ICRA 2024 Papers — Page 2
IEEE International Conference on Robotics and Automation · 1760 papers
A three-dimensional compliant bowtie-shaped mechanical amplifier to magnify coaxial displacement in a confined space
Jintaek Im, Cheol Song
Physics Related
🎯 What it does: Proposed a three-dimensional coaxial butterfly-shaped mechanical amplifier that improves the Sarrus linkage structure using a lever mechanism to achieve parallel displacement amplification of the target plate along a single axis.
A Time-Optimal Energy Planner for Safe Human-Robot Collaboration
Andrea Pupa, Cristian Secchi
OptimizationSafty and PrivacyRobotic Intelligence
🎯 What it does: Proposed a trajectory planner that utilizes the variable inertia of robots to achieve time-optimal energy planning, enabling tasks to be completed at higher speeds while ensuring safety constraints.
A Track-based Colon Endoscopic Robot with Depth Perception Stereo Cameras for Haustral Fold Detection during Colonic Navigation
Shujing He, Chengzhi Hu
Depth EstimationRobotic IntelligenceImageBiomedical Data
🎯 What it does: A track-based stereoscopic disparity endoscopic robot (TSER) was developed for active navigation in the colon and detection of cecal folds.
A Trajectory-based Flight Assistive System for Novice Pilots in Drone Racing Scenario
Yuhang Zhong, Fei Gao
OptimizationRobotic Intelligence
🎯 What it does: Proposed a trajectory-based flying assistance system to help novice drone pilots achieve high-speed flight in racing scenarios; the system includes offline generation of globally optimal spatiotemporal trajectories and dense flight corridors, as well as online remote control mapping primitives, time-mapped trajectory progress, intent-aligned smooth safe trajectory planning, and optimal viewpoint heading planning; verified through simulation and field experiments, achieving a maximum flight speed of 6.0 m/s.
A transtibial prosthesis using a parallel spring mechanism
Donggyun Jung, Junho Choi
Biomedical Data
🎯 What it does: Proposed an electric ankle-foot prosthesis using parallel elastic actuators (PEA) and nonlinear spring mechanisms to generate the required joint torque for walking.
A Tube-Based Reinforcement Learning Approach for Optimal Motion Planning in Unknown Workspaces
Panagiotis Rousseas, Kostas J. Kyriakopoulos
OptimizationRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a pipeline-based approximate optimal motion planning method applicable to unknown workspaces, combining reactive motion planning with policy iteration reinforcement learning.
A Turning Radius Prediction Scheme for Sailing Robots under Complex Marine Environment
Weimin Qi, Huihuan Qian
Robotic IntelligenceTabular
🎯 What it does: Proposed a turning radius prediction scheme for sailboat robots that considers aerodynamic and hydrodynamic disturbances in the marine environment
A Two-step Nonlinear Factor Sparsification for Scalable Long-term SLAM Backend
Binqian Jiang, Shaojie Shen
Simultaneous Localization and Mapping
🎯 What it does: Proposes a two-step nonlinear factor sparsification process for a long-term SLAM backend targeting a general feature basis, aiming to limit the number of time-indexed poses while maintaining the impact of important landmarks.
A User-Centered Shared Control Scheme with Learning from Demonstration for Robotic Surgery
Haoyi Zheng, Etienne Burdet
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a user-centered shared control framework enabling robots to learn from expert demonstrations, predict operator intent, and naturally adjust control weights when necessary.
A Vision-Centric Approach for Static Map Element Annotation
Jiaxin Zhang, Wei Sui
Autonomous DrivingImage
🎯 What it does: Developed a LiDAR-free input, vision-centric method for annotating static map elements called CAMA, generating high-quality 3D annotations while ensuring spatiotemporal consistency.
A Wearable Robotic Hand for Hand-over-Hand Imitation Learning
Deh-Chang Wei, Huazhe Xu
Robotic Intelligence
🎯 What it does: Developed a wearable dexterous robotic hand, HIRO Hand, for hand-to-hand imitation learning, capable of collecting expert data and achieving grasping and in-hand manipulation; simultaneously designed non-learning and visual behavior cloning controllers.
Accelerating Long-Horizon Planning with Affordance-Directed Dynamic Grounding of Abstract Strategies
Khen Elimelech, L. Kavraki
Optimization
🎯 What it does: Proposed a planning framework enhanced with abstract strategies, which introduces strategy affordance to guide planning and policy induction, ultimately achieving a usability-oriented lazy search algorithm capable of seamlessly combining strategies and actions in long-horizon tasks to solve problems such as object reordering.
Accelerating Robotic Picking of Rigid Objects with a Compliant Pneumatic Gripper and an Impact-Aware Trajectory Plan
Frederik Ostyn, G. Crevecoeur
OptimizationRobotic Intelligence
🎯 What it does: Proposes a robotic grasping scheme that achieves higher-speed contact using compliant pneumatic grippers and improved trajectory planning without exceeding hardware limitations.
