arXivSub Start free trial

IROS 2024 Papers — Page 3

IEEE/RSJ International Conference on Intelligent Robots and Systems · 1581 papers

Automatic Dietary Monitoring Using Inertial Sensor in Smartwatch

K. Pavlov, Elena Volkova

ClassificationTime Series

🎯 What it does: Detecting eating activities using inertial sensor data from smartwatches and developing and integrating the corresponding detection algorithms.

Automatic Field of View Adjustment of an RCM Constraint-Free Continuum Laparoscopic Robot

Jing Zhang, Mengtang Li

OptimizationRobotic Intelligence

🎯 What it does: Proposed a tension-driven continuous laparoscope that eliminates remote center motion constraints, and designed an automatic field-of-view adjustment method considering the position, size, and eye-hand consistency of surgical instruments; optimized the method through MATLAB and Webots simulation platforms, and built the first-generation prototype to verify real-time performance.

Automatic Image Annotation for Mapped Features Detection

Maxime Noizet, P. Bonnifait

Object DetectionSegmentationSupervised Fine-TuningImageMultimodalityPoint Cloud

🎯 What it does: Integrate three automatic annotation methods: map projection, image segmentation, and LiDAR segmentation for multimodal automatic annotation of road poles, and use this annotated data to fine-tune the pole detection model.

Automatic Spatial Calibration of Near-Field MIMO Radar With Respect to Optical Depth Sensors

V. Wirth, Marc Stamminger

Object DetectionRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud

🎯 What it does: A joint calibration method for simultaneously calibrating optical RGB-D sensors and MIMO radar in the radar near field is proposed, including dedicated calibration targets, automatic target detection and localization, and spatial registration between the two sensor coordinate systems.

Automating ROS2 Security Policies Extraction through Static Analysis

Giacomo Zanatta, Ruffin White

🎯 What it does: Automatically extract ROS2 security policies using static analysis to generate minimal and correct security configurations.

Automating Trophectoderm Cells Aspiration and Separation in Embryo Biopsy at the Blastocyst Stage: A Vision-Based Control Approach

Ihab Abu Ajamieh, James K. Mills

ImageBiomedical Data

🎯 What it does: Developed a vision-based automated system for the extraction and separation of trophectoderm (TE) cells at the blastocyst stage;

AutoNeRF: Training Implicit Scene Representations with Autonomous Agents

Pierre Marza, D. Chaplot

Robotic IntelligenceReinforcement LearningNeural Radiance Field

🎯 What it does: Proposed an automated exploration method that utilizes autonomous agents to actively collect data in unknown three-dimensional environments, training a NeRF model without human intervention to construct high-quality implicit scene representations.

Autonomous Behavior Planning For Humanoid Loco-manipulation Through Grounded Language Model

Jin Wang (Italian Institute of Technology), N. Tsagarakis (Italian Institute of Technology)

Robotic IntelligenceTransformerLarge Language ModelMultimodality

🎯 What it does: Propose a framework based on a large language model (LLM) that enables humanoid robots to autonomously plan and execute displacement and manipulation tasks in unstructured environments, capable of observing and correcting failures during execution; and create a library of robot actions and perception behaviors for task planning, conducting mobility manipulation experiments using the CENTAURO robot in both simulated and real environments.

Autonomous Guidewire Navigation in Dynamic Environments

Valentina Scarponi, Stéphane Cotin

Robotic IntelligenceReinforcement LearningBiomedical Data

🎯 What it does: Train guide wires to autonomously navigate in dynamic vascular environments using deep reinforcement learning, and propose a method to estimate vascular motion from single-view fluorescent images.

Autonomous localization of multiple ionizing radiation sources using miniature single-layer Compton cameras onboard a group of micro aerial vehicles

Michal Werner, M. Saska

Robotic IntelligenceSimultaneous Localization and MappingPhysics Related

🎯 What it does: Propose a method for autonomously localizing multiple γ-ray sources using miniature single-layer Compton cameras on multiple micro-drones.

Autonomous Power Line Tracking with mmWave Radar

N. Malle, E. Ebeid

Object TrackingPose EstimationRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Propose a drone system using millimeter-wave radar to achieve autonomous tracking and 3D reconstruction of power lines;

Autonomous Robotic Assembly: From Part Singulation to Precise Assembly

Keita Ota, Diego Romeres

Robotic IntelligenceImageMultimodality

🎯 What it does: Achieving autonomous assembly of a gearbox under the condition of minimal structural requirements;

Autonomous Storytelling for Social Robot with Human-Centered Reinforcement Learning

Lei Zhang, Guangliang Li

Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: Enable the social robot Haru to learn personalized storytelling styles tailored to users' emotional states in real-time interactions through human-centric reinforcement learning.

