IROS 2024 Papers — Page 15
IEEE/RSJ International Conference on Intelligent Robots and Systems · 1581 papers
SwiftBase: A Ddataset based on High-Frequency Visual Measurement for Visual-Inertial Localization in High-Speed Motion Scenes
Zhenghao Zou, Haochen Chai
Autonomous DrivingSimultaneous Localization and MappingImageTime SeriesBenchmark
🎯 What it does: Created the SwiftBase dataset for visual-inertial localization research in high-speed motion scenarios.
SwiftEagle: An Advanced Open-Source, Miniaturized FPGA UAS Platform with Dual DVS/Frame Camera for Cutting-Edge Low-Latency Autonomous Algorithms
Christian Vogt, Michele Magno
Autonomous DrivingComputational EfficiencyRobotic IntelligenceImageVideo
🎯 What it does: Proposed and implemented the open-source FPGA UAS platform named SwiftEagle, equipped with dual cameras (RGB/DVS) and developed software/firmware stacks for high-precision ML dataset recording and FPGA-side rendering of DVS frames.
SwinMTL: A Shared Architecture for Simultaneous Depth Estimation and Semantic Segmentation from Monocular Camera Images
P. Taghavi, Gaurav Pandey
SegmentationDepth EstimationGenerative Adversarial NetworkImage
🎯 What it does: Proposes a multi-task learning framework that enables monocular cameras to perform simultaneous depth estimation and semantic segmentation.
Swiss DINO: Efficient and Versatile Vision Framework for On-device Personal Object Search
Kirill Paramonov, Mete Ozay
RetrievalComputational EfficiencyTransformerContrastive LearningImage
🎯 What it does: Proposes Swiss DINO, a single-sample personal object search framework based on the DINOv2 Transformer, designed for personalized scene understanding on devices without requiring adaptation training.
Switching Sampling Space of Model Predictive Path-Integral Controller to Balance Efficiency and Safety in 4WIDS Vehicle Navigation
Mizuho Aoki, Tatsuya Suzuki
Autonomous DrivingOptimization
🎯 What it does: Implementing navigation for a four-wheel independently driven vehicle to avoid collisions and reach the target point among obstacles of arbitrary shapes using the MPPI control algorithm
Synergistic Reinforcement and Imitation Learning for Vision-driven Autonomous Flight of UAV Along River
Zihan Wang, N. Mahmoudian
Robotic IntelligenceReinforcement Learning
🎯 What it does: Develop a trainable, realistic dynamics-free river simulation environment and propose a framework that synergistically combines reinforcement learning (RL) and imitation learning (IL) for vision-driven autonomous flight and obstacle avoidance of drones in river environments.
Synergizing Morphological Computation and Generative Design: Automatic Synthesis of Tendon-Driven Grippers
Kirill Zharkov, S. Kolyubin
Robotic Intelligence
🎯 What it does: Proposes a method for automatically generating robotic linkages using context-free grammars and heuristic search, and applies it to design a tendon-driven, low-torque gripper.
TacLink-Integrated Robot Arm toward Safe Human-Robot Interaction
Q. Luu, Van Anh Ho
Domain AdaptationSafty and PrivacyRobotic IntelligenceMultimodality
🎯 What it does: Developed and tested the integration of TacLink soft tactile sensing connection into a robotic arm to achieve safe interaction control during collisions.
Tactile Active Inference Reinforcement Learning for Efficient Robotic Manipulation Skill Acquisition
Zihao Liu, Panfeng Huang
Robotic IntelligenceReinforcement LearningMultimodality
🎯 What it does: Proposed the TactileAIRL framework for efficient robot manipulation skill learning through tactile perception, and demonstrated its effectiveness in simulation and physical screw grasping tasks.
Tactile Comfort: Lowering Heart Rate Through Interactions with a Pocket Robot
Morten Roed Frederiksen, Maja J. Mataric
Robotic IntelligenceTime SeriesBiomedical DataElectrocardiogram
🎯 What it does: Developed a portable tactile companion robot and measured heart rate changes when the robot was used versus not used through two experiments (a 14-day pilot study and a main experiment with 18 children).
Tactile Odometry in Aerial Physical Interaction
M. Schuster, S. Hamaza
Pose EstimationRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose a novel aerial tactile odometry method that estimates the drone's attitude by using a coupled end-effector and trackball tactile sensor for contact.
Targeted Image Transformation for Improving Robustness in Long Range Aircraft Detection
Rebecca Martin, S. Scherer
Object DetectionImage
🎯 What it does: Propose an orientation-aware image transformation method to enhance the robustness of distant aircraft detection
Task and Domain Adaptive Reinforcement Learning for Robot Control
Y. Liu, Aamir Ahmad
Domain AdaptationRobotic IntelligenceReinforcement Learning
🎯 What it does: Propose an adaptive agent based on transfer learning that can dynamically adjust strategies to adapt to different tasks and environmental conditions, and verify its multi-task and environmental adaptability in the balloon control challenge.
