arXivSub Start free trial

ICRA 2024 Papers — Page 16

IEEE International Conference on Robotics and Automation · 1760 papers

Stable, Safe, and Passive Teleoperation of Multi-Robot Systems

Gennaro Notomista

OptimizationRobotic Intelligence

🎯 What it does: Propose a unified framework that utilizes optimal control methods to ensure the stability, safety, and passivity of multi-robot remote control systems

Stackelberg Game-Theoretic Trajectory Guidance for Multi-Robot Systems with Koopman Operator

Yuhan Zhao, Quanyan Zhu

OptimizationRobotic Intelligence

🎯 What it does: A learning method based on the Koopman operator and Stackelberg game theory is utilized to plan trajectories for a leading robot to guide follower robots in achieving collaborative goals.

STAGE: Scalable and Traversability-Aware Graph based Exploration Planner for Dynamically Varying Environments

Akash Patel, G. Nikolakopoulos

Autonomous DrivingOptimizationGraph Neural NetworkSimultaneous Localization and MappingMultimodalityPoint Cloud

🎯 What it does: Proposes a scalable and traversability-sensitive navigation framework that utilizes a two-layer graph representation to achieve efficient large-scale exploration in dynamic environments.

Standardization of Cloth Objects and its Relevance in Robotic Manipulation

Irene Garcia-Camacho, Júlia Borràs Sol

Robotic Intelligence

🎯 What it does: Propose a fabric object characterization framework applicable to robotic applications and investigate the impact of fabric properties on robotic manipulation tasks.

STARK: A Unified Framework for Strongly Coupled Simulation of Rigid and Deformable Bodies with Frictional Contact

J. A. Fernández-fernández, J. Bender

OptimizationPhysics Related

🎯 What it does: Proposed a unified simulation framework named Stark for strong coupling between rigid and deformable bodies with frictional contact, validated through experiments involving interaction between a mobile vacuum cleaner and a towel.

Statler: State-Maintaining Language Models for Embodied Reasoning

Takuma Yoneda, Matthew R. Walter

Robotic IntelligenceTransformerLarge Language ModelPrompt Engineering

🎯 What it does: Propose the Statler framework, which utilizes large language models (LLMs) to maintain estimates of unobservable world states through prompts, tracks state changes as new actions are executed, and subsequently determines each step's action based on the current state estimates;

Stein Variational Guided Model Predictive Path Integral Control: Proposal and Experiments with Fast Maneuvering Vehicles

Kohei Honda, Tatsuya Suzuki

Autonomous DrivingOptimization

🎯 What it does: Proposes a novel stochastic optimal control method called SVG-MPPI based on MPPI to handle rapidly changing multi-modal optimal action distributions.

Stereo Image-based Visual Servoing Towards Feature-based Grasping

Albert Enyedy, Michael Gennert

Robotic IntelligenceImage

🎯 What it does: Propose a visual servoing scheme based on 2D stereo images, which directly maps image errors to joint space under an eye-hand configuration, enabling the robot to control the target position in three-dimensional space without requiring stereo reconstruction, prior target information, or 3D data.

Stereo-LiDAR Depth Estimation with Deformable Propagation and Learned Disparity-Depth Conversion

Ang Li, Danping Zou

Depth EstimationImagePoint Cloud

🎯 What it does: Proposed a stereo-LiDAR depth estimation network called SDG-Depth based on semi-dense hint guidance, which generates semi-dense hint maps and confidence maps using a deformable propagation module, and uses these maps to guide cost aggregation in stereo matching, while introducing a disparity-depth conversion module to reduce triangulation errors.

Stereo-NEC: Enhancing Stereo Visual-Inertial SLAM Initialization with Normal Epipolar Constraints

Weihan Wang, Philippos Mordohai

Simultaneous Localization and MappingImage

🎯 What it does: Proposes a stereo vision-inertial SLAM initialization method based on normal monocular constraints

Stiffness-Based Hybrid Motion/ Force Control for Cable-Driven Serpentine Manipulator*

Wenshuo Li, Bin Liang

Robotic Intelligence

🎯 What it does: Designed a cable-driven serpentine manipulator and implemented stiffness-based motion/force hybrid control.

