ICRA 2025 Papers — Page 2
IEEE International Conference on Robotics and Automation · 1604 papers
Adaptive Emotional Expression in Social Robots: A Multimodal Approach to Dynamic Emotion Modeling
Haeun Park, Hui-Sung Lee
Robotic IntelligenceMultimodality
🎯 What it does: Proposed a framework enabling robots to naturally express emotions in a multimodal manner, and implemented a mobile robot prototype capable of being recognized through touch and expressing different emotions via facial expressions and movements.
Adaptive Energy Regularization for Autonomous Gait Transition and Energy-Efficient Quadruped Locomotion
Boyuan Liang, Masayoshi Tomizuka
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed and verified the incorporation of an energy-efficient reward based on average energy consumption per distance in a quadruped robot reinforcement learning framework, enabling the robot to adaptively select appropriate gaits;
Adaptive Grasping of Moving Objects in Dense Clutter via Global-to-Local Detection and Static-to-Dynamic Planning
Hao Chen, Kensuke Harada
OptimizationRobotic IntelligenceImage
🎯 What it does: This paper proposes an adaptive grasping method for novel objects in dense cluttered environments using a monocular RGB-D camera, employing global-to-local visual detection and grasping planning from static to dynamic scenes;
Adaptive Perching and Grasping by Aerial Robot with Light-Weight and High Grip-Force Tendon-Driven Three-Fingered Hand Using Single Actuator
Hisaaki Iida, Moju Zhao
Robotic Intelligence
🎯 What it does: This paper designs a lightweight three-fingered hand and develops an adaptive grasping and hover control method driven by a single actuator, subsequently verifying its feasibility through load, grasping, and aerial hover experiments.
Adaptive Task Allocation in Multi-Human Multi-Robot Teams Under Team Heterogeneity and Dynamic Information Uncertainty
Ziqin Yuan, Byung-Cheol Min
OptimizationRepresentation LearningRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposes an adaptive task allocation framework (ATA-HRL) using hierarchical reinforcement learning (HRL) for task allocation in multi-agent and multi-robot teams.
Adaptive Thresholding for Sequence-Based Place Recognition
O. Vysotska, C. Stachniss
RetrievalSequential
🎯 What it does: Proposes an automatic similarity threshold selection technique and integrates it into a complete sequence-based visual localization system.
Adaptive Walker: User Intention and Terrain Aware Intelligent Walker with High-Resolution Tactile and IMU Sensor
Yunho Choi, Kyung-Joong Kim
Robotic IntelligenceMultimodalityTime Series
🎯 What it does: Propose an adaptive walker system that dynamically responds to user intent and terrain changes by utilizing high-resolution tactile sensors, deep learning algorithms, IMU sensors, and linear motor dynamic responses.
Added Mass and Accuracy of the FF -SLIP Model for Legged Swimming
Max P. Austin, Jonathan E. Clark
Robotic IntelligencePhysics RelatedOrdinary Differential Equation
🎯 What it does: This paper introduces two additional mass models into the elastic spring-driven inverted pendulum model (FF-SLIP) and evaluates their effectiveness on the two-legged swimming robot Tadpole.
Addition of a Peristaltic Wave Improves Multi-Legged Locomotion Performance on Complex Terrains
Massimiliano Iaschi, Daniel I. Goldman
Robotic Intelligence
🎯 What it does: Designed and tested a multi-legged robot with five-segment, static legs, incorporating peristaltic waves and longitudinal waves to evaluate its locomotion performance on flat and rough terrain.
ADMM-MCBF-LCA: A Layered Control Architecture for Safe Real-Time Navigation
Anusha Srikanthan, Nadia Figueroa
Autonomous DrivingOptimizationSafty and Privacy
🎯 What it does: Propose a hierarchical control architecture that includes offline path library generation, online path selection, and a safety layer to achieve safe real-time navigation.
Advanced $X \theta$ Reluctance Electromagnetic Micropositioning System for Precision Motion Control
Michael Pumphrey, M. Janaideh
OptimizationPhysics Related
🎯 What it does: Designed and verified a system utilizing C-core magnetic reluctance actuators and two sets of moving magnetic actuators to achieve X-θ dual-degree-of-freedom micro-positioning.
Advancing Dense Endoscopic Reconstruction with Gaussian Splatting-Driven Surface Normal-Aware Tracking and Mapping
Yiming Huang, Hongliang Ren
Gaussian SplattingSimultaneous Localization and MappingBiomedical Data
🎯 What it does: Proposed a real-time endoscopic SLAM system called Endo-2DTAM, integrating 2D Gaussian scattering and surface normal perception for tracking, mapping, and bundle adjustment modules, achieving geometrically accurate reconstruction.
AERAS: Adaptive Experience Replay with Attention-Based Sequence Embedding for Improved Multi-Agent Reinforcement Learning
Zaipeng Xie, Wenzhan Song
TransformerReinforcement LearningSequential
🎯 What it does: Propose the AERAS framework, combining sequence embedding and attention mechanisms to adaptively weight prioritize experiences.
