social tasks

Interpretable Motion Planner for Urban Driving via Hierarchical Imitation Learning

Bikun Wang, Qian Zhang

Autonomous DrivingExplainability and Interpretability

🎯 What it does: Proposes a hierarchical planning architecture, including a high-level grid-based behavior planner and a low-level trajectory planner, providing explainable and controllable path and trajectory generation solutions for urban autonomous driving.

Interpretable Trajectory Prediction for Autonomous Vehicles via Counterfactual Responsibility

Kai-Chieh Hsu, Marco Pavone

Autonomous DrivingExplainability and InterpretabilityReinforcement LearningPoint Cloud

🎯 What it does: Proposed an explainable trajectory prediction framework based on responsibility assessment, validated through simulation on existing autonomous vehicle prediction models.

InterTracker: Discovering and Tracking General Objects Interacting with Hands in the Wild

Yan Shao, Jiming Chen

Object TrackingVideo

🎯 What it does: Proposes a hand-object interaction object discovery and tracking method based on spatiotemporal information.

Investigating the Usability of Collaborative Robot Control Through Hands-Free Operation Using Eye Gaze and Augmented Reality

Joosun Lee, Wansoo Kim

Robotic Intelligence

🎯 What it does: Using HoloLens 2 to control a mobile robot hands-free through eye movements, and achieving speed and navigation via adaptive impedance control

Investigations into Customizing Bilateral Ankle Exoskeletons to Increase Vertical Jumping Performance

Emily A. Bywater, Elliott J. Rouse

OptimizationBiomedical Data

🎯 What it does: Developed a customized control strategy for a bipedal ankle exoskeleton that considers both individual motion biomechanics and user preferences to enhance vertical jump height.

Investigations into Exploiting the Full Capabilities of a Series-Parallel Hybrid Humanoid Using Whole Body Trajectory Optimization

Melya Boukheddimi, Frank Kirchner

OptimizationRobotic Intelligence

🎯 What it does: This paper conducts research on whole-body trajectory optimization for humanoid robots with serial-parallel hybrid topology, focusing on all homogeneous constraints in closed-loop kinematics.

IOSG: Image-Driven Object Searching and Grasping

Houjian Yu, Changhyun Choi

Robotic IntelligenceVision-Language-Action ModelImage

🎯 What it does: Propose an image-driven object search and grasping method that uses reference images to search for and grasp novel target objects.

IPA-3D1K: A Large Retail 3D Model Dataset for Robot Picking

Jochen Lindermayr, Marco F. Huber

ClassificationObject DetectionData SynthesisRobotic IntelligenceImagePoint CloudMeshBenchmark

🎯 What it does: Proposed and constructed the IPA-3D1K dataset, containing color images, 3D scans, textured 3D models, synthetic scenes, and real-world scenes of over 1,000 retail products, demonstrating robotic picking tasks on this dataset.

Irregular Change Detection in Sparse Bi-Temporal Point Clouds Using Learned Place Recognition Descriptors and Point-to-Voxel Comparison

Nikolaos Stathoulopoulos, G. Nikolakopoulos

Anomaly DetectionSimultaneous Localization and MappingPoint Cloud

🎯 What it does: A method for detecting and extracting irregular changes in 2D sparse point clouds using deep learning-based pose recognition descriptors and voxel-to-point comparison is proposed. First, the map fusion algorithm aligns dual-time point clouds. Then, robust and discriminative features are extracted to identify changes between consecutive frames. Subsequently, the changed regions are sampled and compared between two time points to extract occlusions causing the changes.

Is Weakly-Supervised Action Segmentation Ready for Human-Robot Interaction? No, Let's Improve It with Action-Union Learning

Fan Yang, Shan Jiang

SegmentationRobotic IntelligenceVideo

🎯 What it does: Proposes an action segmentation method based on weakly supervised timestamp labels, namely Action-Union Learning, to enhance model performance.

ITIRRT: A Decoupled Framework for the Integration of Machine Learning Into Path Planning

Thibault Barbie, Shigeharu Mukai

Autonomous Driving

🎯 What it does: Proposes a decoupled framework integrating machine learning models into path planning, divided into prediction and planning stages.

Joint Imitation Learning of Behavior Decision and Control for Autonomous Intersection Navigation

Zeyu Zhu, Huijing Zhao

Autonomous DrivingTime SeriesSequential

🎯 What it does: Proposes a hierarchical imitation learning framework that simultaneously learns high-level behavior decision-making and low-level control for autonomous vehicle navigation in dense intersections;

Joint On-Manifold Gravity and Accelerometer Intrinsics Estimation for Inertially Aligned Mapping

R. Nemiroff, B. Lopez

OptimizationRobotic IntelligenceSimultaneous Localization and MappingTime SeriesSequential

🎯 What it does: Propose a fixed time delay factor graph estimator that can simultaneously estimate the bias, scale factor of the accelerometer, and the gravity vector, thereby achieving alignment of the robot's trajectory or map with the inertial frame.