Stimulate the Potential of Robots via Competition

Kangyao Huang, Huaping Liu

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a competitive learning framework to help robots learn and unlock their potential in competitive environments.

STNet: Spatio-Temporal Fusion-Based SelfAttention for Slip Detection in Visuo-Tactile Sensors

Jin Lu, Jingjing Ji

Anomaly DetectionRobotic IntelligenceTransformerMultimodalityTime Series

🎯 What it does: Propose a self-attention network named STNet based on spatiotemporal sequence fusion for slip detection using visual-tactile sensors.

Stochastic Games for Interactive Manipulation Domains

Karan Muvvala, M. Vardi

Computational EfficiencyReinforcement Learning

🎯 What it does: Proposes using stochastic games as a general model for human-computer interaction, which can cover previous models and reduce assumptions; based on this, semantic definitions are provided, and existing tools (e.g., PRISM-games) are utilized to synthesize strategies that satisfy complex task guarantees; scalability is improved by two orders of magnitude through a novel model construction method.

Stochastic Implicit Neural Signed Distance Functions for Safe Motion Planning under Sensing Uncertainty

Carlos Quintero-Peña, L. Kavraki

OptimizationSafty and PrivacyRobotic IntelligenceStochastic Differential Equation

🎯 What it does: Proposing a method that combines an implicit neural signed distance function, which directly models sensor-specific random uncertainties, with a hierarchical optimization planner for safe motion planning of high-dimensional robots in complex environments

STOPNet: Multiview-based 6-DoF Suction Detection for Transparent Objects on Production Lines

Yuxuan Kuang, He Wang

Pose EstimationRobotic IntelligenceImage

🎯 What it does: Proposes the STOPNet framework for 6-DoF suction detection on production lines, specifically enabling high-quality real-time detection for transparent objects.

Stranger Danger! Identifying and Avoiding Unpredictable Pedestrians in RL-based Social Robot Navigation

Sara Pohland, Claire J. Tomlin

Safty and PrivacyRobotic IntelligenceReinforcement Learning

🎯 What it does: Improve the SARL strategy by systematically introducing pedestrian model bias during training, updating the value network to estimate and utilize pedestrian unpredictability features, and designing a reward function tailored for learning unpredictability, thereby enhancing safety in unfamiliar scenarios and validating effectiveness on physical robots.

Strawberry Weight Estimation Based on Plane-Constrained Binary Division Point Cloud Completion

Yanjiang Huang, Xianmin Zhang

Point CloudAgriculture Related

🎯 What it does: Proposed a strawberry weight estimation method based on binary segmentation of point clouds with plane constraints, and constructed an estimated weight dataset containing strawberries at different heights and angles.

Streamlined Acquisition of Large Sensor Data for Autonomous Mobile Robots to Enable Efficient Creation and Analysis of Datasets

Mark Niemeyer, Joachim Hertzberg

Autonomous DrivingOptimizationRobotic IntelligenceImagePoint Cloud

🎯 What it does: Proposed an efficient method for recording robot sensor data streams, which reduces the total time required for data transmission through the spatiotemporal semantic query interface (SEEREP) and improves the maximum sensor data rate during real-time disk writing, particularly suitable for big data types such as images and point clouds.

Stretch with Stretch: Physical Therapy Exercise Games Led by a Mobile Manipulator

Matthew Lamsey, Charles C. Kemp

Robotic IntelligenceVision-Language-Action ModelImageMultimodalityAudio

🎯 What it does: Designed and evaluated a mobile manipulator-driven PT game system SWS to assist Parkinson's patients in performing stretching exercises.