Aerial Grasping by Multi-Limbed Flying Robot SPIDAR Based on Vectored Thrust Control
Moju Zhao
OptimizationRobotic Intelligence
🎯 What it does: Studies the multi-arm flying robot SPIDAR in achieving grasping and transportation in the air using vector thrust
AeroSafe: Mobile Indoor Air Purification Using Aerosol Residence Time Analysis and Robotic Cough Emulator Testbed
M. Tanjid, Tauhidur Rahman
OptimizationRobotic IntelligenceRecurrent Neural NetworkGraph Neural Network
🎯 What it does: Developed AeroSafe, using a robotic cough simulator test platform and aerosol residence time analysis to enhance indoor air purification effectiveness.
AF-RLIO: Adaptive Fusion of Radar-LiDAR-Inertial Information for Robust Odometry in Challenging Environments
Chenglong Qian, Liang Li
Autonomous DrivingOptimizationRobotic IntelligenceSimultaneous Localization and MappingMultimodalityPoint Cloud
🎯 What it does: Propose AF-RLIO, achieving robust odometry estimation by fusing 4D millimeter-wave radar, LiDAR, IMU, and GPS.
Affordance-Based Explanations of Robot Navigation
Amar Halilovic, Senka Krivic
Explainability and InterpretabilityRobotic IntelligenceTextMultimodality
🎯 What it does: This paper proposes and implements a robot navigation decision explanation method based on affordances, integrating visual and textual explanations and validating its high user satisfaction through user studies.
Agile Continuous Jumping in Discontinuous Terrains
Yuxiang Yang, Byron Boots
Robotic IntelligenceReinforcement Learning
🎯 What it does: This paper studies the technology for achieving agile, continuous, and terrain-adaptive jumping in quadruped robots on discontinuous terrains (e.g., stairs, jump stones).
Agile Mobility with Rapid Online Adaptation via Meta-Learning and Uncertainty-Aware MPPI
Dvij Kalaria, John M. Dolan
Robotic IntelligenceMeta Learning
🎯 What it does: Designed a meta-pretrained model-based learning controller that can quickly adapt to any wheeled robot using a small amount of dynamic data while considering model uncertainty.
AI-Enhanced Automatic Design of Efficient Underwater Gliders
P. Y. Chen, Wojciech Matusik
OptimizationRobotic IntelligenceFlow-based ModelMeshPhysics Related
🎯 What it does: Developed an AI-enhanced automated computational framework for co-optimizing the shape and control signals of underwater gliders, thereby generating non-trivial hull designs.
Air-FAR: Fast and Adaptable Routing for Aerial Navigation in Large-Scale Complex Unknown Environments
Botao He, Ji Zhang
Autonomous DrivingOptimization
🎯 What it does: Proposes a real-time 3D navigation algorithm based on hierarchical 3D visibility graph (V-graph) and efficient path search methods, enabling rapid path planning in large-scale complex unknown environments.
AIR-HLoc: Adaptive Retrieved Images Selection for Efficient Visual Localisation
Changkun Liu, Tristan Braud
RetrievalComputational EfficiencySimultaneous Localization and MappingImage
🎯 What it does: Propose an adaptive strategy for selecting the image retrieval number k, dynamically adjusting k based on the similarity between the global descriptors of the query image and the database image to maintain localization accuracy while reducing computational costs.
Airflow Source Seeking on Small Quadrotors Using a Single Flow Sensor
Lenworth Thomas, Sarah Bergbreiter
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: A custom flow sensor capable of simultaneously sensing flow velocity and direction was developed for small (less than 100μg) quadcopters, and an improved Cast and Surge algorithm was implemented using it to achieve airflow source tracking.
AlignBot: Aligning VLM-Powered Customized Task Planning with User Reminders Through Fine-Tuning for Household Robots
Zhaxizhuoma Zhaxizhuoma, Xuelong Li
Robotic IntelligenceTransformerSupervised Fine-TuningVision Language ModelTextMultimodalityRetrieval-Augmented Generation
🎯 What it does: Developed the AlignBot framework, leveraging a fine-tuned LLaVA-7B as an adapter for GPT-40 to convert user reminders into structured instructions, and enhancing home robot VLM-driven customized task planning through dynamic retrieval of historical successful cases;
AllGaits: Learning All Quadruped Gaits and Transitions
Guillaume Bellegarda, A. Ijspeert
Robotic IntelligenceReinforcement Learning
🎯 What it does: Train a single strategy using deep reinforcement learning to control central pattern generators (CPG) and a pattern formation layer, enabling a quadruped robot to achieve all gaits and their instant transitions, with systematic evaluation of energy efficiency under the relationship between gait and speed.
Ambient Flow Perception of Freely Swimming Robotic Fish Using an Artificial Lateral Line System
Hongru Dai, Yang Wang
ClassificationRobotic IntelligenceRecurrent Neural NetworkTime SeriesSequential
🎯 What it does: Developed an environmental flow classifier based on the artificial lateral line system (ALLS), enabling robotic fish to perceive flow fields while swimming freely.