Task Planning for Long-Horizon Cooking Tasks Based on Large Language Models
Jungkyoo Shin, Eunwoo Kim
TransformerLarge Language ModelText
🎯 What it does: Proposed a long-term cooking task planning framework based on large language models (LLMs), which leverages LLMs to extract text semantic features and combines them with Transformer encoding-decoding to generate object-centered subgoal sequences.
Task-Based Design and Policy Co-Optimization for Tendon-driven Underactuated Kinematic Chains
Sharfin Islam, M. Ciocarlie
OptimizationRobotic IntelligenceReinforcement Learning
🎯 What it does: Implemented a task-driven design and co-optimization method for underactuated tendon-driven joint chains, and performed co-optimization on a three-joint, two-actuator chain using reinforcement learning.
Task-Driven Computational Framework for Simultaneously Optimizing Design and Mounted Pose of Modular Reconfigurable Manipulators
Maolin Lei, N. Tsagarakis
OptimizationRobotic Intelligence
🎯 What it does: Proposed a computational framework capable of simultaneously optimizing the pose and structure of modular reconfigurable robotic arms;
Task-Driven Manipulation with Reconfigurable Parallel Robots
Daniel Morton, Marco Pavone
OptimizationRobotic Intelligence
🎯 What it does: Developed the ReachBot platform with a scalable mast as a limb, enabling reconfiguration of the parallel robot through detachable and repositionable masts to facilitate tool usage, sample handling, and transportation in challenging environments, while designing a hybrid integer pose planner and convex tension planner to maximize the task torque space and account for uncertainties in micro-claw grasping.
Task-Oriented Design Method for Monolithic Flexible Hands with Wire Drive Systems
Rina Kusuhara, Mitsuru Higashimori
Robotic Intelligence
🎯 What it does: A task-oriented design method for a single flexible gripper is proposed, and line coordination equations and posture coordination equations are derived through analytical modeling. Their structural isomorphism is utilized to analytically determine mechanical parameters and line tension, with the feasibility of the method ultimately validated through case studies and experiments.
Task-Oriented Dexterous Hand Pose Synthesis Using Differentiable Grasp Wrench Boundary Estimator
Jiayi Chen, He Wang
GenerationPose EstimationOptimizationRobotic Intelligence
🎯 What it does: Designed a unified framework for synthesizing task-oriented dexterous hand poses that can generate hand postures suitable for specific task torque spaces.
Task-Space Riccati Feedback based Whole Body Control for Underactuated Legged Locomotion
Shunpeng Yang, Hua Chen
OptimizationRobotic Intelligence
🎯 What it does: A whole-body control method based on task space Riccati feedback is proposed for legged robots with unactuated joints. The method constructs an LQR using linearized unactuated dynamics and solves the Riccati gain, replacing manually tuned floating base task feedback gains.
TD-NeRF: Novel Truncated Depth Prior for Joint Camera Pose and Neural Radiance Field Optimization
Zhen Tan, Dewen Hu
Pose EstimationDepth EstimationOptimizationNeural Radiance Field
🎯 What it does: Propose a TD-NeRF method based on truncated depth prior, achieving training from unknown camera poses through joint optimization of camera poses and NeRF.
Teaching Robots Where To Go And How To Act With Human Sketches via Spatial Diagrammatic Instructions
Qilin Sun, Matthew Johnson-Roberson
OptimizationRobotic IntelligenceVision-Language-Action ModelImage
🎯 What it does: Propose Spatial Diagrammatic Instructions (SDIs) by annotating spatial regions on camera images with hand-drawn sketches, projecting them onto three-dimensional coordinates, and learning continuous Spatial Instruction Maps (SIMs). Subsequently, SIMs are incorporated as constraints into an optimization problem to address the base placement issue for mobile manipulators.
Team Coordination on Graphs: Problem, Analysis, and Algorithms
Manshi Limbu, Xuesu Xiao
OptimizationGraph
🎯 What it does: Redefined the Team Collaboration Graph Problem (TCGRE) as a constrained optimization problem, conducted rigorous mathematical analysis, and proposed three categories of algorithms (JSG-based, Collaborative-based, and Recursive Window Subteam-based) to solve this problem.
TeFF: Tracking-enhanced Forgetting-free Few-shot 3D LiDAR Semantic Segmentation
Junbao Zhou, Yu Hu
Object TrackingSegmentationPoint Cloud
🎯 What it does: This paper utilizes a tracking model to generate pseudo labels for data augmentation and combines LoRA to achieve few-shot 3D LiDAR semantic segmentation without forgetting.
TempBEV: Improving Learned BEV Encoders with Combined Image and BEV Space Temporal Aggregation
T. Monninger, Steffen Staab
Object DetectionSegmentationAutonomous DrivingOptical FlowImagePoint Cloud
🎯 What it does: Propose a new temporal BEV encoder, TempBEV, which combines temporal information aggregation in the image space and BEV latent space, and treats consecutive image frames as temporal stereo, utilizing optical flow estimation for temporal stereo encoding.