STT: Stateful Tracking with Transformers for Autonomous Driving

Longlong Jing, Congcong Li

Object TrackingAutonomous DrivingTransformerImagePoint Cloud

🎯 What it does: Proposed a Transformer-based state tracking model called STT that can continuously track targets in 3D space and accurately predict their states.

Subequivariant Reinforcement Learning Framework for Coordinated Motion Control

Haoyu Wang, Chao Qu

Robotic IntelligenceGraph Neural NetworkReinforcement Learning

🎯 What it does: Proposed the CoordiGraph architecture, which enhances coordination in multi-agent motion control by introducing the subequivariant principle from physics

Subgoal Diffuser: Coarse-to-fine Subgoal Generation to Guide Model Predictive Control for Robot Manipulation

Zixuan Huang, Dmitry Berenson

OptimizationRobotic IntelligenceDiffusion model

🎯 What it does: Propose a coarse-to-fine subgoal generation method based on diffusion models, dynamically providing subgoal sequences to MPC planning to achieve long-term robotic operations.

Subsurface Feature-based Ground Robot/Vehicle Localization Using a Ground Penetrating Radar

Haifeng Li, Dezhen Song

Robotic Intelligence

🎯 What it does: This study proposes a subsurface feature localization method that combines ground-penetrating radar (GPR) measurements with a known underground feature map.

SuperFusion: Multilevel LiDAR-Camera Fusion for Long-Range HD Map Generation

Hao Dong, Xieyuanli Chen

GenerationDepth EstimationAutonomous DrivingImageMultimodalityPoint Cloud

🎯 What it does: This paper proposes a multi-level LiDAR-camera fusion network called SuperFusion for generating high-precision semantic maps within short-range (up to 30 meters) and long-range (up to 90 meters) distances.

Supernumerary Robotic Limbs to Support Post-Fall Recoveries for Astronauts

Erik Ballesteros, H. Asada

Robotic Intelligence

🎯 What it does: Research and verified the technical solution of using redundant robotic arms (SuperLimbs) in a partial gravity environment to assist astronauts in recovering from falls during EVAs.

Surgical Gym: A high-performance GPU-based platform for reinforcement learning with surgical robots

Samuel Schmidgall, Jason Eshraghian

Computational EfficiencyRobotic IntelligenceReinforcement Learning

🎯 What it does: Developed Surgical Gym, a high-performance GPU-based platform for direct physics simulation and reinforcement learning on surgical robots;

SWTrack: Multiple Hypothesis Sliding Window 3D Multi-Object Tracking

S. Papais, Steven L. Waslander

Object TrackingAutonomous DrivingGraph Neural NetworkPoint Cloud

🎯 What it does: Propose SWTrack, a 3D multi-object tracking method that utilizes sliding window batch processing of multi-frame sensor data, capable of real-time operation and enhancing the accuracy of association and state estimation.

Symmetric Models for Visual Force Policy Learning

Colin Kohler, Robert W. Platt

Robotic IntelligenceReinforcement LearningMultimodality

🎯 What it does: Proposed a method called Symmetric Visual Force Learning (SVFL), which utilizes visual and force feedback combined with symmetric neural networks for robot control.

Symmetric Stair Preconditioning of Linear Systems for Parallel Trajectory Optimization

Xueyi Bu, B. Plancher

OptimizationComputational Efficiency

🎯 What it does: Proposed and evaluated a novel parallel-friendly symmetric staircase preconditioner to accelerate the iterative solution of moderate-scale sparse linear systems in trajectory optimization problems.

Symmetry Considerations for Learning Task Symmetric Robot Policies

Mayank Mittal, Marco Hutter

Robotic IntelligenceReinforcement Learning

🎯 What it does: This paper studies how to leverage symmetry in goal-oriented robot tasks to enable deep reinforcement learning models to better adhere to symmetry constraints, proposing two methods: data augmentation and mirror loss function, and providing their theoretical foundations under on-policy settings;

Symmetry-aware Reinforcement Learning for Robotic Assembly under Partial Observability with a Soft Wrist

Hai Nguyen, Masashi Hamaya

Robotic IntelligenceRecurrent Neural NetworkReinforcement Learning

🎯 What it does: Using a soft wrist to address plug-and-hole insertion assembly tasks with rich contact, adopting a partially observable reinforcement learning framework, training a memory-based agent to perform operations using only tactile and proprioceptive signals.