An Active Perception Game for Robust Information Gathering
Siming He, Pratik Chaudhari
OptimizationRobotic Intelligence
🎯 What it does: Proposes a game-theoretic active sensing framework that estimates the difference between predicted information gain and actual information gain, achieving sublinear regret through online estimation to reduce the suboptimality of active sensing systems.
An Adversarial Learning Framework for Reliable Myoelectric Force Estimation Under Fatigue
Huiming Pan, Peter B. Shull
Domain AdaptationGenerative Adversarial NetworkBiomedical Data
🎯 What it does: Propose an adversarial learning framework for reliable myoelectric force estimation under muscle fatigue conditions
An Algorithm for Geometric Navigation Planning Under Uncertainty Using Terrain Boundary Detection
Bennett A. Carley, Jason M. O'Kane
OptimizationRobotic Intelligence
🎯 What it does: Studied and proposed an algorithm for geometric navigation planning under uncertainty, applicable to robots with extreme perception and large steering ratio errors, which plans reliable action sequences by detecting terrain boundaries in known terrain map environments;
An Average-Distance Minimizing Motion Sweep for Bounded Spatial Objects and Its Application in Bézier-Like Freeform Motion Generation
Huan Liu, Q. Ge
GenerationOptimization
🎯 What it does: Propose a method to construct motion sweeping using ellipsoid parameters based on the volume inertia moment, generating motion between two given poses with minimal average squared distance (ASD);
An EEG Conformer Model for Error Feedback During Human-Robot Interaction
Jinpei Han, A. Faisal
ClassificationRobotic IntelligenceConvolutional Neural NetworkTransformerBiomedical Data
🎯 What it does: Developed a causal EEG conformer framework for real-time prediction of error-related potentials (ErrP) signals in human-robot interaction (HRI), and conducted cross-validation evaluation on exoskeleton-assisted robots in a pseudo-online environment.
An Efficient NSGA-II-Based Algorithm for Multi-Robot Coverage Path Planning
Ashley J. I. Foster, Hooman Samani
OptimizationRobotic Intelligence
🎯 What it does: Proposed a multi-objective multi-robot coverage path planning algorithm based on NSGA-II, utilizing a donation mutation operator and multi-parent crossover operator to generate solutions that balance the longest path and average path length, while improving offspring quality through an elite path library and tournament selection.
An End-to-End Learning-Based Multi-Sensor Fusion for Autonomous Vehicle Localization
Changhong Lin, Bo Zhang
Autonomous DrivingMultimodality
🎯 What it does: Propose a learning-based multi-sensor fusion method that utilizes high-order neural network features to encode sensor information, constructing an end-to-end network to eliminate uncertainty modeling and parameter tuning.
An Equilibrium Analysis of Magnetic Quadrupole Force Field With Applications to Microrobotic Swarm Coordination
Ioannis Faros, Herbert G. Tanner
Robotic IntelligencePhysics Related
🎯 What it does: Developed a method utilizing the equilibrium points of a magnetic quadrupole force field to achieve overall or subgroup coordinated control of homogeneous magnetic microrobot swarms, and to split the swarm into two groups for guiding them to different positions separately.
An Interactive Hands-Free Controller for a Riding Ballbot to Enable Simple Shared Control Tasks
Chenzhang Xiao, Elizabeth T. Hsiao-Wecksler
Robotic Intelligence
🎯 What it does: Designed and verified an interactive hands-free adaptation control scheme, iHACS, to enhance speed tracking and safe shared control for the riding ball robot PURE.
An Iterative Approach for Heterogeneous Multi-Agent Route Planning with Resource Transportation Uncertainty and Temporal Logic Goals
Gustavo A. Cardona, C. Vasile
Autonomous DrivingOptimization
🎯 What it does: Proposed an iterative method for heterogeneous multi-agent path planning in environments with unknown resource distribution.
An Omnidirectional Non-Tethered Aerial Prototype with Fixed Uni-Directional Thrusters
Mahmoud Hamandi, F. Khorrami
OptimizationRobotic Intelligence
🎯 What it does: Developed the first global functional prototype of an omnidirectional multirotor drone equipped with fixed unidirectional propellers and onboard power; an optimization algorithm was used to calculate the position and orientation of the propellers within the drone's frame to achieve omnidirectional mobility; in experiments, aerodynamic interactions between different propellers were identified and quantified, and the results were incorporated into the optimization algorithm to avoid interference during flight; the drone's torque decoupling, ability to simultaneously track independent position and attitude, and capability to rotate while hovering at a fixed position were validated in real-world experiments.