Temporal Attention for Cross-View Sequential Image Localization
Dong Yuan, Feras Dayoub
Autonomous DrivingTransformerImageSequential
🎯 What it does: Proposed a new method using the Temporal Attention Module (TAM) to achieve cross-view sequential image localization
Temporal- and Viewpoint-Invariant Registration for Under-Canopy Footage using Deep-Learning-based Bird’s-Eye View Prediction
Jiawei Zhou, L. Teixeira
Image TranslationConvolutional Neural NetworkImagePoint CloudAgriculture Related
🎯 What it does: Proposed a cross-seasonal and cross-day under-canopy image sequence registration method that combines GPS and deep learning to generate bird's-eye views for estimating tree positions and achieving registration.
Tension Feedback Control for Musculoskeletal Quadrupedal Locomotion over Uneven Terrain
Hiroaki Tanaka, Koh Hosoda
Robotic Intelligence
🎯 What it does: Quadruped gait control on uneven terrain using PAM tension, developing a durable tension sensor and implementing tension feedback control.
Ternary-Type Opacity and Hybrid Odometry for RGB NeRF-SLAM
Junru Lin, D. Paudel
Pose EstimationAutonomous DrivingNeural Radiance FieldSimultaneous Localization and MappingImage
🎯 What it does: Propose a method to apply NeRF to SLAM using only RGB input, addressing scenarios without depth information; achieve more accurate depth rendering and image warping by introducing a ternary opacity model (TT) and hybrid odometry (HO); ultimately attain state-of-the-art performance with superior speed and accuracy.
Terrain-Attentive Learning for Efficient 6-DoF Kinodynamic Modeling on Vertically Challenging Terrain
A. Datar, Xuesu Xiao
Autonomous DrivingRobotic Intelligence
🎯 What it does: Proposes a 6-DoF dynamics learning method that focuses on critical terrain features to achieve efficient real-time motion planning on vertical challenging terrain.
Test-Time Certifiable Self-Supervision to Bridge the Sim2Real Gap in Event-Based Satellite Pose Estimation
A. M. Jawaid, Tat-Jun Chin
Pose EstimationDomain AdaptationOptimizationPoint Cloud
🎯 What it does: Proposes a test-time self-supervised method under event sensor environments, incorporating a verifier module to bridge the domain gap between simulated and real data.
Text-to-Drive: Diverse Driving Behavior Synthesis via Large Language Models
Phat Nguyen, Daniela Rus
Data SynthesisAutonomous DrivingTransformerLarge Language ModelTextSequential
🎯 What it does: Proposes the Text-to-Drive (T2D) method for synthesizing diverse driving behaviors through large language models (LLMs), employing a two-phase knowledge-driven process: first generating diverse language descriptions using LLMs, then synthesizing behaviors in simulation.
Text2Map: From Navigational Instructions to Graph-Based Indoor Map Representations Using LLMs
Ammar Karkour, Eduardo Feo Flushing
GenerationData SynthesisTransformerLarge Language ModelPrompt EngineeringTextGraph
🎯 What it does: Leverages large language models and few-shot learning to convert natural language navigation instructions into graph-based indoor maps, simplifying the map generation process.
Text3DAug - Prompted Instance Augmentation for LiDAR Perception
Laurenz Reichardt, Oliver Wasenmüller
Data SynthesisPrompt EngineeringPoint Cloud
🎯 What it does: Proposed a LiDAR instance enhancement method based on text generation called Text3DAug
The Control Strategy for Vehicle Transfer Robots in RO/RO Terminal Environments
Zhi Liu, Junzheng Wang
Object DetectionPose EstimationOptimizationRobotic IntelligencePoint Cloud
🎯 What it does: Proposes a new control framework for bipedal hurdle-crossing vehicle transfer robots, including point cloud segmentation processing, traversal-based point cloud matrix fitting for target pose estimation, and a docking controller based on real-time object detection.
The design of a sensorized laryngoscope training system for pediatric intubation
Ningzhe Hou, P. Maiolino
Biomedical Data
🎯 What it does: Designed and verified a perceptual laryngoscope training system equipped with a force-torque sensor, 9-axis inertial measurement unit, and tactile sensor to provide online feedback on angle, force, and grip.
The Design of the Barkour Benchmark for Robot Agility
Wenhao Yu, Kuang-Huei Lee
Robotic IntelligenceBenchmark
🎯 What it does: Proposed and designed the Barkour robot agility benchmark, providing a set of complex obstacle courses to quantify the agility performance of robots across multiple platforms
The Effectiveness of State Representation Model in Multi-Agent Proximal Policy Optimization for Multi-Agent Path Finding
Jaehoon Chung, Homayoun Najjaran
OptimizationReinforcement LearningWorld Model
🎯 What it does: Integrate a state representation model into the multi-agent Proximal Policy Optimization (PPO) framework, leveraging features extracted from the global map to enhance the training effectiveness of multi-agent path planning.