Synchronized Dual-arm Rearrangement via Cooperative mTSP

Wenhao Li, Kai Xu

OptimizationRobotic IntelligenceTransformerReinforcement Learning

🎯 What it does: Propose solving the synchronous dual-arm rearrangement problem through cooperative mTSP and reinforcement learning

SynH2R: Synthesizing Hand-Object Motions for Learning Human-to-Robot Handovers

S. Christen, Jie Song

GenerationData SynthesisRobotic Intelligence

🎯 What it does: Proposed a framework that utilizes a hand-object synthesis method to generate human grasp actions for robot training, addressing the reliance on expensive motion capture data.

Synset Boulevard: A Synthetic Image Dataset for VMMR*

Anne Sielemann, Juergen Beyerer

ClassificationData SynthesisImage

🎯 What it does: Proposed and constructed Synset Boulevard, the first-ever fully synthetic vehicle make and model recognition (VMMR) dataset, and evaluated its performance on existing large-scale real-world datasets.

SynthAct: Towards Generalizable Human Action Recognition based on Synthetic Data

David Schneider, Rainer Stiefelhagen

RecognitionData SynthesisPose EstimationVideoMultimodality

🎯 What it does: Designed the SynthAct synthetic data generation pipeline, leveraging 3D pose estimation techniques to generate rich multi-view synthetic data for human action recognition tasks, significantly reducing reliance on real data.

Synthesis of Temporally-Robust Policies for Signal Temporal Logic Tasks using Reinforcement Learning

Siqi Wang, Xiang Yin

OptimizationReinforcement Learning

🎯 What it does: Design control strategies that satisfy Signal Temporal Logic (STL) specifications using reinforcement learning in unknown, random environments, with a focus on enhancing the temporal robustness of the strategies.

Synthesize Efficient Safety Certificates for Learning-Based Safe Control using Magnitude Regularization

Haotian Zheng, Jianqiang Wang

Robotic IntelligenceReinforcement Learning

🎯 What it does: Propose a learning-based safe controller design method that reduces the conservativeness of the energy function while maintaining provable safety through amplitude regularization techniques.

Synthesizing Robust Walking Gaits via Discrete-Time Barrier Functions with Application to Multi-Contact Exoskeleton Locomotion

Maegan Tucker, A. Ames

Robotic Intelligence

🎯 What it does: Proposes a hybrid forward invariant set estimation method based on discrete-time barrier functions and sampling approximation for synthesizing robust gaits, with experimental validation on the Atalante lower-limb exoskeleton.

System Calibration of a Field Phenotyping Robot with Multiple High-Precision Profile Laser Scanners

Felix Esser, H. Kuhlmann

OptimizationRobotic IntelligencePoint CloudAgriculture Related

🎯 What it does: Implemented a new calibration method for calibrating dual industrial-grade laser scanners in agricultural field robots, and proposed a factor graph-based pose estimation method to achieve high-precision pose determination.

System Identification of Space Manipulator Systems and its Implications on Robust Control Performance*

G. Rekleitis, E. Papadopoulos

Robotic IntelligencePhysics Related

🎯 What it does: A novel system identification scheme for spatial mechanical systems (SMS) was developed, capable of identifying all necessary dynamic parameters even when fuel solution and modal states are unmeasurable. The scheme was validated through two experiments; subsequently, the identified parameters were used to simulate model-based and robust controllers in three-dimensional SMS, demonstrating that accurate system identification significantly enhances robust control performance.