An Ultra-Light Seedling Planting Mechanism for Use in Aerial Reforestation
S. Lloyd, Rasmus Astrup
Agriculture Related
🎯 What it does: Proposed an ultra-lightweight tree-planting mechanism for aerial tree planting, using high-pressure compressed air drive and dual telescoping design, weighing only 8kg, suitable for medium to large drones to carry;
Analysis of Kinematics and Propulsion of a Self-Sensing Multi-DoF Undulating Soft Robotic Fish
Myungsun Park, M. Tolley
Robotic Intelligence
🎯 What it does: This paper studies the kinematics of self-sensing multi-degree-of-freedom soft robotic fish, ranging from eel-like to swallowing-like undulatory patterns, evaluates the propulsion efficiency of different motion patterns by measuring thrust, and ultimately estimates their steady-speed swimming velocity using drag experiments.
Angular Divergent Component of Motion: A Step Towards Planning Spatial DCM Objectives for Legged Robots
Connor W. Herron, Johannes Englsberger
Robotic Intelligence
🎯 What it does: The study proposes a spatial DCM method incorporating angular coordinates, and conducts simulation and hardware validation under the 3D linear + 1D angular DCM framework.
Anisotropic Stiffness and Programmable Actuation for Soft Robots Enabled by an Inflated Rotational Joint
Sicheng Wang, Laura H. Blumenschein
Robotic Intelligence
🎯 What it does: A demonstration of an inflatable soft robotic actuator module that defines a bending plane and regulates final stiffness through the ratio of wrinkled to non-wrinkled regions by enforcing local wrinkling; its stiffness characteristics are validated through first-principles models and experimental characterization, and the module's ability to maintain kinematic constraints under various loading conditions is demonstrated.
Annealed Winner-Takes-All for Motion Forecasting
Yihong Xu, Matthieu Cord
Autonomous DrivingComputational Efficiency
🎯 What it does: Demonstrate integrating the annealed Winner-Takes-All (aWTA) loss into state-of-the-art motion prediction models to minimize the number of hypotheses and improve performance.
Anomalies-by-Synthesis: Anomaly Detection using Generative Diffusion Models for Off-Road Navigation
Siddharth Ancha, Nicholas Roy
Anomaly DetectionAutonomous DrivingVision Language ModelDiffusion modelImage
🎯 What it does: Propose a pixel-level anomaly detection method based on analysis-synthesis, which synthesizes edited images with anomalies removed from input images using diffusion generative models, and identifies anomalies by analyzing the modified image segments.
Anti-Sensing: Defense Against Unauthorized Radar-Based Human Vital Sign Sensing with Physically Realizable Wearable Oscillators
Md. Farhan Tasnim Oshim, Tauhidur Rahman
OptimizationSafty and PrivacyBiomedical Data
🎯 What it does: Proposes a defense mechanism called Anti-Sensing that utilizes wearable oscillators to generate physically realizable perturbations to mimic natural heartbeats and mislead UWB radar heart rate estimation.
Anticipatory Planning for Performant Long-Lived Robot in Large-Scale Home-Like Environments
Md Ridwan Hossain Talukder, Gregory J. Stein
Robotic IntelligenceGraph Neural NetworkGraph
🎯 What it does: Proposes a model-driven prospective task planning framework for large-scale, long-term home-like environments, aiming to simultaneously optimize the planning cost of current tasks and the expected cost of future tasks.
AnyCar to Anywhere: Learning Universal Dynamics Model for Agile and Adaptive Mobility
Wenli Xiao, Guanya Shi
Data SynthesisAutonomous DrivingOptimizationRobotic IntelligenceTransformerSupervised Fine-TuningWorld Model
🎯 What it does: Proposed a Transformer-based general dynamics model called AnyCar, which can achieve agile control on various wheeled robots, and through unifying multiple simulators to generate diverse training data, combined with robust training and real-world fine-tuning to achieve precise adaptation.
AnySkin: Plug-and-Play Skin Sensing for Robotic Touch
Raunaq M. Bhirangi, Lerrel Pinto
Robotic Intelligence
🎯 What it does: Proposed a detachable, adhesive-free magnetic tactile sensor called AnySkin, along with a simplified manufacturing process and design tools; conducted characterization analysis of slip detection and control strategy learning for the sensor; demonstrated the possibility of zero-shot transfer across different instances;
APA-BI: Adaptive Partition Aggregation and Bidirectional Integration for UAV-View Geo-Localization
Xichen Zhang, Yizhong Zhang
RetrievalConvolutional Neural NetworkImage
🎯 What it does: Proposed a UAV perspective geolocation method APA-BI, which includes adaptive partition aggregation and bidirectional integration modules.
Application of Koopman Direct Encoding-Based Model Predictive Control to Nonlinear Electromechanical Systems
Sungbin Park, Jung Kim
OptimizationPhysics Related
🎯 What it does: A model predictive control (KDE-MPC) method based on Koopman direct encoding is studied and verified for nonlinear electromechanical systems with piecewise dynamic conditions.
AquaMILR: Mechanical Intelligence Simplifies Control of Undulatory Robots in Cluttered Fluid Environments
Tianyu Wang, Daniel I. Goldman
Robotic Intelligence
🎯 What it does: Developed a limbless undulating robot on water using a dual-cable drive mechanism with programmable anisotropic flexibility, and experimentally validated the feasibility of mechanical intelligence principles in complex fluid environments.