The Power of Input: Benchmarking Zero-Shot Sim-to-Real Transfer of Reinforcement Learning Control Policies for Quadrotor Control
Alberto Dionigi, Giuseppe Loianno
Domain AdaptationRobotic IntelligenceReinforcement LearningBenchmark
🎯 What it does: Benchmark evaluation of deep reinforcement learning (DRL) drone control strategies under different observation space configurations, optimizing multiple agents and studying their robustness and zero-shot transfer capability from simulation to real-world environments.
The Power of the Senses: Generalizable Manipulation from Vision and Touch through Masked Multimodal Learning
Carmelo Sferrazza, Pieter Abbeel
Representation LearningRobotic IntelligenceReinforcement LearningAuto EncoderContrastive LearningMultimodality
🎯 What it does: Proposed and evaluated a new multi-modal learning framework called Masked Multimodal Learning (M3L) for jointly learning visual and tactile representations and policies in reinforcement learning environments, further verifying its effectiveness in three simulated environments (robotic insertion, door opening, and in-hand multi-finger manipulation).
The subtle line between personalization and user manipulation in a European regulatory perspective. A proposal for a technology-assessment methodology for Artificial Intelligence Systems *
Andrea Bertolini
🎯 What it does: Proposes a technical evaluation method aimed at balancing personalization and user control, assessing AI systems under the European regulatory framework
Theoretical Modeling and Bio-inspired Trajectory Optimization of A Multiple-locomotion Origami Robot
Keqi Zhu, Huixu Dong
OptimizationRobotic Intelligence
🎯 What it does: Proposes mathematical models for crawling and swimming of multi-modal origami robots, and designs a swimming kinematic model through coordinate transformation, as well as an algorithm for systematic planning of human-like swimming gaits with maximum thrust.
Thermal Ablation Therapy Control with Tissue Necrosis-driven Temperature Feedback Enabled by Neural State Space Model with Extended Kalman Filter
Ryo Murakami, Haichong K. Zhang
🎯 What it does: Proposed a temperature estimation technique based on the progression of tissue necrosis for temperature feedback control in thermal ablation therapy.
Thermal-NeRF: Neural Radiance Fields from an Infrared Camera
Tian Ye, Ling Pei
GenerationNeural Radiance FieldImage
🎯 What it does: Developed the Thermal-NeRF method, which generates a NeRF volume field representation using only infrared images, and employs thermal mapping and structural thermal constraints to enhance recovery quality in low-light/visually impaired scenarios.
Thermally-Resilient Soft Gripper for On-Orbit Operations
F. Ruiz, A. Ollero
Robotic IntelligencePhysics Related
🎯 What it does: Conducted experimental validation of a soft gripper in rail operations, completed thermal dynamic analysis, designed a multi-layer experimental prototype using TPU, silicone, PTFE, and aerogel, and tested its grasping performance under temperature variations in the laboratory using liquid nitrogen and a heat gun.
This is the Way: Mitigating the Roll of an Autonomous Uncrewed Surface Vessel in Wavy Conditions Using Model Predictive Control *
D. L. Jenkins, Joshua A. Marshall
Autonomous Driving
🎯 What it does: Developed a nonlinear model predictive control (NMPC) system to reduce the average roll angle of unmanned surface vessels (USVs) under wave conditions;
Three-Dimensional Vehicle Dynamics State Estimation for High-Speed Race Cars under varying Signal Quality
Sven Goblirsch, Johannes Betz
Autonomous Driving
🎯 What it does: Proposed and implemented a three-dimensional vehicle dynamics state estimation method under varying signal quality conditions, taking into account road inclination and slope, and validated its deployment on a high-speed track.
Tightly Coupled Passive UWB Localization for Low-density Anchor Networks
Nushen M. Senevirathna, R. Gosine
Simultaneous Localization and MappingTime Series
🎯 What it does: Proposed and evaluated a passive tightly coupled UWB inertial navigation system for indoor localization on mobile platforms.
Tightly-Coupled Factor Graph Formulation For Radar-Inertial Odometry
J. Michalczyk, Stephan Weiss
OptimizationSimultaneous Localization and Mapping
🎯 What it does: Propose a radar-inertial odometry (RIO) method based on factor graph nonlinear optimization, implemented within a sliding window framework.
Time-Optimal Path Parameterization for Cooperative Multi-Arm Robotic Systems with Third-Order Constraints
Maximilian Dio, Andreas Völz
OptimizationRobotic Intelligence
🎯 What it does: Proposed a time-optimal path parameterization (TOPP) method for collaborative multi-arm robot systems (MARS), considering third-order constraints involving impact, torque rate, and torque rate limits.
Time-Optimal TCP and Robot Base Placement for Pick-and-Place Tasks in Highly Constrained Environments
Alexander Wachter, C. Hartl-Nesic
OptimizationRobotic Intelligence
🎯 What it does: In a highly constrained environment, simultaneous optimization of the robot base and Tool Center Point (TCP) within a robotic workcell is performed to achieve the shortest cycle time for a series of pick-and-place tasks.