TacShade: A New 3D-printed Soft Optical Tactile Sensor Based on Light, Shadow and Greyscale for Shape Reconstruction

Zhenyu Lu, Chenguang Yang

Depth EstimationRobotic IntelligenceImage

🎯 What it does: Designed and implemented a 3D printed soft phototactile sensor named TacShade, and improved the Shape from Shading (SFS) algorithm to achieve rapid rough shape reconstruction of contact objects.

Tactile Embeddings for Multi-Task Learning

Yiyue Luo, Brian Okorn

Representation LearningRobotic Intelligence

🎯 What it does: Construct a unified tactile embedding space for multi-task learning, predicting task phase classification, object dynamics estimation, and tactile dynamics prediction

Tactile Estimation of Extrinsic Contact Patch for Stable Placement

Keita Ota, J. Tenenbaum

Robotic Intelligence

🎯 What it does: Design feedback skills to enable robots to estimate the stability of placement during light-touch interactions, specifically by measuring the contact area between the grasped object and the environment to infer stability.

Tactile Robot Programming: Transferring Task Constraints into Constraint-Based Unified Force-Impedance Control

Kübra Karacan, Sami Haddadin

Robotic Intelligence

🎯 What it does: Propose a task-based tactile robot programming paradigm, which directly maps object constraints in tasks to constraint-based unified force impedance control through object-centered tactile skill definitions; by abstracting task constraints, transmitting them to the robot's workspace, synthesizing unified force impedance control with formalized homogeneous constraints, providing flexible task execution; also propose quantitative analysis metrics, using examples such as lever manipulation and screw removal as flexible operation disassembly skills, combined with Franka Emika robot experiments for real-world validation.

Tactile-Informed Action Primitives Mitigate Jamming in Dense Clutter

Dane Brouwer, M. Cutkosky

Robotic Intelligence

🎯 What it does: Proposed and utilized two action primitives—digging and tunneling—to form a closed-loop hybrid control using tactile and proprioceptive information, enabling obstacle unlocking and continuous advancement in dense crowded environments;

TactileAR: Active Tactile Pattern Reconstruction

Bing Wu, Qian Liu

RestorationSuper ResolutionRobotic IntelligenceSequential

🎯 What it does: Reconstructing the 2D high-resolution contact surface shape using measurement sequences from low-resolution tactile sensors through a Kalman filter-based framework and an active exploration strategy.

Talk2BEV: Language-enhanced Bird’s-eye View Maps for Autonomous Driving

Vikrant Dewangan, K. Krishna

Autonomous DrivingVision Language ModelImageTextMultimodalityBenchmark

🎯 What it does: Propose Talk2BEV, a language-enhanced audio-visual model interface for bird's-eye-view (BEV) in autonomous driving, and evaluate its performance on multiple scene understanding tasks.

TartanDrive 2.0: More Modalities and Better Infrastructure to Further Self-Supervised Learning Research in Off-Road Driving Tasks

Matthew Sivaprakasam, Sebastian A. Scherer

Autonomous DrivingRepresentation LearningReinforcement LearningImageMultimodalityPoint CloudBenchmark

🎯 What it does: Released the TartanDrive 2.0 large-scale offline road driving dataset, and provided tools and metadata systems for data collection, processing, and querying.

Task Allocation in Heterogeneous Multi-Robot Systems Based on Preference-Driven Hedonic Game

Liwang Zhang, Shaowu Yang

OptimizationRobotic Intelligence

🎯 What it does: Propose a preference-driven hedonic game approach for task allocation in multi-robot systems and present a distributed framework

Task-Driven Domain-Agnostic Learning with Information Bottleneck for Autonomous Steering

Yu Shen, Ming C. Lin

Domain AdaptationAutonomous Driving

🎯 What it does: Construct a causal graph using information bottleneck analysis, define the framework and loss function, and propose a domain-agnostic learning method for autonomous driving steering.