Accounting for Travel Time and Arrival Time Coordination During Task Allocations in Legged-Robot Teams
Shengqiang Chen, Satyandra K. Gupta
OptimizationRobotic Intelligence
🎯 What it does: Proposed a task allocation model based on Mixed Integer Linear Programming (MILP), considering the coordination of travel time and arrival time for legged robot teams to achieve efficient execution of collaborative tasks.
Accurate Kinematic Modeling using Autoencoders on Differentiable Joints
Nikolas J. Wilhelm, Maximilian Karl
Robotic IntelligenceAuto EncoderBiomedical Data
🎯 What it does: Developed a kinematic optimizer based on an autoencoder, utilizing a neural network to simulate inverse kinematics, encoding measurement data into joint parameters, and decoding through a differentiable forward kinematic model, validated on knee and hand experimental data.
Accurate Prior-centric Monocular Positioning with Offline LiDAR Fusion
Jinhao He, Ming Liu
Pose EstimationAutonomous DrivingSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: Utilizing a monocular camera combined with LiDAR-enhanced visual prior maps, and employing deep learning-based visual feature tracking to achieve centimeter-level global localization
Achieving Autonomous Cloth Manipulation with Optimal Control via Differentiable Physics-Aware Regularization and Safety Constraints
Yutong Zhang, Michael C. Yip
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: Proposed a differentiable fabric dynamics modeling framework that combines physics-aware regularization terms and safety constraints, achieving autonomous fabric manipulation through gradient optimization;
Achieving Mechanical Transparency Using Fusion Hybrid Linear Actuator for Shoulder Flexion and Extension in Exoskeleton Robot
Takuma Shimoyama, Yoshihiro Nakata
Robotic Intelligence
🎯 What it does: A hybrid linear actuator integrating pneumatic actuation and electromagnetic force compensation was developed for the motion of the shoulder joint in upper limb exoskeletons, generating force through pneumatic pressure and compensating friction via electromagnetic force to achieve mechanical transparency;
Acoustic Soft Tactile Skin (AST Skin)
Vishnu Rajendran S, A. E.
ClassificationRobotic IntelligenceAudio
🎯 What it does: Developed and evaluated a novel acoustic flexible tactile skin (AST Skin) that utilizes sound waves to detect touch pressure and contact position.
Acoustically Driven Micropipette for Hydrodynamic Manipulation of Mouse Oocytes
Zhaofeng Zuo, T. Arai
Biomedical DataUltrasound
🎯 What it does: Developed a non-contact hydrodynamic manipulation method based on sound waves and micropipettes, capable of precisely rotating and transporting mouse oocytes.
AcTExplore: Active Tactile Exploration on Unknown Objects
A. Shahidzadeh, Y. Aloimonos
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a reinforcement learning-based active tactile exploration method called AcTExplore, which can automatically explore unknown object surfaces within limited steps, collect tactile data, and perform 3D shape reconstruction.
ActFormer: Scalable Collaborative Perception via Active Queries
Suozhi Huang, Chen Feng
Object DetectionAutonomous DrivingTransformerImage
🎯 What it does: Propose a Transformer architecture named ActFormer, which uses active querying to select relevant cameras and learns bird's-eye view (BEV) representations to achieve scalable collaborative perception.
Action Segmentation Using 2D Skeleton Heatmaps and Multi-Modality Fusion
S. Hyder, Quoc-Huy Tran
SegmentationConvolutional Neural NetworkMultimodality
🎯 What it does: Propose an action segmentation method based on 2D skeleton heatmap, and perform multi-modal fusion between 2D skeleton heatmap and RGB videos.
Action-By-Detection: Efficient Forklift Action Detection for Autonomous Mobile Robots in Warehouses
Alexander Prutsch, Horst Bischof
Object DetectionAutonomous DrivingComputational EfficiencyConvolutional Neural NetworkImage
🎯 What it does: A method is proposed that infers forklift actions by utilizing a shallow image classification model through bird's-eye view scene cropping. The method projects 3D object detection results onto a context map, learns map contextual information, and aggregates time series information without requiring object tracking, offering efficiency and ease of deployment on embedded AMR hardware.
Active Automotive Augmented Reality Displays using Reinforcement Learning
Jungjae Ryu, Seong-Woo Kim
Autonomous DrivingReinforcement Learning
🎯 What it does: Studied and addressed the vertical mismatch problem in automotive augmented reality (AR) displays caused by uneven road surfaces, verifying that reinforcement learning-based control methods can significantly reduce mismatch and improve visibility
Active Collision-Based Navigation for Wheeled Robots
Jingjing Li, Fei Gao
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: This paper studies methods for active navigation and localization using collision information when a robot's external sensors fail, demonstrating the positive role of collisions in improving the localization accuracy of intrinsic sensors.