Time-Ordered Ad-hoc Resource Sharing for Independent Robotic Agents
Arjo Chakravarty, M. R. Elara
OptimizationRobotic Intelligence
🎯 What it does: Proposed a resource sharing method based on Boolean satisfiability (SAT), presented an algorithm to convert restricted allocation into weighted-SAT optimization, introduced a theorem that enables solving optimal resource allocation problems by repeatedly invoking a SAT solver, and demonstrated a method to encode continuous time ordering constraints using CNF; benchmarked the performance of this method in non-fixed collaboration settings and validated its feasibility on simulated and real robot fleets.
Time-varying Control Barrier Function for Safe and Precise Landing of a UAV on a Moving Target
V. N. Sankaranarayanan, G. Nikolakopoulos
OptimizationRobotic Intelligence
🎯 What it does: Proposed a safe and precise landing control scheme for UAVs on moving targets based on time-varying control barrier functions (CBF)
TinyLidarNet: 2D LiDAR-based End-to-End Deep Learning Model for F1TENTH Autonomous Racing
Mohammed Misbah Zarrar, Heechul Yun
Autonomous DrivingComputational EfficiencyConvolutional Neural NetworkPoint Cloud
🎯 What it does: Proposed and evaluated a lightweight 2D LiDAR end-to-end deep learning model called TinyLidarNet for F1TENTH autonomous racing.
TivNe-SLAM: Dynamic Mapping and Tracking via Time-Varying Neural Radiance Fields
Chengyao Duan, Zhiliu Yang
Autonomous DrivingNeural Radiance FieldSimultaneous Localization and Mapping
🎯 What it does: Proposes a time-varying representation for simultaneously tracking and reconstructing dynamic scenes under a self-supervised framework; maintains tracking and mapping processes in parallel, uses motion masks to focus on dynamic pixels; introduces a two-stage parameter optimization, mapping time to 3D positions to obtain a canonical scene, then mapping to the canonical scene's embedding to acquire color and SDF; and employs a keyframe selection strategy based on overlap rate.
To Help or Not to Help: LLM-based Attentive Support for Human-Robot Group Interactions
Daniel Tanneberg, M. Gienger
Robotic IntelligenceLarge Language Model
🎯 What it does: Proposes the concept of Attentive Support, integrating scene perception, dialogue acquisition, contextual understanding, behavior generation, and combining the commonsense reasoning of large language models (LLMs), to provide non-intrusive physical support for human-robot group interaction.
ToolEENet: Tool Affordance 6D Pose Estimation
Yunlong Wang, Jianwei Zhang
Pose EstimationDiffusion modelImage
🎯 What it does: Developed the ToolEENet framework, which utilizes RGB-D data to segment tool end-effectors and achieve 6D pose estimation, and released the first TOOLEE dataset containing tool end-effector affinity segmentation and 6D pose based on usage scenarios.
TopoNav: Topological Navigation for Efficient Exploration in Sparse Reward Environments
Jumman Hossain, Theron Trout
Autonomous DrivingReinforcement LearningSimultaneous Localization and Mapping
🎯 What it does: Proposed the TopoNav framework, achieving efficient goal-oriented exploration and navigation in sparse reward environments.
TOPPQuad: Dynamically-Feasible Time-Optimal Path Parametrization for Quadrotors
Katherine Mao, Vijay Kumar
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: Proposes the TOPPQuad algorithm, which performs time-optimal parameterization of a given collision-free geometric path without violating quadrotor actuation and state constraints.
Torque Ripple Reduction in Quasi-Direct Drive Motors Through Angle-Based Repetitive Learning Observer and Model Predictive Torque Controller
Hefei Zhang, Shiwu Zhang
OptimizationRobotic Intelligence
🎯 What it does: Propose a scheme combining the angle-based repetitive learning observer (ARLO) with a model predictive control field-oriented controller (MPC-FOC) to reduce torque ripple in quasi-direct drive (QDD) motors.
Touch-GS: Visual-Tactile Supervised 3D Gaussian Splatting
Aiden Swann, Monroe Kennedy
Depth EstimationGaussian SplattingImageMultimodalityPoint Cloud
🎯 What it does: Proposes a method that utilizes an optical tactile sensor to supervise 3D Gaussian Splatting scene reconstruction, integrating tactile and visual depth information to build a unified object representation.
Toward An Analytic Theory of Intrinsic Robustness for Dexterous Grasping
Albert H. Li, Aaron D. Ames
Robotic Intelligence
🎯 What it does: Proposed a novel analytical theory to evaluate intrinsic robustness under object pose or geometric uncertainty in grasp planning, and validated its effectiveness through hardware experiments.