AquaMILR+: Design of an Untethered Limbless Robot for Complex Aquatic Terrain Navigation
M. Fernandez, Daniel I. Goldman
Robotic Intelligence
🎯 What it does: Designed, manufactured, and tested a cableless and legless underwater robot named AquaMILR+, demonstrating its maneuverability and depth control capabilities in complex aquatic environments.
ARCap: Collecting High-Quality Human Demonstrations for Robot Learning with Augmented Reality Feedback
Sirui Chen, C. K. Liu
Data-Centric LearningRobotic Intelligence
🎯 What it does: Propose a portable data acquisition system called ARCap that uses augmented reality (AR) visual feedback and haptic warnings to guide users in collecting high-quality robot-executable demonstrations;
Arena 4.0: a Comprehensive Ros2 Development and Benchmarking Platform for Human-Centric Navigation Using Generative-Model-Based Environment Generation
V. Shcherbyna, Harold Soh
GenerationData SynthesisAutonomous DrivingLarge Language ModelDiffusion modelImageTextMeshBenchmark
🎯 What it does: This paper proposes and implements Arena 4.0, a development and benchmark platform for human-centered navigation based on ROS 2, integrating generative models for environment and scenario generation, a complete 3D model database, and a comprehensive migration to ROS 2.
ARS-SLAM: Accurate Robust Spinning LiDAR SLAM for a Quadruped Robot in Large-Scale Scenario
Member Ieee Jiehao Li, Fellow Ieee Chenguang Yang
OptimizationRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposed a precise and robust rotational LiDAR SLAM (ARS-SLAM) algorithm for large-scale scenarios, integrating the tight-coupled iterative Kalman filter from FAST-LIO2 into the frontend of the Cartographer framework, and introducing a pose threshold optimization algorithm to reduce redundant information in loop detection and improve computational efficiency.
ASCENT: Autonomous Skill Learning Toward Complex Embodied Tasks With Foundation Models
Haolin Wu, Shunbo Zhou
Robotic IntelligenceTransformerLarge Language ModelAgentic AI
🎯 What it does: Proposed and implemented the ASCENT framework, which leverages large language models (LLMs) and large multimodal models (LMMs) to automatically generate simulated scenarios and tasks. The framework enables AI agents to select appropriate atomic skills from an atomic skill library to complete complex embodied tasks and generate trajectory data.
Assembly Order Planning for Modular Structures by Autonomous Multi-Robot Systems
Tom Peters, Irina Kostitsyna
OptimizationRobotic Intelligence
🎯 What it does: Proposed an algorithm for computing construction plans for robots under the constraints of the ARMADAS model, and evaluated the quality of the plan in experiments
AstroLoc2: Fast Sequential Depth-Enhanced Localization for Free-Flying Robots
Ryan Soussan, Trey Smith
Pose EstimationRobotic IntelligenceSimultaneous Localization and MappingImage
🎯 What it does: Proposed AstroLoc2, a visual-inertial graph localizer integrating monocular vision and time-of-flight (ToF) sensors, achieving high-precision positioning and efficient operation for the Astrobee free-flying robot on the International Space Station (ISS)
Asymptotically Optimal Sampling-Based Motion Planning Through Anytime Incremental Lazy Bidirectional Heuristic Search
Yi Wang, Oren Salzman
Autonomous DrivingOptimization
🎯 What it does: Proposed the Bidirectional Lazy Informed Trees (BLIT*) algorithm, first integrating the on-the-fly incremental lazy bidirectional heuristic search (Bi-HS) into batch sampling-based motion planning (Bw-SBMP);
Asymptotically-Optimal Multi-Query Path Planning for a Polygonal Robot
Duo Zhang, Jingjin Yu
OptimizationRobotic Intelligence
🎯 What it does: For multi-query 2D environment polyhedral global robot path planning, the rotational stacked visibility graph (RVG) algorithm is proposed, achieving fast computation of near-optimal paths while supporting simultaneous translation and rotation.
Asynchronous Multi-Object Tracking with an Event Camera
Angus Apps, Robert Mahony
Object TrackingOptical FlowVideo
🎯 What it does: Proposed the AEMOT algorithm, which achieves the detection and tracking of multiple objects by asynchronously processing raw events using an event camera.
Atom: Adaptive Theory-of-Mind-Based Human Motion Prediction in Long-Term Human-Robot Interactions
Yuwen Liao, Lihua Xie
Explainability and InterpretabilityRobotic Intelligence
🎯 What it does: Propose an adaptive human motion prediction model based on Theory-of-Mind (ToM) for long-term human-robot interaction scenarios
Automated Hybrid Reward Scheduling Via Large Language Models for Robotic Skill Learning
Changxin Huang, Jianqiang Li
OptimizationRobotic IntelligenceTransformerLarge Language ModelReinforcement Learning
🎯 What it does: Propose an automatic hybrid reward scheduling framework based on large language models (LLMs), dynamically adjusting the learning intensity of each reward component to achieve progressive learning of high degrees of freedom (dof) skills in robots.