Active Exploration for Real-Time Haptic Training
Jake Ketchum, Todd D. Murphey
Data-Centric LearningRobotic Intelligence
🎯 What it does: Using active learning and coverage-based ergodic controllers, training a tactile perception model in a near-real-time environment by utilizing the entropy of sensor state variables conditioned as a measure of uncertainty.
Active Implicit Reconstruction Using One-Shot View Planning
Hao Hu, Maren Bennewitz
RestorationOptimizationPoint Cloud
🎯 What it does: Proposes a method that combines implicit representations with one-shot view planning (OSVP), using a deep neural network to directly predict a set of views for object reconstruction, filling missing surfaces through implicit representations rather than acquiring additional views.
Active Inference for Reactive Temporal Logic Motion Planning
Ziyang Chen, Zheng Kan
Robotic Intelligence
🎯 What it does: Developed a real-time decision-making and motion planning framework that enables robots to make proactive decisions and automatically generate temporal logic specifications for local reactive tasks when encountering dynamic events during the execution of globally planned tasks from offline planning;
Active Learning with Dual Model Predictive Path-Integral Control for Interaction-Aware Autonomous Highway On-ramp Merging
Jacob W. Knaup, P. Tsiotras
Autonomous Driving
🎯 What it does: Proposes a dual control framework based on Model Predictive Path-Integral Control (MPCPI), combined with Bayesian inference for active learning of other drivers' model parameters, to generate interactive merging trajectories;
Active Visual Localization for Multi-Agent Collaboration: A Data-Driven Approach
M. Hanlon, Hermann Blum
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Studied the challenge of overcoming viewpoint changes in multi-robot or human-robot collaboration scenarios through active visual localization methods, focusing on selecting the optimal observation perspective at a given location, and comparing existing methods with a newly proposed data-driven approach.
Active-Perceptive Motion Generation for Mobile Manipulation
Snehal Jauhri, Georgia Chalvatzaki
OptimizationRobotic Intelligence
🎯 What it does: Proposed ActPerMoMa, an activity-aware pipeline that generates informative motion paths for mobile operating systems in uncertain and cluttered environments; validated the method on TIAGo++ dual-arm mobile grasping experiments and demonstrated its ability to transfer to real-world scenarios.
Active, Quasi-Passive, Pneumatic, and Portable Knee Exoskeleton with Bidirectional Energy Flow for Efficient Air Recovery in Sit-Stand Tasks
L. Mišković, T. Petrič
OptimizationRobotic Intelligence
🎯 What it does: Designed and implemented a portable pneumatic knee exoskeleton that provides energy output for standing actions in active mode and absorbs and stores energy for sitting actions in quasi-passive mode; bidirectional energy flow achieves energy recovery and reuse.
Actor-Critic Model Predictive Control
Angel Romero, D. Scaramuzza
Reinforcement Learning
🎯 What it does: This paper proposes and implements an Actor-Critic model predictive control (AC-MPC) framework, integrating differentiable MPC into the actor-critic reinforcement learning (RL) structure to achieve real-time control and complex behavior learning;
AD4RL: Autonomous Driving Benchmarks for Offline Reinforcement Learning with Value-based Dataset
Dongsu Lee, Minhae Kwon
Autonomous DrivingReinforcement LearningBenchmark
🎯 What it does: Provides an autonomous driving dataset and benchmark for offline reinforcement learning;
Ada-Tracker: Soft Tissue Tracking via Inter-Frame and Adaptive-template Matching
Jiaxin Guo, Yun-hui Liu
Object TrackingOptical FlowVideoBiomedical DataBenchmark
🎯 What it does: Propose a soft tissue tracking method called Ada-Tracker, which uses optical flow for adjacent frame matching to obtain a rough ROI, and dynamically updates the tracking template based on estimated reliability through adaptive template matching.
AdaptAUG: Adaptive Data Augmentation Framework for Multi-Agent Reinforcement Learning
Xin Yu, Rongye Shi
Reinforcement Learning
🎯 What it does: Developed an adaptive data augmentation framework called AdaptAUG to improve sample efficiency and overall performance of multi-agent reinforcement learning (MARL) in multi-robot systems.
Adapting for Calibration Disturbances: A Neural Uncalibrated Visual Servoing Policy
Hongxiang Yu, Rong Xiong
Robotic Intelligence
🎯 What it does: Proposes a neural network uncalibrated visual servoing strategy (NUVS) that can adapt to camera calibration disturbances through adaptation mechanisms and control-oriented approaches.