Toward Control of Wheeled Humanoid Robots with Unknown Payloads: Equilibrium Point Estimation via Real-to-Sim Adaptation
D. Baek, João Ramos
Domain AdaptationOptimizationRobotic Intelligence
🎯 What it does: Proposed a model-based control framework for predicting the balance points of wheeled humanoid robots under unknown loads; estimated total mass and center of gravity via the system's response to unknown dynamics, utilizing precise nonlinear dynamic models with real-simulation adaptation, optimized parameters through particle swarm optimization (PSO), and validated on a physical wheeled inverted pendulum.
Toward Micro Eye Movement Detection in Practice: Stand-alone Eye Tracker with High Resolution and Wide Measurement Range
Keiko Yokoyama, Masatoshi Ishikawa
Video
🎯 What it does: Developed an independently usable eye-tracking system for detecting micro-movements of the eyes;
Toward Perpetual Occlusion-Aware Observation of Comb States in Living Honeybee Colonies
Jan Blaha, Tomáš Krajník
Robotic IntelligenceVideoAgriculture Related
🎯 What it does: In the RoboRoyale project, a robot was built and operated to continuously track the queen bee using a camera and monitor the state of the beehive, particularly areas of the hive frequently obstructed by worker bees; meanwhile, multiple mapping methods combining previous observations and predicted worker bee density were evaluated.
Toward Precise Robotic Weed Flaming Using a Mobile Manipulator with a Blowtorch
Di Wang, Dezhen Song
Robotic IntelligenceImageAgriculture Related
🎯 What it does: Built a mobile manipulator system equipped with a blowtorch and proposed an algorithm for real-time detection and prediction of flame coverage area;
Toward Understanding Key Estimation in Learning Robust Humanoid Locomotion
Zhicheng Wang, Qiuguo Zhu
OptimizationRobotic Intelligence
🎯 What it does: Studied the impact of different types of state estimation on the decision-making of learning-based humanoid robot control strategies, and utilized significance analysis to quantitatively evaluate the effectiveness of learned state estimation, identify key estimation variables, and optimize their combinations to enhance the robustness of humanoid locomotion tasks.
Toward Universal and Scalable Road Graph Partitioning for Efficient Multi-Robot Path Planning
Xingyao Han, Hesheng Wang
Autonomous DrivingOptimizationGraph
🎯 What it does: Proposes a generic, scalable road map partitioning method that automatically divides any real-world environment into multiple regions, converting global path planning into distributed sub-region path planning and inter-regional strategies.
Towards a Factor Graph-Based Method using Angular Rates for Full Magnetometer Calibration and Gyroscope Bias Estimation
Sebastián Rodríguez-Martínez, G. Troni
Time SeriesPhysics Related
🎯 What it does: Proposed a factor graph-based magnetometer calibration and gyroscope bias estimation method (MAGYC) that utilizes triaxial angular rate measurements to improve the accuracy of batch and online calibration.
Towards a Surgeon-in-the-Loop Ophthalmic Robotic Apprentice using Reinforcement and Imitation Learning
Amr Gomaa, Antonio Krüger
Robotic IntelligenceReinforcement LearningImageBiomedical Data
🎯 What it does: Developed an image-based surgeon-centered autonomous robotic assistant, utilizing both reinforcement learning and imitation learning for simultaneous training, specifically for autonomous operation during the incision phase of cataract surgery in ophthalmology.
Towards Accurate And Robust Dynamics and Reward Modeling for Model-Based Offline Inverse Reinforcement Learning
Gengyu Zhang, Yan Yan
Robotic IntelligenceReinforcement LearningDiffusion modelScore-based Model
🎯 What it does: Enhance model-based offline inverse reinforcement learning by improving the conservative MDP framework and utilizing score-based diffusion generative models to improve dynamics and reward modeling.
Towards Cross-View-Consistent Self-Supervised Surround Depth Estimation
Laiyan Ding, Rui Huang
Depth EstimationAutonomous DrivingConvolutional Neural NetworkImage
🎯 What it does: Proposed a cross-view consistent self-supervised surround view depth estimation method, designed an efficient pose estimation scheme using only the front view, and introduced dense depth consistency loss, multi-view reconstruction consistency loss, and flip augmentation techniques.
Towards Design and Development of a Soft Pressure Sensing Sleeve for Performing Safe Colonoscopic Procedures
Mohammad Rafiee Javazm, F. Alambeigi
Safty and PrivacyBiomedical Data
🎯 What it does: Designed and developed a soft pressure sensing sleeve (SPSS) that seamlessly integrates into existing colonoscopy devices, utilizing resistance changes in microchannels filled with liquid metal (gallium) to detect pressure and position contact between the colonoscope and the colon surface.
Towards Designing a Low-Cost Humanoid Robot with Flex Sensors-Based Movement*
Muhammad H. Al Omoush, Tracey Mehigan
Robotic Intelligence
🎯 What it does: Designed and built a low-cost humanoid robot using flexible sensors to achieve the motion mechanism, and tested its functionality and efficiency in an educational environment at a school in Dubai.