AVM-SLAM: Semantic Visual SLAM with Multi-Sensor Fusion in a Bird’s Eye View for Automated Valet Parking

Ye Li, Xiaolin Qin

Autonomous DrivingConvolutional Neural NetworkSimultaneous Localization and MappingImageMultimodality

🎯 What it does: Proposes AVM-SLAM, a BEV semantic visual SLAM system that integrates multi-sensors (four fisheye cameras, wheel encoders, IMU) to achieve precise localization in underground parking environments with poor lighting, sparse textures, repetitive structures, and no GPS.

Avoiding Object Damage in Robotic Manipulation

Erica Aduh, M. Nambi

ClassificationRobotic IntelligenceTransformerLarge Language ModelVision Language ModelImageTextMultimodality

🎯 What it does: A classification model system was developed to predict whether objects would be damaged during warehouse robot operations, and its effectiveness was validated through experiments.

Backpropagation-Based Analytical Derivatives of EKF Covariance for Active Sensing

Jonas Benhamou, Camille Chapdelaine

OptimizationComputational EfficiencyRobotic Intelligence

🎯 What it does: Proposes an analytical method using backpropagation to compute derivatives of the extended Kalman filter (EKF) covariance with respect to all inputs, and generates perception-driven optimal motion plans based on these gradients.

BAM: Box Abstraction Monitors for Real-time OoD Detection in Object Detection

Changshun Wu, S. Bensalem

Object DetectionAnomaly DetectionImage

🎯 What it does: Proposes Box Abstraction-based Monitors (BAM), which performs real-time OoD detection on object detection models using convex box abstractions with finite unions, without requiring retraining or modifying the network architecture.

Barely-Visible Surface Crack Detection for Wind Turbine Sustainability

Sourav Agrawal, Jonathan Lwowski

Object DetectionAnomaly DetectionImage

🎯 What it does: This paper proposes a dataset of barely visible hairline cracks for wind turbine blades and establishes an end-to-end crack detection pipeline from image acquisition to automatic maintenance recommendations.

BaRiFlex: A Robotic Gripper with Versatility and Collision Robustness for Robot Learning

Gu-Cheol Jeong, Roberto Mart'in-Mart'in

Robotic IntelligenceReinforcement Learning

🎯 What it does: Designed and evaluated a hybrid soft and rigid robotic hand named BaRiFlex to enhance collision robustness and grasp diversity in robot learning processes.

BaSeNet: A Learning-based Mobile Manipulator Base Pose Sequence Planning for Pickup Tasks

Lakshadeep Naik, Norbert Kruger

Robotic IntelligenceGraph Neural NetworkReinforcement Learning

🎯 What it does: Proposes BaSeNet, a learning-based mobile manipulator base pose sequence planning method, which uses reinforcement learning to learn the base pose of a single object and the grasping order, and employs hierarchical learning and graph neural networks to handle variable states and actions;

Bayesian Deep Predictive Coding for Snake-like Robotic Control in Unknown Terrains

William Ziming Qu, Yuanyuan Jia

Robotic IntelligenceTransformer

🎯 What it does: Proposes the SnakeFormer model based on deep predictive coding for effectively modeling the spatiotemporal interactions of multi-link snake robots.

Bayesian Floor Field: Transferring people flow predictions across environments

Francesco Verdoja, Ville Kyrki

Domain Adaptation

🎯 What it does: Proposed a Bayesian method that combines environmental geometric information with human trajectory observations, using an occupancy depth prior to build an initial transition model, followed by Bayesian inference to update the model, achieving data-efficient and cross-environment generalizable pedestrian flow prediction.

Bayesian Optimization for Sample-Efficient Policy Improvement in Robotic Manipulation

Adrian Röfer, A. Valada

OptimizationRobotic Intelligence

🎯 What it does: Proposed a hybrid method BOpt-GMM that combines imitation learning with autonomous experience collection, training a Gaussian Mixture Model with a small number of demonstrations and improving the model through Bayesian optimization in sparse reward environments to achieve sample-efficient robotic manipulation skill learning.

BayRnTune: Adaptive Bayesian Domain Randomization via Strategic Fine-tuning

Tianle Huang, Sehoon Ha

Domain AdaptationSupervised Fine-TuningReinforcement Learning

🎯 What it does: Proposes an algorithm called BayRnTune that accelerates Bayesian domain randomization learning through strategic fine-tuning, leveraging previously trained policies as priors for fine-tuning.

BE-SLAM: BEV-Enhanced Dynamic Semantic SLAM with Static Object Reconstruction

Jun Luo

Autonomous DrivingOptimizationSimultaneous Localization and Mapping

🎯 What it does: Proposes a visual SLAM framework called BE-SLAM based on BEV perception results, which can handle dynamic objects, occlusions, and partially observable objects, and optimizes poses by reconstructing static objects.