Adapting to the “Open World”: The Utility of Hybrid Hierarchical Reinforcement Learning and Symbolic Planning
Pierrick Lorang, Matthias Scheutz
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed an enhanced hybrid system with nested hierarchical action abstraction to rapidly adapt to unknown events in open-world robotic tasks.
Adaptive Contact-Implicit Model Predictive Control with Online Residual Learning
Wei-Cheng Huang, Michael Posa
OptimizationRobotic Intelligence
🎯 What it does: Proposed a real-time adaptive multi-contact model predictive control framework capable of online learning the residual of the hybrid model and real-time improving control performance.
Adaptive Control for Triadic Human-Robot-FES Collaboration in Gait Rehabilitation: A Pilot Study
Andreas Christou, S. Vijayakumar
Robotic IntelligenceBiomedical Data
🎯 What it does: Designed an adaptive hybrid robot-FES controller to achieve tripartite collaboration among the patient, robot, and FES
Adaptive Gait Modeling and Optimization for Principally Kinematic Systems
Siming Deng, Brian A. Bittner
Optimization
🎯 What it does: Use an adaptive system identification framework to enhance the performance of the dominant motion device, demonstrating its adaptability and iterative improvement capabilities across diverse terrains within a behavior optimization framework; particularly for the nine-link Purcell swimmer, optimizing the gait over approximately 10 segment cycles achieved a tenfold improvement in optimization speed.
Adaptive Haptic Control Interface for Safeguarding Robotic Teleoperation in Hazardous Steelmaking Environments
Jaehyun Park, Keehoon Kim
Robotic Intelligence
🎯 What it does: For the task of removing iron blocks in the hazardous environment of a steel plant, two novel tactile control interfaces (POstick-KF and POstick-VF) were designed and implemented, and their effectiveness was verified through user experiments.
Adaptive Landmark Color for AUV Docking in Visually Dynamic Environments
Corey Knutson, Junaed Sattar
Autonomous DrivingRobotic IntelligenceImage
🎯 What it does: Proposed a vision-based adaptive LED color marking and dynamic color filtering method to improve the visibility between AUV and docking stations under different water quality conditions
Adaptive Model Predictive Control with Data-driven Error Model for Quadrupedal Locomotion
Xuanqi Zeng, Yunhui Liu
OptimizationRobotic IntelligenceTime Series
🎯 What it does: Integrating a data-driven error model into traditional MPC to enhance quadruped robot locomotion performance.
Adaptive Motion Scaling for Robot-Assisted Microsurgery Based on Hybrid Offline Reinforcement Learning and Damping Control
Peiyang Jiang, Dandan Zhang
Robotic IntelligenceReinforcement LearningBiomedical Data
🎯 What it does: Developed an adaptive motion scaling method based on hybrid offline reinforcement learning and damping control for teleoperation in robot-assisted minimally invasive surgery;
Adaptive Outlier Thresholding for Bundle Adjustment in Visual SLAM
Alejandro Fontan, Michael Milford
OptimizationSimultaneous Localization and Mapping
🎯 What it does: Propose a distribution-based online outlier rejection method for bundle adjustment in visual SLAM;
Adaptive Passive Biped Dynamic Walking on Unknown Uneven Terrain
Lishen Pu, Chunquan Xu
OptimizationRobotic Intelligence
🎯 What it does: An adaptive controller is proposed for dynamic walking of a virtual passive biped robot on unknown uneven terrain, with its effectiveness verified through simulation.
Adaptive Pedestrian Agent Modeling for Scenario-based Testing of Autonomous Vehicles through Behavior Retargeting
G. Muktadir, Jim Whitehead
Autonomous Driving
🎯 What it does: Proposes a new pedestrian crossing scene representation and a hybrid modeling method called RePed, which can transfer micro-behavior models to macro trajectories, enabling diversified human crossing actions in real-world scenarios through controllable modeling-based perturbation-driven scenario expansion; meanwhile, this representation is based on the Ego vehicle coordinate system and logical road structure, supporting scenario redirection for different roads, traffic conditions, and Ego vehicle behaviors, thereby enhancing scenario-based testing.
Adaptive Planning and Control with Time-Varying Tire Models for Autonomous Racing Using Extreme Learning Machine
Dvij Kalaria, J. Dolan
Autonomous DrivingOptimization
🎯 What it does: By real-time data collection, a tire model is constructed and adaptively updated to estimate friction, and based on this, the optimal track is selected from a pre-generated track library to achieve adaptive speed planning.
Adaptive State Estimation with Constant-Curvature Dynamics Using Force-Torque Sensors with Application to a Soft Pneumatic Actuator
Maximilian Mehl, Moritz Schappler
Robotic IntelligenceTime SeriesPhysics Related
🎯 What it does: This paper studies state estimation using a torque sensor located at the base of a soft pneumatic actuator, achieving real-time estimation through an unscented Kalman filter with unconstrained planar or quasi-static motion, and verifying it via Cosserat rod simulations and actual physical systems.