Automated Planning Domain Inference for Task and Motion Planning
Jinbang Huang, Florian Shkurti
Robotic IntelligenceSequential
🎯 What it does: Automatically infer the planning domain required by the task and motion planning (TAMP) framework using a small number of demonstration trajectories, generating new planning domains to improve the efficiency of robots executing complex tasks.
Automated Video Object Detection of Motile Cells Under Microscopy
Haocong Song, Yu Sun
Object DetectionTransformerBiomedical Data
🎯 What it does: Proposed a video object detection network combining static and dynamic queries for detecting moving cells in microscope videos; adopted a two-stage framework, first generating high-quality queries for reference frames using a pre-trained static Transformer decoder, then modeling the current frame with a dynamic Transformer decoder, and introducing a Reference Query Relation Module to enhance query aggregation.
Automatic Behavior Tree Expansion with LLMs for Robotic Manipulation
J. Styrud, Christian Smith
Robotic IntelligenceTransformerLarge Language Model
🎯 What it does: Proposed an automatic expansion method of behavior trees based on large language models for control strategy configuration in robotic manipulation tasks.
Automatic Robotic-Assisted Diffuse Reflectance Spectroscopy Scanning System
Kaizhong Deng, Daniel S. Elson
Robotic IntelligenceBiomedical Data
🎯 What it does: Proposed an automated robot-assisted diffuse reflectance spectroscopy scanning system for large-area spectral acquisition during surgery.
Automating Tension Calibration for Tendon-Driven Continuum Robots: A Low-Cost Approach Towards Consistent Teleoperation
Kyum Lee, J. Burgner-Kahrs
Robotic Intelligence
🎯 What it does: Propose a low-cost automatic tension calibration method suitable for tendon-driven continuous robots lacking tension sensing.
Autonomous Bimanual Manipulation of Deformable Objects Using Deep Reinforcement Learning Guided Adaptive Control
Jiayi Liu, Han Ding
Domain AdaptationRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a deep reinforcement learning guided adaptive control framework for autonomous manual manipulation of deformable objects.
Autonomous Continuous Capsulorhexis Based on a Force-Vision-Guided Robot System
Hongli Liang, Kai Huang
Robotic IntelligenceBiomedical Data
🎯 What it does: Proposed a force-visual guided robotic system to automate the continuous curve capsulotomy (CCC) in cataract surgery
Autonomous Drone for Dynamic Smoke Plume Tracking
Srijan Kumar Pal, Jiarong Hong
Object TrackingReinforcement LearningImage
🎯 What it does: Developed an automated smoke plume tracking system based on quadrotor drones, capable of navigation, detection, and real-time tracking under highly unstable atmospheric conditions.
Autonomous Excavation of Challenging Terrain using Oscillatory Primitives and Adaptive Impedance Control
Noah Franceschini, Kris Hauser
Robotic Intelligence
🎯 What it does: Proposes an autonomous digging method for challenging terrains with clogging and particle adhesion, combining human-inspired oscillating rotating elements (swivel, twist, dive) and an adaptive impedance controller (RAIC) to reduce clogging and enhance excavation efficiency.
Autonomous Navigation in Crowded Space Using Multi-Sensory Data Fusion
Nourin Siddique Ananna, Md. Golam Rabiul Alam
Autonomous DrivingRobotic IntelligenceConvolutional Neural NetworkGenerative Adversarial NetworkMultimodality
🎯 What it does: Proposes a crowd navigation method that achieves socially compliant navigation by utilizing human pose tracking, trajectory prediction, and obstacle avoidance techniques
Autonomous Navigation in Ice-Covered Waters with Learned Predictions on Ship-Ice Interactions
Ninghan Zhong, Stephen L. Smith
Autonomous DrivingRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Proposed a deep learning-based ice movement prediction model integrated into a graph search planner, achieving real-time autonomous navigation in icy areas, with verification showing a significant reduction in collisions with ice floes.
Autonomous Sensor Exchange and Calibration for Cornstalk Nitrate Monitoring Robot
J. Lee, Oliver Kroemer
Robotic IntelligenceAgriculture Related
🎯 What it does: Developed an autonomous sensor replacement and calibration system for robots used in nitrate monitoring of corn stalks.
Autonomous Wheel Loader Navigation Using Goal-Conditioned Actor-Critic MPC
Aleksi Mäki-Penttilä, Reza Ghabcheloo
Autonomous DrivingOptimizationReinforcement Learning
🎯 What it does: Proposed and verified an MPC control method using a critic trained by Actor-Critic reinforcement learning as the cost function for efficient target pose navigation in wheeled loaders.
AutoPeel: Adhesion-Aware Safe Peeling Trajectory Optimization for Robotic Wound Care
Xiao Liang, Michael C. Yip
OptimizationRobotic IntelligenceBiomedical Data
🎯 What it does: Developed the first robotic system for trauma dressing removal, utilizing differentiable physics simulation to achieve gradient-based peeling trajectory optimization.