Towards Dynamic and Small Objects Refinement for Unsupervised Domain Adaptative Nighttime Semantic Segmentation
Jingyi Pan, Lin Wang
SegmentationDomain AdaptationContrastive LearningImageBenchmark
🎯 What it does: Proposes a new unsupervised domain adaptation method that refines dynamic and small objects in nighttime semantic segmentation at both the label and feature levels.
Towards Electricity-free Pneumatic Miniature Rotation Actuator for Optical Coherence Tomography Endoscopy
Tinghua Zhang, Wu Yuan
Biomedical Data
🎯 What it does: Designed and verified an electricity- and magnet-free pneumatic micro-rotary actuator for optical coherence tomography (OCT) endoscopy, demonstrating its feasibility in finger imaging.
Towards Enhanced Fairness and Sample Efficiency in Traffic Signal Control
Xingshuai Huang, Benoit Boulet
Autonomous DrivingReinforcement LearningWorld Model
🎯 What it does: Propose FM2Light, a fairness-aware model-based multi-agent reinforcement learning framework for traffic signal control.
Towards Human-Centered Construction Robotics: A Reinforcement Learning-Driven Companion Robot for Contextually Assisting Carpentry Workers
Yuning Wu, Daniel Cardoso Llach
Robotic IntelligenceReinforcement Learning
🎯 What it does: Deployed a 'work companion' robot in carpentry formwork construction, demonstrating a modular framework prototype based on context-aware reinforcement learning to enhance safety and workflow smoothness.
Towards intelligent robotic sole deburring: from burrs identification to path planning
Alessandra Tafuro, Paolo Rocco
SegmentationOptimizationRobotic IntelligenceConvolutional Neural NetworkImage
🎯 What it does: Developed an autonomous path planning pipeline for shoe sole deburring, comprising a vision-based segmentation and burr detection module, as well as a path planning module obtained through learning from demonstration (LfD).
Towards Kbps-level Vehicle Teleoperation via Persistent-Transient Environment Modelling
Chunyang Zhao, Danwei Wang
Autonomous DrivingWorld Model
🎯 What it does: Propose a non-video vehicle remote operation framework based on persistent-transient environment modeling for remote control of autonomous vehicles in low-bandwidth environments.
Towards Long Term SLAM on Thermal Imagery
Colin Keil, H. Singh
Simultaneous Localization and MappingImage
🎯 What it does: Developed a learning-based feature thermal imaging visual SLAM baseline system, demonstrating good tracking and day-night relocalization performance on thermal imaging.
Towards Robotised Palpation for Cancer Detection through Online Tissue Viscoelastic Characterisation with a Collaborative Robotic Arm
Luca Beber, Luigi Palopoli
Robotic IntelligenceBiomedical Data
🎯 What it does: Propose an online palpation method using a collaborative robot arm to estimate the penetration depth of the end-effector and the viscoelastic properties of soft tissues.
Towards the New Generation of Smart Home-Care with Cloud-Based Internet of Humans and Robotic Things
Dandan Zhang, Jin Zheng
Robotic Intelligence
🎯 What it does: Proposed and implemented a cloud-based Internet of Humans and Robotic Things (IoHRT) framework integrating human-robot collaboration control mechanisms to enhance the efficiency and quality of home care services.
Towards Unconstrained Collision Injury Protection Data Sets: Initial Surrogate Experiments for the Human Hand
R. Kirschner, Sami Haddadin
Data SynthesisBiomedical Data
🎯 What it does: Proposed an experimental scheme using two pendulums and a robot to simulate unconstrained collisions between a human hand and sharp objects, and constructed an initial injury database using pig feet as in vitro substitute samples.
Tracking Control with Uncertainty Smoothing Estimation under Aggressive Maneuvers of Aerial Vehicles
Hao Zhang, Zhi Zheng
Physics RelatedOrdinary Differential Equation
🎯 What it does: Designed a high dynamic tracking control framework to achieve precise tracking of aggressive trajectories with speeds up to 15 m/s and accelerations of 2g.
Tracking Tumors under Deformation from Partial Point Clouds using Occupancy Networks
P. Henrich, Axel Krieger
Object TrackingPoint CloudBiomedical Data
🎯 What it does: Propose a method based on the occupancy network for quickly locating tumors from local point clouds when the kidney model deforms.
Trajectory Optimization with Global Yaw Parameterization for Field-of-View Constrained Autonomous Flight
Yuwei Wu, Vijay Kumar
OptimizationRobotic Intelligence
🎯 What it does: Proposes a global yaw parameterization method for joint yaw and position optimization of quadrotor trajectories under limited field-of-view sensor conditions, implementing the method in various application scenarios with simulation and field experiment evaluations.
Trajectory Planning for Non-Prehensile Object Transportation
Lingyun Chen, Sami Haddadin
OptimizationRobotic Intelligence
🎯 What it does: Proposes two trajectory planning methods for non-grasping transport using sampling and dynamic programming algorithms, tested on transportation tasks involving unstable objects with the 7-DoF Franka Emika robot.