Adaptive Whole-body Robotic Tool-use Learning on Low-rigidity Plastic-made Humanoids Using Vision and Tactile Sensors
Kento Kawaharazuka, Masayuki Inaba
Robotic IntelligenceImageMultimodality
🎯 What it does: Proposes a method that utilizes neural networks to model the relationships between joint angles, visual information, and foot tactile information, in order to control the tool tip position of the low-stiffness plastic humanoid robot KXR during tool usage.
Advancements in 3D Lane Detection Using LiDAR Point Clouds: From Data Collection to Model Development
Runkai Zhao, Weidong (Tom) Cai
Object DetectionAutonomous DrivingPoint Cloud
🎯 What it does: Propose the LiSV-3DLane large-scale 3D lane dataset and the LiLaDet 3D lane detection model, leveraging the geometric features and spatial attributes of LiDAR point clouds to achieve automatic annotation and detection;
Advancing Virtual Reality Interaction: A Ring-Shaped Controller and Pose Tracking
Zhuqing Zhang, Yue Wang
Pose EstimationOptical FlowImage
🎯 What it does: Proposes a robust tracking algorithm based on an innovative wearable ring-shaped controller (equipped with IMU and LED), utilizing various visual measurements (6DoF, 5DoF pose, 3DoF position and orientation, and 2DoF image) to achieve precise tracking of controller motion;
AdvGPS: Adversarial GPS for Multi-Agent Perception Attack
Jinlong Li, Hongkai Yu
Object DetectionAutonomous DrivingAdversarial AttackPoint Cloud
🎯 What it does: Proposed the ADVGPS method to generate stealthy GPS adversarial signals, attacking multi-agent perception systems and significantly reducing target detection accuracy.
Aerial Image-based Inter-day Registration for Precision Agriculture
Chen Gao, L. Teixeira
TransformerImageAgriculture Related
🎯 What it does: Developed a UAV image-to-image daily registration method that uses a ground control point (GCP) only once at the beginning of the season for precision agriculture analysis;
Aerial Interaction with Tactile Sensing
Xiaofeng Guo, Sebastian Scherer
Robotic Intelligence
🎯 What it does: Achieved hybrid motion-force control with tactile feedback and wall texture detection in fully actuated UAV dynamic manipulation tasks using visual tactile sensors.
Aerial Tensile Perching and Disentangling Mechanism for Long-Term Environmental Monitoring
Tian Lan, B. B. Kocer
Robotic Intelligence
🎯 What it does: Proposed a multi-modal drone system that utilizes tension-based shutdown and suspended driving cabin to achieve long-term environmental monitoring.
AeroDima: Cheetah-Inspired Aerodynamic Tail Design for Rapid Maneuverability
D. Bright, Amir Patel
Robotic Intelligence
🎯 What it does: Designed and verified a lightweight sail tail (AeroDima) that utilizes aerodynamic drag to generate stable torque, enhancing the robot's maneuverability during high-speed turns.
AG-Cvg: Coverage Planning with a Mobile Recharging UGV and an Energy-Constrained UAV
Nare Karapetyan, Pratap Tokekar
Autonomous DrivingRobotic Intelligence
🎯 What it does: This paper proposes a coverage path planning method for energy-constrained unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs).
AgriSORT: A Simple Online Real-time Tracking-by-Detection framework for robotics in precision agriculture
Leonardo Saraceni, Thomas Alessandro Ciarfuglia
Object TrackingVideoAgriculture Related
🎯 What it does: Propose AgriSORT, a simple, online, real-time multi-object tracking-by-detection framework based on motion information, specifically designed for precision agriculture robots.
AGRNav: Efficient and Energy-Saving Autonomous Navigation for Air-Ground Robots in Occlusion-Prone Environments
Junming Wang, Heming Cui
OptimizationRobotic IntelligenceTransformer
🎯 What it does: Proposed the AGRNav framework for aerial-ground robots to find safe and energy-efficient hybrid paths in occlusion-prone environments.
AHPPEBot: Autonomous Robot for Tomato Harvesting based on Phenotyping and Pose Estimation
Xingxu Li, Siyi Zheng
Object DetectionPose EstimationConvolutional Neural NetworkAgriculture Related
🎯 What it does: Designed and implemented an autonomous tomato harvesting robot named AHPPEBot based on crop phenotyping and pose estimation
AiAReSeg: Catheter Detection and Segmentation in Interventional Ultrasound using Transformers
Alex Ranne, F. R. Y. Baena
SegmentationData SynthesisTransformerImageBiomedical DataUltrasound
🎯 What it does: Proposes a state-of-the-art architecture based on Transformer for detecting and segmenting catheters in axial interventional ultrasound image sequences.