Task-Oriented Active Learning of Model Preconditions for Inaccurate Dynamics Models

Alex LaGrassa, Oliver Kroemer

OptimizationData-Centric LearningSequentialAgriculture Related

🎯 What it does: Proposes an active selection algorithm for trajectories to learn model preconditions of inaccurate predefined dynamics models, restricting planning to model-accurate state-action spaces.

Task-space Control of a Powered Ankle Prosthesis

David J. Kelly, Patrick M. Wensing

Robotic IntelligenceBiomedical Data

🎯 What it does: Implemented and evaluated a task space control strategy based on CoM and GRF trajectories, testing the performance of an electric ankle prosthesis when a healthy gaiter used a bypass adapter during walking.

TBD Pedestrian Data Collection: Towards Rich, Portable, and Large-Scale Natural Pedestrian Data

Allan Wang, Aaron Steinfeld

Object TrackingOptical FlowImageVideo

🎯 What it does: Designed and implemented a portable panoramic and first-person perspective data collection system, along with a semi-automated annotation pipeline and a web-based label correction application, enabling rapid, large-scale collection of pedestrian trajectory data with human-validated labels.

Teach and Repeat Navigation: A Robust Control Approach

Payam Nourizadeh, Tobias Fischer

Robotic Intelligence

🎯 What it does: Proposed a robust Teach and Repeat navigation system based on sliding mode control, applicable to skid-steer mobile robots;

TELESIM: A Modular and Plug-and-Play Framework for Robotic Arm Teleoperation using a Digital Twin

Florent P. Audonnet, Gerardo Aragon-Camarasa

Robotic Intelligence

🎯 What it does: Proposed the TELESIM framework, which realizes the use of digital twins as an interface between users and robotic arms, enabling modular, plug-and-play remote operation.

Tendon-Driven Continuum Robot for Deep-Sea Application

Cora Maria Sourkounis, Annika Raatz

Robotic Intelligence

🎯 What it does: Designed a lightweight tendon-driven continuously deformable robot for deep-sea vacuum sampling to reduce sediment sampling costs.

TerrainSense: Vision-Driven Mapless Navigation for Unstructured Off-Road Environments

Bilal Hassan, Jorge Dias

Autonomous DrivingTransformerImage

🎯 What it does: Propose TerrainSense, an end-to-end vision-driven map-free navigation framework capable of detecting lane semantics and topology in rugged off-road environments to achieve map-free path planning.

Terrestrial Locomotion of PogoX: From Hardware Design to Energy Shaping and Step-to-step Dynamics Based Control

Yi Wang, Xiaobin Xiong

Robotic Intelligence

🎯 What it does: A novel controller was designed to enable the quadrotor robot PogoX with spring legs to perform ground bouncing locomotion under low thrust-to-weight ratio (TWR < 1).

Tethered Lifting-Wing Multicopter Landing Like Kite

Haoyu Wei, Quan Quan

Robotic Intelligence

🎯 What it does: A tethered lift-wing multirotor drone landing method without the need for positioning or velocity sensors

TEXterity: Tactile Extrinsic deXterity

Sangwoon Kim, Alberto Rodriguez

Pose EstimationRobotic IntelligenceImage

🎯 What it does: Proposed a framework that combines tactile estimation with control for in-hand manipulation, achieving object pose estimation and tracking by fusing robot kinematics with image-based tactile sensors, and generating motion planning within a recursive horizon to control the grasping posture of objects.

That’s My Point: Compact Object-centric LiDAR Pose Estimation for Large-scale Outdoor Localisation

Georgi Pramatarov, Daniele De Martini

Pose EstimationAutonomous DrivingPoint Cloud

🎯 What it does: By performing semantic clustering on segmented LiDAR point clouds, retaining only the centroids and semantic categories of each object, each LiDAR scan is compressed into a set of quaternion vectors. A self-correlation and cross-correlation based object matching network is then utilized, combined with weighted SVD and RANSAC to achieve 3D pose estimation.