Trans-Rotor: An Active Omnidirectional Aerial-Ground Vehicle With Differential Gear Joint Transformation Mechanism
Xuankang Wu, Zheng Fang
Robotic IntelligenceBenchmark
🎯 What it does: Designed and implemented a fully omnidirectional aerial-ground vehicle named TransRotor, introducing a differential gear joint, four-wheel steering model, and intermediate mode transition mechanism to achieve omnidirectional movement in both air and ground modes with rapid switching; a decoupled controller was also developed to enable autonomous navigation.
Transcrib3D: 3D Referring Expression Resolution through Large Language Models
Jiading Fang, Matthew R. Walter
Robotic IntelligenceTransformerLarge Language ModelSupervised Fine-TuningVision-Language-Action ModelText
🎯 What it does: Propose the Transcrib3D method, integrating 3D detection technology with large language models (LLM), using text as a unified medium to solve 3D reference expression parsing problems, and enabling actual robotic grasping and placement tasks.
Transformer-based Multi-Agent Reinforcement Learning for Generalization of Heterogeneous Multi-Robot Cooperation
Yuxin Cai, Chen Lv
Robotic IntelligenceGraph Neural NetworkTransformerReinforcement Learning
🎯 What it does: Proposed a transformer-based multi-agent reinforcement learning method for heterogeneous multi-robot collaboration.
Transformer-Based Relationship Inference Model for Household Object Organization by Integrating Graph Topology and Ontology
Xiaodong Li, Yu Gu
Robotic IntelligenceGraph Neural NetworkTransformerLarge Language ModelGraph
🎯 What it does: By constructing a dataset containing ontological attributes and relationships of household items, and using a feature extraction method that combines Graph Attention Network (GAT) with BERT, training under the Transformer framework to achieve relationship inference between items, thereby improving the systematic organization of household items by service robots.
Translating Agent-Environment Interactions from Humans to Robots
Tanmay Shankar, Jean Oh
Robotic IntelligenceAgentic AI
🎯 What it does: Proposed the TransAct framework, which first learns and abstracts temporal representations of common agent-environment interactions, then converts human demonstration sequences on unknown tasks into executable robot interaction strategies, achieving zero-shot task execution.
Transparency evaluation for the Kinematic Design of the Harnesses through Human-Exoskeleton Interaction Modeling
Riccardo Bezzini, Alessandro Filippeschi
OptimizationRobotic Intelligence
🎯 What it does: Proposes a configuration comparison method for wearable leg exoskeletons based on a flexible simulation tool, which evaluates various connection mechanisms, kinematic, and drive characteristics through dynamic interaction between the simulation setup and the user; the evaluation criteria are based on minimizing interaction torque through an optimization process, while considering the similarity between exoskeleton joint trajectories and user joint movements.
TRAVERSE: Traffic-Responsive Autonomous Vehicle Experience & Rare-event Simulation for Enhanced safety
Sandeep Thalapanane, Ming C. Lin
Autonomous Driving
🎯 What it does: By allowing participants to drive in a VR vehicle simulator and experience various accident scenarios, the study investigates human responses and behaviors in simulated accidents.
Tree-Based Reconfiguration of Metamorphic Robots
Patrick Ondika, Jiri Barnat
OptimizationRobotic Intelligence
🎯 What it does: Proposes a reconfiguration algorithm for a chain-type deformable robot that utilizes the formation of 'tentacles' and transports modules through them.
TriHelper: Zero-Shot Object Navigation with Dynamic Assistance
Lingfeng Zhang, Renjing Xu
Robotic Intelligence
🎯 What it does: Propose three auxiliary components, TriHelper, which dynamically assist robots in completing zero-shot target navigation tasks such as collision avoidance, exploration, and target detection
TriLoc-NetVLAD: Enhancing Long-term Place Recognition in Orchards with a Novel LiDAR-Based Approach
Na Sun, Chunjiang Zhao
RetrievalPoint CloudAgriculture Related
🎯 What it does: Proposed and implemented the TriLoc-NetVLAD LiDAR long-term location recognition method to address the localization challenges caused by high-frequency repeated features in orchards.
TřiVis: Versatile, Reliable, and High-Performance Tool for Computing Visibility in Polygonal Environments
Jan Mikula, L. Přeučil
MeshBenchmark
🎯 What it does: Developed the TřiVis library for computing visibility-related queries in complex polygonal environments.
TrustNavGPT: Modeling Uncertainty to Improve Trustworthiness of Audio-Guided LLM-Based Robot Navigation
Xingpeng Sun, Aniket Bera
Safty and PrivacyRobotic IntelligenceTransformerLarge Language ModelAgentic AIAudio
🎯 What it does: Propose TrustNavGPT, an audio-guided navigation agent based on LLM, which evaluates the credibility of human instructions by leveraging emotional cues in speech (tone, emphasis) and makes safe decisions.