Air Bumper: A Collision Detection and Reaction Framework for Autonomous MAV Navigation
Ruoyu Wang, Ben M. Chen
Robotic IntelligenceTime Series
🎯 What it does: Proposes the Air Bumper collision detection and response framework, achieving collision detection and estimation using only IMU, and designs rapid recovery control and collision-aware mapping to enable safer autonomous flight for drones in 3D environments.
AirExo: Low-Cost Exoskeletons for Learning Whole-Arm Manipulation in the Wild
Hongjie Fang, Cewu Lu
Data-Centric LearningRobotic IntelligenceVideo
🎯 What it does: Developed a low-cost, adaptable, portable dual-arm exoskeleton called AirExo for remote control operations and demonstration collection, utilizing it to conduct large-scale low-cost demonstration data collection in the wild to train robots for full-arm manipulation.
AirFisheye Dataset: A Multi-Model Fisheye Dataset for UAV Applications
Pravin Kumar Jaisawal, Volker Gollnick
SegmentationData SynthesisDepth EstimationImagePoint CloudBenchmark
🎯 What it does: Proposed and publicly released the AirFisheye multi-model fisheye dataset, applicable to drone segmentation, depth estimation, depth completion tasks, while providing a generic framework for synthetic fisheye image generation and an occlusion correction algorithm.
Aligning Knowledge Graph with Visual Perception for Object-goal Navigation
Nuo Xu, Chao Li
Robotic IntelligenceGraph Neural NetworkVision Language ModelMultimodalityGraphBenchmark
🎯 What it does: Propose a method that aligns knowledge graphs with visual perception for object goal navigation, constructing a continuous hierarchical scene architecture and aligning language with vision.
ALPHA: Attention-based Long-horizon Pathfinding in Highly-structured Areas
Chengyang He, Guillaume Sartoretti
OptimizationTransformer
🎯 What it does: Proposed a new framework called ALPHA for multi-agent path planning, which combines real local information with fuzzy global information and allows agents to make short-term predictions about each other's paths to improve overall collaboration and decision quality.
Amodal Optical Flow
Maximilian Luz, Abhinav Valada
TransformerOptical FlowImage
🎯 What it does: Proposed the Amodal Optical Flow task, achieving unified modeling of visible and occluded regions by expanding the AmodalSynthDrive dataset and defining a multi-layer pixel-level motion field.
Amortized Inference for Efficient Grasp Model Adaptation
Michael Noseworthy, Nicholas Roy
Domain AdaptationRobotic Intelligence
🎯 What it does: Proposes an encoder/decoder action feasibility model that estimates unobserved attributes of new objects through interaction, enabling efficient adaptation.
AMSwarmX: Safe Swarm Coordination in CompleX Environments via Implicit Non-Convex Decomposition of the Obstacle-Free Space
V. K. Adajania, Angela P. Schoellig
OptimizationRobotic Intelligence
🎯 What it does: Propose a safety coordination method for UAV swarms based on alternating minimization, which does not rely on conservative free space approximations, uniformly handles static and dynamic collision constraints, and achieves implicit non-convex free space decomposition through Octomap distance queries.
An adaptable ankle trajectory generation method for lower-limb exoskeletons by means of safety constraints computation and minimum jerk planning
Raffaele Giannattasio, L. Michieli
OptimizationRobotic Intelligence
🎯 What it does: A method is proposed to calculate smooth ankle trajectories for lower-limb exoskeletons using a fourth-order polynomial function, which computes trajectories in real-time according to different safety constraints and user-defined gait parameters;
An Analytic Solution to the 3D CSC Dubins Path Problem
Victor M. Baez, Aaron T. Becker
OptimizationRobotic Intelligence
🎯 What it does: Propose an analytical solution for 3D Dubins CSC paths and convert it into an inverse kinematics problem for an RRPRR robotic arm
An Augmented Catenary Model for Underwater Tethered Robots
Martin Filliung, V. Hugel
Robotic IntelligencePhysics Related
🎯 What it does: An enhanced catenary model is introduced to account for the impact of hydrodynamic damping on the shape of underwater cables, and its effectiveness is validated through experiments.
An Efficient Model-Based Approach on Learning Agile Motor Skills without Reinforcement
Hao-bin Shi, Max Q.-H. Meng
Computational EfficiencyRobotic IntelligenceAuto EncoderWorld Model
🎯 What it does: Propose a model-based efficient learning framework that combines differentiable world models with a Variational Autoencoder (VAE)-based policy network to imitate real animal behavior, reducing the demand for real interaction data.