AutoSplat: Constrained Gaussian Splatting for Autonomous Driving Scene Reconstruction
Mustafa Khan, Bingbing Liu
Autonomous DrivingGaussian SplattingImage
🎯 What it does: Propose the AutoSplat framework, which utilizes Gaussian splatting technology to achieve realistic scene reconstruction and view synthesis in autonomous driving scenarios.
AVD2: Accident Video Diffusion for Accident Video Description
Cheng Li, Hao Zhao
GenerationData SynthesisVision Language ModelDiffusion modelVideoTextMultimodality
🎯 What it does: Proposed the AVD2 framework, which generates accident videos consistent with detailed natural language descriptions and reasoning through an accident video diffusion model, and contributed the EMM-AU multi-modal accident video understanding dataset.
Back to the Cartesian: Pilot Study for Assessing Human Stiffness in 3D Cartesian Space by Transforming from Muscle Space in a Peg-In-Hole Scenario for Tele-Impedance
Sabine Thürauf, Marek Sierotowicz
Robotic IntelligenceBiomedical Data
🎯 What it does: Introduces a full 3D Cartesian space human stiffness measurement method based on electromyography (EMG) and validates its effectiveness in insertion tasks with different directions.
Bandwidth-Adaptive Spatiotemporal Correspondence Identification for Collaborative Perception
Peng Gao, Hao Zhang
Autonomous DrivingOptimizationComputational EfficiencyRobotic Intelligence
🎯 What it does: Proposes a bandwidth-adaptive spatiotemporal correspondence identification (CoID) method, enabling robots to gradually select and share partial spatiotemporal observations under dynamic communication constraints;
Bat-VUFN: Bat-Inspired Visual-and-Ultrasound Fusion Network for Robust Perception in Adverse Conditions
Gyeongrok Lim, Hyeon-Min Bae
Robotic IntelligenceImageMultimodalityUltrasound
🎯 What it does: Proposed a bat-inspired multi-sensor system called Bat-VUFN, which fuses camera and ultrasound sensor data through input quality scoring (IQS) to enhance near-field perception in harsh environments.
Bayesian Optimal Experimental Design for Robot Kinematic Calibration
Ersin Daş, J. W. Burdick
OptimizationRobotic Intelligence
🎯 What it does: Proposed a Bayesian optimal experimental design method for robot rigid body motion calibration.
Behav: Behavioral Rule Guided Autonomy Using VLMs for Robot Navigation in Outdoor Scenes
Kasun Weerakoon, Dinesh Manocha
Autonomous DrivingOptimizationRobotic IntelligenceLarge Language ModelVision Language ModelPoint Cloud
🎯 What it does: Propose the BehAV system, which uses Vision Language Models (VLM) and Large Language Models (LLM) for navigation and behavior decomposition. The system employs VLM for zero-shot localization of objects and constructs a behavior cost map, integrates LiDAR occupancy maps, and adopts unconstrained Model Predictive Control (MPC) planning to achieve autonomous navigation for outdoor robots.
Behavioral Manifolds: Representing the Landscape of Grasp Affordances in Relative Pose Space
Michael Zechmair, Yannick Morel
Representation LearningRobotic Intelligence
🎯 What it does: Proposes a new method that combines grasp adaptability learning with grasp synthesis (i.e., achieving grasping through robotic arm kinematics), explicitly mapping the grasp strategy space based on generated grasp types and corresponding grasp quality.
BEINGS: Bayesian Embodied Image-Goal Navigation With Gaussian Splatting
Wugang Meng, Fumin Zhang
Autonomous DrivingOptimizationGaussian SplattingImage
🎯 What it does: Propose an image-based target navigation method based on Bayesian update and 3D Gaussian Splatting, modeling the image navigation problem as an optimal control problem within the Model Predictive Control (MPC) framework
Belief Roadmaps with Uncertain Landmark Evanescence
Erick Fuentes, Nicholas Roy
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Proposes a robot navigation method called BRULE that considers landmark evanescence;
Benchmarking Different QP Formulations and Solvers for Dynamic Quadrupedal Walking
Franek Stark, Frank Kirchner
OptimizationComputational EfficiencyRobotic IntelligenceBenchmark
🎯 What it does: Compare dense and sparse QP forms and various solving methods on different hardware, and propose the SFPW metric to evaluate computational efficiency under dynamic gaits of quadruped robots.
Berkeley Humanoid: A Research Platform for Learning-Based Control
Qiayuan Liao, K. Sreenath
Robotic IntelligenceReinforcement Learning
🎯 What it does: Introduces the Berkeley Humanoid robot platform, which aims to support control using learning algorithms and demonstrates its robust walking capabilities on various outdoor terrains.
BETTY Dataset: A Multi-Modal Dataset for Full-Stack Autonomy
Micah Nye, Sebastian A. Scherer
Autonomous DrivingMultimodalityBenchmark
🎯 What it does: Proposed and released the BETTY dataset, a large-scale multimodal dataset collected on multiple autonomous racing platforms, encompassing sensor inputs, software stack outputs, semantic metadata, and ground truth information, supporting tasks such as supervised and self-supervised state estimation, dynamics modeling, motion prediction, and perception.