Behavior Tree Based Decentralized Multi-agent Coordination for Balanced Servicing of Time Varying Task Queues

Niklas Dahlquist, G. Nikolakopoulos

OptimizationRobotic IntelligenceAgentic AI

🎯 What it does: Proposes a reactive multi-agent coordination architecture tailored for scenarios such as underground mines and automated factories, designed to manage material flow from the production/extraction phase to the delivery/deployment phase.

Behavior-Actor: Behavioral Decomposition and Efficient-Training for Robotic Manipulation

Wenyi Jiang, Zhihao Cui

Robotic IntelligenceTransformerLarge Language ModelTextRetrieval-Augmented Generation

🎯 What it does: Propose a Behavior-Actor (BehAct) framework that utilizes a large language model (LLM) to decompose language instructions into executable behaviors and trains an end-to-end executor (actor) to perform these behaviors, achieving robotic manipulation under language conditions.

Belief-Aided Navigation using Bayesian Reinforcement Learning for Avoiding Humans in Blind Spots

Jinyeob Kim, Donghan Kim

Autonomous DrivingReinforcement Learning

🎯 What it does: Proposed and verified a BNBRL+ algorithm based on Bayesian reinforcement learning for avoiding humans in blind spots and achieving socially aware navigation.

Benchmarking Smoothness and Reducing High-Frequency Oscillations in Continuous Control Policies

Guilherme Christmann, Wei-Chao Chen

Robotic IntelligenceReinforcement LearningBenchmark

🎯 What it does: This paper identifies, classifies, and compares methods to mitigate high-frequency oscillations in deep reinforcement learning, and proposes a hybrid approach combining loss regularization with architectural methods.

Best of Both Worlds: Hybrid SNN-ANN Architecture for Event-based Optical Flow Estimation

Shubham Negi, K. Roy

Spiking Neural NetworkOptical Flow

🎯 What it does: Designed and evaluated a hybrid SNN-ANN architecture for optical flow estimation in event cameras.

BEV Image-based Lane Tracking Control System for Autonomous Lane Repainting Robot

Junghyun Seo, Yongsik Jin

Autonomous DrivingRobotic IntelligenceImage

🎯 What it does: Proposed a lane tracking control system based on bird's-eye view (BEV) images for autonomous lane painting robots, achieving high-precision lane detection using row and column anchor points and implementing a position-based visual pure pursuit algorithm (PV-PP), constructing a high-performance lane painting system.

BEV-CV: Birds-Eye-View Transform for Cross-View Geo-Localisation

Tavis Shore, Oscar Mendez

Image TranslationRetrievalAutonomous DrivingComputational EfficiencyRepresentation LearningContrastive LearningImage

🎯 What it does: Propose the BEV-CV method, which converts ground-level images into semantic bird's-eye views and then performs embedding matching with aerial images to achieve cross-view geolocation.

BEV-ODOM: Reducing Scale Drift in Monocular Visual Odometry with BEV Representation

Yufei Wei, Yue Wang

Pose EstimationAutonomous DrivingConvolutional Neural NetworkSimultaneous Localization and MappingImage

🎯 What it does: Propose the BEV-ODOM framework to reduce scale drift in monocular visual odometry

BEV2PR: BEV-Enhanced Visual Place Recognition with Structural Cues

Fudong Ge, Jin Gao

RecognitionAutonomous DrivingConvolutional Neural NetworkImage

🎯 What it does: Propose a monocular camera-based BEV-enhanced visual localization framework that generates global descriptors combining visual and structural information;

BEVCar: Camera-Radar Fusion for BEV Map and Object Segmentation

J. Schramm, Abhinav Valada

SegmentationAutonomous DrivingImagePoint Cloud

🎯 What it does: Proposed a new method called BEVCar for joint target and map segmentation from a bird's-eye view, combining camera and automotive radar data fusion

BEVLoc: Cross-View Localization and Matching via Birds-Eye-View Synthesis

Christopher Klammer, Michael Kaess

GenerationData SynthesisAutonomous DrivingContrastive LearningImage

🎯 What it does: Proposed a ground-air matching framework based on BEV synthesis for localization in offline environments.

BEVPose: Unveiling Scene Semantics through Pose-Guided Multi-Modal BEV Alignment

Mehdi Hosseinzadeh, Ian D. Reid

SegmentationAutonomous DrivingTransformerImageMultimodalityPoint Cloud

🎯 What it does: The BEVPose framework aligns and fuses BEV representations from cameras and LiDAR by guiding with sensor poses, learning BEV embeddings that integrate both geometric and semantic information.