An Efficient Solution to the 2D Visibility Problem in Cartesian Grid Maps and its Application in Heuristic Path Planning
Ibrahim Ibrahim, Jan Swevers
OptimizationComputational EfficiencyMesh
🎯 What it does: Propose a lightweight method that evaluates the visibility existence from the source point to all grid points with a single traversal on a 2D grid without preprocessing, with computational and memory complexity of O(n).
An Electromagnetism-Inspired Method for Estimating In-Grasp Torque from Visuotactile Sensors
Yuni Fuchioka, Masashi Hamaya
Robotic IntelligenceMultimodalityPhysics Related
🎯 What it does: Proposed and verified a learning-free, mechanical- and optics-model-free method for analyzing two-dimensional marker displacement vector fields based on tactile dipole moments, to estimate tilt torque during grasping from gel-based visuo-tactile sensors.
An Environmental-Complexity-Based Navigation Method Based on Hierarchical Deep Reinforcement Learning
Pengbin Chen, Shuaikang Ma
Autonomous DrivingReinforcement LearningPoint Cloud
🎯 What it does: Proposes two metrics for quantifying environmental complexity based on laser scanning, and constructs a hierarchical deep reinforcement learning navigation method based on these metrics.
An Equivariant Approach to Robust State Estimation for the ArduPilot Autopilot System
A. Fornasier, Stephan Weiss
Autonomous DrivingTime Series
🎯 What it does: Propose a new equivariant filter (EqF) to achieve inertial navigation system state estimation for the ArduPilot autopilot system, and verify its robustness under common challenges using real-world data.
An Extrinsic Calibration Method between LiDAR and GNSS/INS for Autonomous Driving
Jiahao Pi, Botian Shi
Autonomous DrivingPoint Cloud
🎯 What it does: A three-stage extrinsic parameter calibration method between LiDAR and GNSS/INS was developed, enabling rapid and accurate calibration of the relative attitude between sensors.
An Image Acquisition Scheme for Visual Odometry based on Image Bracketing and Online Attribute Control
Shuyang Zhang, Ming Liu
Pose EstimationImage
🎯 What it does: Proposes an image acquisition scheme based on image intervalization and online attribute control, continuously capturing images at different exposure levels and adjusting exposure online to enhance the robustness and accuracy of visual odometry.
An Integrated Position-velocity-force Method for Safety-enhanced Shared Control in Robot-assisted Surgical Cutting
Xilin Xiao, Hangjie Mo
Safty and PrivacyRobotic Intelligence
🎯 What it does: A safety-enhanced human-robot shared control method was developed, which intelligently allocates control authority while ensuring the surgeon remains in dominance; precise tracking and efficient assistance are achieved by designing a master manipulator position controller, master manipulator velocity controller, and planned trajectory tracking controller; a motion fusion mechanism is adopted to optimize the evaluation function based on future states for control mode switching, and a force feedback mechanism is proposed to help humans understand autonomous control intentions.
An Intelligent Robotic Endoscope Control System Based on Fusing Natural Language Processing and Vision Models
Beili Dong, George P. Mylonas
Object DetectionRobotic IntelligenceConvolutional Neural NetworkVision-Language-Action ModelMultimodality
🎯 What it does: Proposed a hybrid robotic endoscope control system based on the fusion of natural language processing and an improved YOLO-V8 visual model.
An Intuitive Manual Guidance Scheme to Operate Rotation and Translation Simultaneously
Fan Shao, Fanny Ficuciello
OptimizationRobotic Intelligence
🎯 What it does: Proposed a virtual fixture space guidance framework that achieves intuitive synchronized control of rotation and translation
An Investigation of Multi-feature Extraction and Super-resolution with Fast Microphone Arrays
Eric T. Chang, M. Ciocarlie
ClassificationTransformerTime Series
🎯 What it does: Using MEMS microphone arrays as vibration sensors for texture classification and estimation of contact position and velocity.
An Iterative Approach for Heterogeneous Multi-Agent Route Planning with Temporal Logic Goals and Travel Duration Uncertainty
Kaier Liang, C. Vasile
OptimizationGraph
🎯 What it does: Proposes an iterative method for probabilistic constraint-based path planning on weighted graphs with uncertain travel times for heterogeneous multi-agents, aiming to maximize specification robustness, minimize travel time, and maximize success probability.
An LLM-driven Framework for Multiple-Vehicle Dispatching and Navigation in Smart City Landscapes
Ruiqing Chen, Jun Wang
Autonomous DrivingOptimizationTransformerLarge Language ModelReinforcement Learning
🎯 What it does: Proposes LiMeda, a multi-vehicle scheduling and navigation framework based on large language models (LLMs), for coordinating and navigating multi-functional vehicles in smart cities.