Beyond Bare Queries: Open-Vocabulary Object Grounding with 3D Scene Graph
Sergey Linok, Aleksei Valenkov
Object DetectionSegmentationGraph Neural NetworkLarge Language ModelVision Language ModelContrastive LearningTextPoint Cloud
🎯 What it does: Proposes a modular method called BBQ, which achieves object localization based on natural language descriptions by constructing 3D scene graphs and utilizing large language models for deductive scene reasoning.
Beyond Robustness: Learning Unknown Dynamic Load Adaptation for Quadruped Locomotion on Rough Terrain
Leixin Chang, Liangjing Yang
Robotic IntelligenceReinforcement Learning
🎯 What it does: A general load characteristic modeling method is proposed and combined with reinforcement learning control to achieve adaptive carrying of unknown dynamic loads and stable load walking on rough terrain by quadruped robots without external sensors.
Beyond Sight: Finetuning Generalist Robot Policies with Heterogeneous Sensors via Language Grounding
Joshua Jones, Sergey Levine
Robotic IntelligenceLarge Language ModelSupervised Fine-TuningPrompt EngineeringVision-Language-Action ModelDiffusion modelContrastive LearningTextMultimodality
🎯 What it does: Propose the FuSe method, which utilizes natural language as a cross-modal benchmark to fine-tune large-scale general robot policies, integrating heterogeneous sensor information such as vision, touch, and sound.
Beyond Simulation: Benchmarking World Models for Planning and Causality in Autonomous Driving
Hunter Schofield, Jinjun Shan
Autonomous DrivingContrastive LearningBenchmark
🎯 What it does: Analyze and evaluate the robustness of world models in traffic simulation metrics, compare their performance in partial replay scenarios versus standard scenarios, and propose new evaluation metrics.
Beyond Traversing in a Thin Pipe: Self-Sensing Odometry of a Pipeline Robot Driven by High-Frequency Dielectric Elastomer Actuators
Ran Cheng, Huichan Zhao
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: A mole-like pipeline robot was proposed, utilizing dielectric elastomer actuators and passive anchoring to achieve single-directional non-slipping motion, and self-sensing to realize step-wise odometry with a 0.05 mm resolution at high frequencies (20 Hz).
Bi-Stream Knowledge Transfer for Semi-Supervised 3D Point Cloud Object Detection
Jilai Zheng, Chao Ma
Object DetectionAutonomous DrivingKnowledge DistillationDiffusion modelPoint Cloud
🎯 What it does: Proposed the Bi-Stream Knowledge Transfer (BiKT) framework for semi-supervised 3D point cloud object detection, utilizing two knowledge streams (reliable and ambiguous) to jointly guide the student network;
BiFold: Bimanual Cloth Folding with Language Guidance
Oriol Barbany, Carme Torras
Robotic IntelligenceTransformerVision Language ModelVision-Language-Action ModelMultimodality
🎯 What it does: This paper proposes the BiFold system, which learns bimanual folding actions based on language instructions by utilizing pre-trained vision-language models to predict grasping and folding actions.
Bio-Inspired Distributed Neural Locomotion Controller (D-NLC) for Robust Locomotion and Emergent Behaviors
Zhikai Zhang, Lu Li
Robotic Intelligence
🎯 What it does: Proposed and implemented a distributed neural gait controller (D-NLC), utilizing local proprioceptive feedback to modulate joint-level CPG signals to generate adaptive maneuvering behaviors
Bio-Inspired Soft Magnetic Swimming Robot for Flexible Motions
Xiaosa Li, Wenbo Ding
Robotic IntelligencePhysics Related
🎯 What it does: This paper designs and experiments with a bionic soft magnetic fish-shaped robot, which achieves hovering and surface swimming through fin movement driven by magnetic elastic body muscles under an external magnetic field, and realizes eight-shaped trajectory motion using field gradients generated by a dense planar electromagnetic coil array. The robot is equipped with inertial and temperature sensors, and transmits motion data in real-time via Bluetooth.
Bipedal Walking with Continuously Compliant Robotic Legs
Robin Bendfeld, M. I. C. David Remy
Robotic Intelligence
🎯 What it does: Introduce continuous deformable structures in the lower limbs of bipedal robots to replace traditional rigid components, designing and implementing a novel flexible leg structure, low-level force and kinematic controllers, as well as high-level posture controllers and gait schedulers, achieving successful walking in experiments;
Bird-Inspired Tendon Coupling Improves Paddling Efficiency by Shortening Phase Transition Times
Jianfeng Lin, Alexander Badri-Spröwitz
Robotic IntelligencePhysics Related
🎯 What it does: A hardware experiment studying the utilization of avian tendon coupling mechanisms to shorten the transition time between the recovery and propulsion phases in water-based propulsion, thereby enhancing the efficiency of dragging paddles.