Joint Out-of-Distribution Detection and Uncertainty Estimation for Trajectory Prediction

Julian Wiederer, Vasileios Belagiannis

Anomaly DetectionAutonomous DrivingSequential

🎯 What it does: Proposes a trajectory prediction method that combines Gaussian Mixture Models for outlier detection and error regression networks for uncertainty estimation.

KD-EKF: A Consistent Cooperative Localization Estimator Based on Kalman Decomposition

Ning Hao, Weilong Xia

Autonomous DrivingSimultaneous Localization and Mapping

🎯 What it does: Proposed a KD-EKF algorithm based on the Kalman observable canonical form to enhance the consistency of EKF in cooperative localization

Keypoints-Based Adaptive Visual Servoing for Control of Robotic Manipulators in Configuration Space

Sreejani Chatterjee, B. Çalli

Robotic IntelligenceImage

🎯 What it does: Proposes an adaptive visual servoing method based on keypoints, achieving robotic arm control in configuration space using natural features;

KGNv2: Separating Scale and Pose Prediction for Keypoint-Based 6-DoF Grasp Synthesis on RGB-D Input

Yiye Chen, P. Vela

GenerationPose EstimationImage

🎯 What it does: An improved keypoint-based 6-DoF grasping pose synthesis network is proposed, which can simultaneously estimate grasping pose and camera-grasp length scale from RGB-D inputs.

Kinematically-Decoupled Impedance Control for Fast Object Visual Servoing and Grasping on Quadruped Manipulators

Riccardo Parosi, V. Barasuol

Robotic IntelligenceImage

🎯 What it does: Proposed a fast visual servoing and grasping control process based on joint chain decoupling and impedance control, combining image-based visual servoing (IBVS) to achieve object search, approach, and grasping (SAG)

Kinematics-Only Differential Flatness Based Trajectory Tracking for Autonomous Racing

Yashom Dighe, Karthik Dantu

Autonomous DrivingOptimizationComputational Efficiency

🎯 What it does: Studied the effect of differential flatness control in high-speed trajectory tracking, proposed a kinematic-based controller, and compared it with NMPC.

Knowledge Distillation for Efficient Panoptic Semantic Segmentation: Applied to Agriculture

Maohui Li, Chris McCool

SegmentationKnowledge DistillationConvolutional Neural NetworkImageAgriculture Related

🎯 What it does: Using knowledge distillation to enhance the efficiency and performance of panoptic segmentation models

L3MVN: Leveraging Large Language Models for Visual Target Navigation

Bangguo Yu, M. Cao

Autonomous DrivingRobotic IntelligenceLarge Language ModelVision-Language-Action Model

🎯 What it does: Propose a new framework called L3MVN that infuses common sense into visual goal navigation using large language models (LLMs), incorporating two language-based frontier selection paradigms: zero-shot and feed-forward;

Labelling Lightweight Robot Energy Consumption: A Mechatronics-Based Benchmarking Metric Set

Juan Heredia, M. Kjærgaard

Robotic IntelligenceBenchmark

🎯 What it does: A complete lightweight industrial robot energy consumption benchmark framework was developed, and seven static and dynamic energy consumption indicators were proposed to evaluate five types of robots (UR3e, UR5e, FR3, M0609, Gen3).

LAMP: Leveraging Language Prompts for Multi-Person Pose Estimation

Shengnan Hu, G. Sukthankar

Pose EstimationPrompt EngineeringVision Language ModelImageText

🎯 What it does: Proposed a language prompt-based multi-person human pose estimation method called LAMP, which utilizes text representations generated by CLIP to assist in pose reasoning.

Landmark Based Bronchoscope Localization for Needle Insertion Under Respiratory Deformation

Inbar Fried, Ron Alterovitz

Pose EstimationBiomedical Data

🎯 What it does: Proposed a real-time camera-based deep learning method that utilizes anatomical landmarks such as airway bifurcations to accurately localize the bronchoscope relative to the planned needle insertion pose.

Language Guided Robotic Grasping with Fine-Grained Instructions

Qiang Sun, Xiangyang Xue

Object DetectionSegmentationPose EstimationRobotic IntelligenceTransformerVision Language ModelVision-Language-Action ModelMultimodalityBenchmark

🎯 What it does: A fine-grained language-guided robotic grasping (FLarG) problem is proposed for single RGB images and attribute-rich language instructions, and an end-to-end DyCRIS model based on CLIP is designed to achieve target object localization and 6-DoF pose estimation.

Language Guided Temporally Adaptive Perception for Efficient Natural Language Grounding in Cluttered Dynamic Worlds

Siddharth Patki, Thomas M. Howard

Computational EfficiencyRobotic IntelligenceVision-Language-Action ModelWorld Model

🎯 What it does: Propose a language-guided, time-adaptive perception model that constructs a temporally compact dynamic world model through closed-loop localization and perception, enabling effective natural language localization in cluttered dynamic environments.

Language-Conditioned Observation Models for Visual Object Search

Thao Nguyen, Stefanie Tellex

Object DetectionVision Language ModelImageText

🎯 What it does: Propose a language-conditioned observation model (LCOM) that unifies a visual detector with a noise model into a single deep neural network, embedding it into a partially observable Markov decision process (POMDP) to address visual object search tasks with complex language descriptions.