An NMPC Framework for Tracking and Releasing a Cable-suspended Load to a Ground Target Using a Multirotor UAV
Fotis Panetsos, Kostas J. Kyriakopoulos
OptimizationRobotic IntelligenceImage
🎯 What it does: A nonlinear model predictive control (NMPC) framework is proposed for tracking ground targets and releasing loads when suspended by a multirotor drone.
An offline learning of behavior correction policy for vision-based robotic manipulation
Qingxiu Dong, Masashi Sugiyama
Robotic IntelligenceReinforcement LearningImage
🎯 What it does: Propose a two-stage agent for visual robotic manipulation tasks, which first makes preliminary action decisions from raw camera images and then uses cropped images to correct actions, achieving offline learning with small datasets.
An Onboard Framework for Staircases Modeling Based on Point Clouds
Chun Qing, Gan Ma
Pose EstimationAutonomous DrivingPoint Cloud
🎯 What it does: Developed a vehicle-mounted framework for detecting navigable areas and physical property modeling of stairs based on point cloud data.
An Online Self-calibrating Refractive Camera Model with Application to Underwater Odometry
Mohit Singh, Kostas Alexis
Robotic IntelligenceSimultaneous Localization and MappingPhysics Related
🎯 What it does: Proposed a camera model applicable to refractive media (e.g., water), achieving real-time self-calibration without requiring known correspondences or calibration targets, and subsequently applied it to underwater visual-inertial odometry; experimental results verified the model's effectiveness in estimating refractive index, correcting distortion, and enhancing robustness of odometry, while demonstrating capabilities in media conversion and online refractive index estimation.
An Open and Flexible Robot Perception Framework for Mobile Manipulation Tasks
Patrick Mania, Michael Beetz
Object DetectionPose EstimationRobotic Intelligence
🎯 What it does: Proposed the RoboKudo framework, which supports perception processing in mobile manipulation tasks and can combine multiple visual algorithms as needed;
An Open-Source Solution for Fast and Accurate Underwater Mapping with a Low-Cost Mechanical Scanning Sonar
Tim Hansen, Andreas Birk
Robotic IntelligenceSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Propose an open-source software framework to realize real-time underwater mapping based on BlueROV2, Ping360 mechanical scanning sonar (MSS), and A50 Doppler Velocity Log (DVL) on Raspberry Pi 4.
Analysis and Validation of Stiffness and Payload of Nematode-Inspired Cable Routing Method for Cable Driven Redundant Manipulator
Hoyoung Kim, Jungwon Yoon
Robotic Intelligence
🎯 What it does: This paper calculates the equivalent stiffness of the nematode-inspired cable routing method, derives and simulates its kinematics and effective inverse kinematics algorithms, and verifies the stiffness and load capacity through a prototype machine.
Analyzing Accessibility in Robot-Assisted Vitreoretinal Surgery: Integrating Eye Posture and Robot Position
Satoshi Inagaki, M. Nasseri
Robotic Intelligence
🎯 What it does: Studied and proposed strategies for expanding robot reachability and visibility by integrating eye movements with robot posture/position adjustments to optimize reachability in retinal surgery;
Angler: An Autonomy Framework for Intervention Tasks with Lightweight Underwater Vehicle Manipulator Systems
Evan Palmer, Geoffrey A. Hollinger
Robotic Intelligence
🎯 What it does: Developed the Angler software framework to support localization, control, and decision-making algorithms for lightweight underwater manipulator systems (UVMS), achieving simulation-to-reality transfer; verified the framework's performance in stabilization and waypoint tracking tasks.
Anisotropic body compliance facilitates robotic sidewinding in complex environments
Velin Kojouharov, D. Goldman
Robotic Intelligence
🎯 What it does: Designed and implemented a legless robot with a decentralized bilateral cable-driven system, achieving lateral and vertical body swinging for lateral locomotion through programmable anisotropic body compliance; the authors also developed a feedforward controller that integrates body compliance into a lateral gait template, and validated its performance in experiments.
Ankle Exoskeleton with a Symmetric 3 DoF Structure for Plantarflexion Assistance
Miha Dezman, Tamim Asfour
Robotic Intelligence
🎯 What it does: Designed and evaluated an ankle exoskeleton with a symmetric parallel frame structure, capable of achieving three degrees of freedom motion, and realizing ankle plantar flexion assistance through cable transmission.
Anticipate & Act: Integrating LLMs and Classical Planning for Efficient Task Execution in Household Environments†
Raghav Arora, Madhava Krishna
OptimizationComputational EfficiencyRobotic IntelligenceTransformerLarge Language ModelPrompt Engineering
🎯 What it does: Designed a framework that integrates large language models (LLM) with classical planning, leveraging the LLM's ability to perform high-level task prediction with minimal prompts, then using the predicted results as input goals for the classical planning system to generate fine-grained action sequences that complete multiple household tasks in one go.