The Double-Scoop Gripper: A Tendon-Driven Soft-Rigid End-Effector for Food Handling Exploiting Constraints in Narrow Spaces

L. Franco, G. Salvietti

Robotic IntelligenceImage

🎯 What it does: This paper proposes a double-scoop, tendinous actuation-driven soft-hard hybrid gripper (Double-Scoop Gripper) designed to gently grasp various shaped ingredients in narrow spaces.

The effect of rejection strategy on trust and shopping choices in robot-assisted shopping *

Matthias Rehm, Carlos Gomez Cubero

Robotic Intelligence

🎯 What it does: Studied how service robots support decision-making in shopping interactions, especially how different rejection strategies influence customer behavior.

The Fractal Hand-II: Reviving a Classic Mechanism for Contemporary Grasping Challenges

Malcolm G. A. Tisdale, J. Burdick

Robotic Intelligence

🎯 What it does: Proposed a novel fractal gripper based on Fractal Vise, providing a low-cost, easy-to-manufacture, large workspace, and high-conformity Fractal Finger design method; by combining two Fractal Fingers with a closed actuator, an adaptive collaborative Fractal Hand was constructed; this gripper can be easily integrated into existing parallel grippers, reducing reliance on sensing and grasping planning;

The Fractal Hand–I: A Non-anthropomorphic, but Synergistic, Adaptable Gripper

J. Burdick, Malcolm G. A. Tisdale

Robotic Intelligence

🎯 What it does: Introduces a tree-like fractal gripper with (2n+1-1) joints actuated by a single actuator, and demonstrates its ability to achieve stable grasping under mild constraints through geometric, kinematic, and quasi-static modeling

The GEM-C controller for Load Compensation in Object Manipulation

Emmanouil Papadakis, P. Trahanias

OptimizationRobotic Intelligence

🎯 What it does: Proposed the GEM-C controller, which improves grasp quality and enhances robot performance by real-time adjustment of the suction cup MIGHTY's posture to compensate for gravity, external forces, and motion.

The GOOSE Dataset for Perception in Unstructured Environments

Peter Mortimer, Hans-Joachim Wuensche

SegmentationAutonomous DrivingRobotic IntelligenceImagePoint CloudBenchmark

🎯 What it does: Constructed and made public the German Outdoor and Off-road Dataset (GOOSE) for unstructured outdoor environments, containing 10,000 pairs of annotated images and point cloud data, and trained multiple advanced segmentation models.

The Grasp Loop Signature: A Topological Representation for Manipulation Planning with Ropes and Cables

P. Mitrano, Dmitry Berenson

Representation LearningRobotic Intelligence

🎯 What it does: Studied the manipulation of deformable one-dimensional objects (e.g., ropes or cables) by robots, proposed a representation method based on grasp closed-loop topology, and applied it to planning guidance, completing simulation and real-world dual-arm cable manipulation experiments.

The Grasp Reset Mechanism: An Automated Apparatus for Conducting Grasping Trials

Kyle DuFrene, Cindy Grimm

Robotic IntelligenceVideoTabular

🎯 What it does: Proposed the Grasp Reset Mechanism (GRM), a fully automated device for large-scale grasping experiments, and provided the corresponding software and hardware design.

The Impact of Evolutionary Computation on Robotic Design: A Case Study with an Underactuated Hand Exoskeleton

Baris Akbas, Fabio Stroppa

OptimizationRobotic Intelligence

🎯 What it does: Optimize the design of the unpowered arm exoskeleton (U-HEx) using evolutionary computation methods (genetic algorithm and Big Bang-Big Crunch algorithm), and compare it with traditional brute-force search methods.

The Importance of Coordinate Frames in Dynamic SLAM

Jesse Morris, V. Ila

Simultaneous Localization and Mapping

🎯 What it does: This paper systematically analyzes the two main schemes for representing dynamic object points in dynamic SLAM (world coordinate system and object coordinate system), and points out which scheme is more accurate and robust; meanwhile, it proposes a frontend-agnostic evaluation framework, using GTSAM to verify and compare various dynamic SLAM approaches.