BEVRender: Vision-based Cross-view Vehicle Registration in Off-road GNSS-denied Environment

Lihong Jin, Michael Kaess

Pose EstimationAutonomous DrivingSimultaneous Localization and MappingImage

🎯 What it does: Proposes a learning-based method called BEVRender for ground vehicle localization in off-road environments where GNSS fails.

Beyond Feasibility: Efficiently Planning Robotic Assembly Sequences That Minimize Assembly Path Lengths

Alexander Cebulla, Torsten Kröger

Robotic IntelligenceGraph Neural NetworkReinforcement LearningGraph

🎯 What it does: Propose an assembly sequence planning method that combines Monte Carlo Tree Search (MCTS) with deep Q-learning, and optimize robot assembly path length through knowledge transfer.

Beyond Success: Quantifying Demonstration Quality in Learning from Demonstration

Muhammad Bilal, W. Johal

Robotic Intelligence

🎯 What it does: In the simulation experiments, the authors quantified and ranked the quality of each demonstration step by introducing motion-related quality features such as manipulability and joint space vibration; subsequently, they trained manipulation task models for actions like cube lifting and cola can fetching, evaluating the impact of demonstration quality on task execution.

Beyond the Cascade: Juggling Vanilla Siteswap Patterns

Mario Gomez Andreu, Jan Peters

Robotic Intelligence

🎯 What it does: Studied juggling patterns with mixed throwing heights, and implemented continuous throws and transitions of Vanilla Siteswap patterns for 3-9 balls in a simulation environment.

Bi-CL: A Reinforcement Learning Framework for Robots Coordination Through Bi-level Optimization

Zechen Hu, Xuan Wang

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Propose the Bi-CL framework, which uses a two-layer optimization structure to learn coordinated behaviors for multi-robot systems

Bi-level Trajectory Optimization on Uneven Terrains with Differentiable Wheel-Terrain Interaction Model

Amith Manoharan, Arun Kumar Singh

Autonomous DrivingOptimizationImage

🎯 What it does: Proposes a purely model-based differentiable wheel-ground interaction model using digital elevation data for six-degree-of-freedom trajectory optimization on rough terrain.

Bidirectional Partial-to-Full Non-Rigid Point Set Registration with Non-Overlapping Filtering

Hao Yu, Zhe Min

Supervised Fine-TuningPoint CloudBiomedical Data

🎯 What it does: Proposed the Bi-NOFNet network for non-rigid registration between partial intraoperative point sets and complete preoperative point sets, improving registration accuracy through non-overlapping filtering.

Bifurcation Identification for Ultrasound-driven Robotic Cannulation

Cecilia G. Morales, Artur Dubrawski

Object DetectionRobotic IntelligenceImageBiomedical DataUltrasound

🎯 What it does: Developed an algorithm called BIFURC that can automatically identify vascular bifurcation points in ultrasound images and provide optimal needle insertion positions for autonomous robotic puncture systems.

Binary Amplitude-Only Hologram Generation for Acoustic End-Effector Design by Physics-based deep learning

Qing Liu, Song Liu

GenerationPhysics RelatedAudio

🎯 What it does: Proposes a deep learning-based method for generating Binary Amplitude-Phase Interferograms (BAOH) to construct precise, high-resolution acoustic end-effectors.

Biodegradable Gliding Paper Flyers Fabricated Through Inkjet Printing

Luca Girardi, Stefano Mintchev

🎯 What it does: The study prepares biodegradable paper glider platforms using an origami movable type printing method;

Bipedal Safe Navigation over Uncertain Rough Terrain: Unifying Terrain Mapping and Locomotion Stability

Kasidit Muenprasitivej, Ye Zhao

Robotic Intelligence

🎯 What it does: Evaluate three Gaussian Process (GP) kernels to learn the height of uncertain rough terrain and motion deviations caused by terrain and planning model errors; propose a hierarchical, dynamics-aware sampling-based navigation planner; design a new trajectory evaluation metric that minimizes motion deviation while maximizing terrain information gain; validate the proposed framework on the Digit bipedal robot in MuJoCo simulations.

Bird’s-Eye-View Trajectory Planning of Multiple Robots using Continuous Deep Reinforcement Learning and Model Predictive Control

Kristian Ceder, Knut Åkesson

OptimizationRobotic IntelligenceReinforcement LearningImage

🎯 What it does: Propose a hybrid sequential method combining Bird’s-Eye-View visual continuous deep reinforcement learning (DRL) with model predictive control (MPC) for trajectory planning and control of multi-robot systems in industrial automation and indoor logistics environments.

Bistable valve for electronics-free soft robots

Longxin Kan, Cecilia Laschi

Robotic Intelligence

🎯 What it does: Proposed and implemented a simple yet powerful electronic valve for soft robots, capable of operating individually, in pairs, or in large groups, and supporting high-frequency synchronized reverse output, valve state storage, and efficient maintenance.

Blending Distributed NeRFs with Tri-stage Robust Pose Optimization

Baijun Ye, Guyue Zhou

Pose EstimationOptimizationNeural Radiance Field

🎯 What it does: Proposed a distributed NeRF system that employs a three-stage pose optimization (bundle adjust Mip-NeRF 360, Frame2Model optimization, and Model2Model optimization) to achieve precise image poses and NeRF transformations, and complete NeRF fusion;

BOMP: Bin-Optimized Motion Planning

Zachary Tam, Ken Goldberg

OptimizationRobotic IntelligenceSupervised Fine-TuningImage

🎯 What it does: Propose a motion planning framework called BOMP for depth box pick-and-place tasks involving six-axis industrial robots equipped with long-nose suction cups, capable of rapidly generating time-optimized, jitter-limited, and collision-free trajectories.

BonnBeetClouds3D: A Dataset Towards Point Cloud-Based Organ-Level Phenotyping of Sugar Beet Plants Under Real Field Conditions

E. Marks, C. Stachniss

SegmentationPoint CloudAgriculture Related

🎯 What it does: Constructed the BonnBeetClouds3D dataset, providing high-resolution multi-angle images generated by UAV-based sugar beet plant 3D point clouds, annotated with plant, leaf, and key points such as tips and bases for instance segmentation and fine organ-level geometric analysis.

Boosting 3D Visual Grounding by Object-Centric Referring Network

Ruilong Ren, Xing Zhang

RetrievalContrastive LearningPoint Cloud

🎯 What it does: Proposed the Object-Centric Reference Network (3D-OCR) for 3D vision localization, combining the Fine-Grained Semantic Enhancement (FSE) module, Deep Relation Modeling (DRM) module, and vision-language contrastive loss to improve the matching performance between 3D point clouds and text descriptions.

Boosting Generalizability towards Zero-Shot Cross-Dataset Single-Image Indoor Depth by Meta-Initialization

Cho-Ying Wu, Ulrich Neumann

Depth EstimationMeta LearningImage

🎯 What it does: Improving the generalization ability of single-image indoor depth estimation using gradient-based meta-learning, particularly for zero-shot cross-dataset inference.

Bridging Language, Vision and Action: Multimodal VAEs in Robotic Manipulation Tasks

G. Sejnova, Karla Stépánová

Robotic IntelligenceVision-Language-Action ModelAuto EncoderMultimodality

🎯 What it does: Explore unsupervised visual-language-action mapping using multimodal VAEs in simulated environments, and propose a model-agnostic training method; systematically evaluate task challenges.

Bridging the Gap to Natural Language-based Grasp Predictions through Semantic Information Extraction

N. Kleer, Antonio Krüger

Robotic IntelligenceLarge Language ModelVision-Language-Action ModelText

🎯 What it does: Propose a text-based grasp prediction method that utilizes Named Entity Recognition (NER) to automatically extract semantic information, and inputs it as features into the grasp prediction model through multi-stage learning.

Bridging the Sim-to-Real Gap with Bayesian Inference

Jonas Rothfuss, Andreas Krause

Domain AdaptationAutonomous DrivingRobotic IntelligenceReinforcement Learning

🎯 What it does: Use the Sim-FSVGD algorithm combined with low-fidelity physical priors to learn robot dynamics and implement high dynamic parking drift on an RC car.

BronchoCopilot: Towards Autonomous Robotic Bronchoscopy via Multimodal Reinforcement Learning

Jianbo Zhao, Hongbin Liu

Robotic IntelligenceReinforcement LearningMultimodalityBiomedical Data

🎯 What it does: Proposed BronchoCopilot, a multi-modal reinforcement learning agent that integrates bronchoscope video images and robotic posture for autonomous bronchoscopy.

BTGenBot: Behavior Tree Generation for Robotic Tasks with Lightweight LLMs

Riccardo Andrea Izzo, Matteo Matteucci

Robotic IntelligenceTransformerLarge Language ModelSupervised Fine-TuningText

🎯 What it does: Fine-tuning a lightweight LLM (up to 7 billion parameters) on a specific dataset to generate robot behavior trees

BuzzRacer: A Palm-sized Autonomous Vehicle Platform for Testing Multi-Agent Adversarial Decision-Making

Zhiyuan Zhang, P. Tsiotras

Autonomous Driving

🎯 What it does: Developed a compact autonomous driving vehicle platform and its corresponding software framework for experimental validation of multi-agent adversarial decision-making

C3P-VoxelMap: Compact, Cumulative and Coalescible Probabilistic Voxel Mapping

Xu Yang, Bo Wang

Autonomous DrivingComputational EfficiencySimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes a compact, cumulative, and coalescible probabilistic voxel mapping method to enhance the performance, accuracy, and memory efficiency of LiDAR odometry.

CaFNet: A Confidence-Driven Framework for Radar Camera Depth Estimation

Huawei Sun, Robert Wille

Depth EstimationImageMultimodalityPoint Cloud

🎯 What it does: Proposes a two-stage end-to-end trained confidence-driven radar-camera fusion network called CaFNet for dense depth estimation, fusing RGB images with sparse noisy radar point clouds.

CAIS: Culvert Autonomous Inspection Robotic System

C. Le, H. La

Depth EstimationAnomaly DetectionRobotic IntelligenceSimultaneous Localization and MappingImage

🎯 What it does: Proposes an autonomous culvert inspection robot system (CAIS) that utilizes RGBD cameras, deep learning, illumination systems, and non-destructive testing (NDT) technologies to generate three-dimensional maps of culvert defects and inspection results.

Calibration-Free Vision-Assisted Container Loading of RTG Cranes

Jianbin Yang, Danwei Wang

Object DetectionSegmentationRobotic IntelligenceConvolutional Neural Network

🎯 What it does: Proposed an integrated solution for target detection and alignment control in RTG crane container loading, implemented on real cranes and validated at Ningbo Port.

Camera Pose Estimation from Bounding Boxes

Václav Vávra, Zuzana Kukelova

Pose Estimation

🎯 What it does: Propose a method for camera pose estimation that utilizes the correspondence between 2D and 3D bounding boxes, constructing a compact scene representation to reduce memory consumption and privacy risks.

Camera-Based Belief Space Planning in Discrete Partially-Observable Domains

Janis Eric Freund, M. Toussaint

OptimizationRobotic IntelligenceImage

🎯 What it does: Extended the camera-based belief space planning method, improved the path tree optimization (PTO) implementation

Can Reasons Help Improve Pedestrian Intent Estimation? A Cross-Modal Approach

Vaishnavi Khindkar, C. Jawahar

Autonomous DrivingExplainability and InterpretabilityVision-Language-Action ModelVideoMultimodality

🎯 What it does: Proposed the reason-rich PIE++ dataset and designed a multimodal multitask learning framework called MINDREAD for simultaneously predicting pedestrian crossing intent and the underlying reasons.

Can Vehicle Motion Planning Generalize to Realistic Long-tail Scenarios?

Marcel Hallgarten, Andreas Zell

Autonomous DrivingLarge Language ModelBenchmark

🎯 What it does: Proposes a new closed-loop benchmark called interPlan, which includes various edge cases and challenging driving scenarios; evaluates existing state-of-the-art rule-based and learning-based planners, demonstrating their inability to safely navigate interPlan scenarios; evaluates planners using only large language models (LLMs) and proposes a novel hybrid planner combining LLM-based behavior planning with rule-driven motion planning, achieving state-of-the-art performance on interPlan.

Cartesian Impedance Control Generalized to One-Parameter Splines

Ignacio Montesino, Alberto Jardón Huete

Robotic Intelligence

🎯 What it does: Extend Cartesian damping control to one-parameter spline curves, proposing a path damping control method for robotic upper limb rehabilitation.

CASRL: Collision Avoidance with Spiking Reinforcement Learning Among Dynamic, Decision-Making Agents

Chengjun Zhang, Huajin Tang

Spiking Neural NetworkTransformerReinforcement Learning

🎯 What it does: Proposed a collision avoidance model CASRL based on spiking reinforcement learning, using Spiking Neural Networks (SNN) as the actor, Deep Neural Networks (DNN) as the critic, and implementing processing of any number of dynamic decision-making agents through SpikeGTr transformer and ATF module

CaT: Constraints as Terminations for Legged Locomotion Reinforcement Learning

Elliot Chane-Sane, N. Mansard

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a reinforcement learning algorithm called CaT that treats constraints as random termination to achieve efficient compliance with hard constraints;

CathFlow: Self-Supervised Segmentation of Catheters in Interventional Ultrasound Using Optical Flow and Transformers

Alex Ranne, F. R. Y. Baena

SegmentationData SynthesisTransformerOptical FlowImageBiomedical DataUltrasound

🎯 What it does: Proposed a self-supervised deep learning architecture for segmenting ducts in longitudinal ultrasound images without requiring manual annotation data.

CATOA: Cooperative Calibration of Timestamp Measurements for Distributed Multi-Robot Localization

Feiyang Wen, Yuan Shen

OptimizationRobotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Designed and implemented a cooperative clock calibration method to reduce UWB positioning errors.

CBFkit: A Control Barrier Function Toolbox for Robotics Applications

Mitchell Black, Danil Prokhorov

Safty and PrivacyRobotic Intelligence

🎯 What it does: Developed and verified the CBFkit toolbox for achieving safe robot planning and control in uncertain environments.

CBGL: Fast Monte Carlo Passive Global Localisation of 2D LIDAR Sensor

Alexandros Filotheou

Autonomous DrivingComputational EfficiencySimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes a fast Monte Carlo passive global localization method for 2D LIDAR using Cumulative Absolute Error per Ray (CAER) and scan–to–map–scan matching.

Centroidal State Estimation Based on the Koopman Embedding for Dynamic Legged Locomotion

Shahram Khorshidi, Maren Bennewitz

Robotic Intelligence

🎯 What it does: Proposes a center of mass state estimation method based on Koopman operator embedding, validated through simulations under various dynamic gaits.

CFD-enabled Approach for Optimizing CPG Control Network for Underwater Soft Robotic Fish

Yunfei Wang, Juntian Qu

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: Propose an optimization method combining CFD with an improved CPG network, design a three-joint bionic soft robotic fish, and embed a CPG network based on the Hopf model into the control system. First, tune parameters on the CFD simulation platform, then validate on the robot.

CGA: Corridor Generating Algorithm for Multi-Agent Environments

Arseniy Pertzovsky, Roie Zivan

Autonomous DrivingOptimization

🎯 What it does: Study multi-agent path planning, propose the single-agent corridor generation (SACG) problem, and design two baseline algorithms and a new corridor generation algorithm (CGA), as well as its extension to lifelong MAPF.

Channel-wise Motion Features for Efficient Motion Segmentation

Riku Inoue, Yuji Yasui

SegmentationAutonomous DrivingConvolutional Neural NetworkOptical FlowImage

🎯 What it does: Propose a channel-wise motion feature representation based on cost volume, achieving efficient motion segmentation by extracting depth features for each instance in feature maps and capturing 3D motion information of the scene;

ChatMap: A Wearable Platform Based on the Multi-modal Foundation Model to Augment Spatial Cognition for People with Blindness and Low Vision

Yu Hao, Yi Fang

Explainability and InterpretabilityTransformerLarge Language ModelPrompt EngineeringVision Language ModelMultimodalityChain-of-Thought

🎯 What it does: Developed a wearable platform integrating a camera, audio module, and multi-modal foundation models such as GPT-4/GPT-4V to enhance spatial cognition for visually impaired individuals.

CLAT: Convolutional Local Attention Tracker for Real-time UAV Target Tracking System with Feedback Information

Xiaolou Sun, Meng Shen

Object TrackingConvolutional Neural NetworkVideo

🎯 What it does: Proposes a real-time UAV target tracking framework named CLAT (Convolutional Local Attention Tracker), combining a hierarchical convolutional local attention structure, a streamlined feature fusion network, and a redesigned upper controller to enhance tracking speed and control robustness.

Click to Grasp: Zero-Shot Precise Manipulation via Visual Diffusion Descriptors

Nikolaos Tsagkas, C. X. Lu

Pose EstimationRobotic IntelligenceDiffusion modelImageText

🎯 What it does: This paper utilizes a zero-shot learning framework based on text-to-image diffusion models, automatically inferring gripper poses by locating target parts through user clicks on source images to achieve precise manipulation;

Climbing Gait for a Snake Robot by Adapting to a Flexible Net

Kodai Yoshida, Motoyasu Tanaka

Robotic Intelligence

🎯 What it does: Propose a climbing gait in a variable grid environment, enabling the snake robot to adapt to grid deformation by tilting its body and adjusting the head position, achieving vertical and diagonal movement.

CLIPSwarm: Generating Drone Shows from Text Prompts with Vision-Language Models

Pablo Pueyo, Mac Schwager

GenerationVision Language ModelVideoText

🎯 What it does: Automatically generate drone formation performances based on natural language prompts.

Clutter-Aware Spill-Free Liquid Transport via Learned Dynamics

Ava Abderezaei, Alessandro Roncone

OptimizationRobotic IntelligenceTransformerPhysics Related

🎯 What it does: Propose an algorithm for handling liquid containers without spilling in crowded environments, allowing containers to tilt at larger angles and move along all axes to expand the reachable space and maneuverability.

CMR-Agent: Learning a Cross-Modal Agent for Iterative Image-to-Point Cloud Registration

Gongxin Yao, Yu Pan

Autonomous DrivingOptimizationReinforcement LearningImagePoint Cloud

🎯 What it does: Designed a cross-modal registration agent, CMRAgent, which models the registration from images to point clouds as an iterative Markov decision process using reinforcement learning and imitation learning, and proposes a 2D-3D hybrid state representation along with a one-time cross-modal embedding mechanism.

Co-RaL: Complementary Radar-Leg Odometry with 4-DoF Optimization and Rolling Contact

Sangwoo Jung, Ayoung Kim

OptimizationRobotic IntelligenceSimultaneous Localization and MappingMultimodality

🎯 What it does: Proposes a cooperative radar-leg odometry algorithm integrating chip radar with leg-based systems to achieve robust and accurate localization

Coalition Formation Game Approach for Task Allocation in Heterogeneous Multi-Robot Systems under Resource Constraints

Liwang Zhang, Shaowu Yang

OptimizationRobotic Intelligence

🎯 What it does: Model the multi-robot task allocation problem as a leader-follower coalition formation game and propose a corresponding coalition formation algorithm.

Coarse-to-Fine Detection of Multiple Seams for Robotic Welding

Pengkun Wei, Wei Zhang

Object DetectionRobotic IntelligenceImagePoint Cloud

🎯 What it does: Coarse-to-fine weld detection and extraction for multi-seam welding workpieces using RGB images and 3D point clouds

CoBL-Diffusion: Diffusion-Based Conditional Robot Planning in Dynamic Environments Using Control Barrier and Lyapunov Functions

Kazuki Mizuta, Karen Leung

Robotic IntelligenceDiffusion model

🎯 What it does: Proposed a safety-aware robotic planning method called CoBL-Diffusion based on diffusion models, which leverages control barrier functions and Lyapunov functions to guide the denoising process, generating safe and smooth control sequences in dynamic environments.

CoBOS: Constraint-Based Online Scheduler for Human-Robot Collaboration

Marina Ionova, Jan Kristof Behrens

OptimizationRobotic Intelligence

🎯 What it does: Proposed an online constraint scheduling framework CoBOS for human-robot collaborative assembly processes, enabling robots to dynamically adjust their behavior based on human actions and uncertain events.

CoDe : A Cooperative and Decentralized Collision Avoidance Algorithm for Small-Scale UAV Swarms Considering Energy Efficiency

Shuangyao Huang, Zhiyi Huang

Reinforcement Learning

🎯 What it does: Proposed a collaborative and decentralized collision avoidance algorithm named CoDe, targeting small-scale UAV formations with up to three UAVs, to improve energy efficiency and achieve effective collaboration.

CollabLoc: Collaborative Information Sharing for Real-Time Multiuser Visual Localization System

Teng-Te Yu, Kuan-Wen Chen

Pose EstimationComputational EfficiencyOptical FlowImage

🎯 What it does: Proposes CollabLoc, a system supporting multi-user real-time visual localization, which reduces computational overhead and improves efficiency and accuracy by leveraging collaborative information sharing.

Collaboration Strategies for Two Heterogeneous Pursuers in A Pursuit-Evasion Game Using Deep Reinforcement Learning

Zhanping Zhong, Jianping He

Reinforcement Learning

🎯 What it does: A pursuit-evasion game was studied in an unbounded 3D space, involving a flexible pursuer and a fast pursuer collaborating to capture a flexible evader. A hybrid strategy based on the Soft Actor-Critic framework was designed, enabling the pursuer to alter dynamics and decide when to switch to a faster but less maneuverable pursuer.

Collaborative Conversation in Safe Multimodal Human-Robot Collaboration

Davide Ferrari, Cristian Secchi

Safty and PrivacyRobotic IntelligenceWorld ModelMultimodality

🎯 What it does: Proposed a new architecture for achieving collaborative dialogue in secure multimodal human-robot collaboration, enabling operators and robots to communicate naturally and efficiently while handling safety issues.

Collaborative Object Manipulation on the Water Surface by a UAV-USV Team Using Tethers

Filip Novák, M. Saska

OptimizationRobotic Intelligence

🎯 What it does: Propose a method for collaboratively manipulating floating objects on water using UAV and USV through a rope.

Collision Detection between Smooth Convex Bodies via Riemannian Optimization Framework

Seoki An, Dongjun Lee

OptimizationMesh

🎯 What it does: Propose a method to convert the smooth convex body collision detection problem into an unconstrained Riemannian optimization problem, implemented based on double differentiable support functions and Riemannian trust region (RTR) methods.

Collision-Free Robot Navigation in Crowded Environments using Learning based Convex Model Predictive Control

Zhuanglei Wen, Xiai Chen

Robotic IntelligenceReinforcement LearningPoint Cloud

🎯 What it does: Propose a unified robot navigation framework that integrates perception, decision-making, and control, using convex obstacle-free regions computed from 2D LiDAR as observational inputs; achieve inherently collision-free reference point sampling by designing an action space based on kinematic constraints; employ MPC to track the reference trajectory to satisfy constraints.

Combining Ontological Knowledge and Large Language Model for User-Friendly Service Robots

Haru Nakajima, Jun Miura

Robotic IntelligenceTransformerLarge Language ModelText

🎯 What it does: Integrating ontology knowledge with large language models to improve the robot's 'give me' task, reduce ambiguity, and enhance system usability.