The Joint-Space Reconstruction of Human Fingers by using a Highly Under-Actuated Exoskeleton

Yuan Su, Jiming Chen

Robotic Intelligence

🎯 What it does: Reconstructed human finger joint angles using a highly underactuated hand exoskeleton and a curved fitting algorithm, achieving high-precision joint configurations based on inverse kinematics.

The LuViRA Dataset: Synchronized Vision, Radio, and Audio Sensors for Indoor Localization

Ilayda Yaman, Liang Liu

Pose EstimationSimultaneous Localization and MappingMultimodalityBenchmark

🎯 What it does: Proposed and released the LuViRA dataset, which synchronously records data from visual, RF, audio, and IMU sensors along with 6DOF pose ground truth.

The New Dexterity Modular, Dexterous, Anthropomorphic, Open-Source, Bimanual Manipulation Platform: Combining Adaptive and Hybrid Actuation Systems with Lockable Joints

Che-Ming Chang, Minas Liarokapis

Robotic Intelligence

🎯 What it does: Proposed a new modular, dexterous, human-like dual-arm manipulation platform applicable to robot grasping, dexterous manipulation, and bimanual coordination experiments.

The Price of a Safe Flight: Risk Cost Based Path Planning

Aliaksei Pilko, James P. Scanlan

Autonomous DrivingOptimizationSafty and Privacy

🎯 What it does: Propose a risk-aware UAV path planning method using monetary value as the sole cost metric, generating non-uniform cost maps with a third-party ground risk model, and modifying the A* heuristic search for path planning.

The Un-Kidnappable Robot: Acoustic Localization of Sneaking People

Mengyu Yang, James Hays

Robotic IntelligenceVideoMultimodalityAudio

🎯 What it does: Collect a robot dataset with high-quality four-channel audio and 360° RGB data, train a model to detect and locate nearby moving people using only audio, and implement silent moving person tracking on a robot.

The Virtues of Laziness: Multi-Query Kinodynamic Motion Planning with Lazy Methods

Anuj Pasricha, Alessandro Roncone

Robotic Intelligence

🎯 What it does: Introduce LazyBoE, a multi-query kinodynamic motion planning method that uses forward propagation to simultaneously explore the robot's state space and control space;

Thermal Voyager: A Comparative Study of RGB and Thermal Cameras for Night-Time Autonomous Navigation

NG Aditya, Zubin Jacob

Autonomous DrivingImageVideo

🎯 What it does: Proposed and implemented an end-to-end nighttime autonomous navigation system called Thermal Voyager, which uses infrared thermal vision for passive perception to generate driving trajectories, and then determines the optimal steering angle through model predictive control.

Thermally-activated Biochemically-sustained Reactor for Soft Fluidic Actuation

Jialun Liu, D. D. Damian

Robotic IntelligencePhysics Related

🎯 What it does: Constructed a thermally activated bioreactor using biocompatible hydrogel valves to regulate saccharomyces yeast chemical reactions, generating pressure-driven fluid soft effectors.

Thermoformed electronic skins for conformal tactile sensor arrays

Peng Lu, W. Lee

🎯 What it does: Investigated a thermal forming technology for manufacturing electronic skin that can conform to curved surfaces and has tactile sensing capabilities, with detailed characterization of sensor performance, repeatability, and uniformity.

Thin-film NiTi Microactuator With A Magnetic Spring For A Tiny Launcher Mechanism

Sukjun Kim, Sarah Bergbreiter

Robotic IntelligencePhysics Related

🎯 What it does: Propose a thin film NiTi shape memory alloy micro actuator utilizing magnetic springs and integrate it into a micro launch device

Think, Act, and Ask: Open-World Interactive Personalized Robot Navigation

Yinpei Dai, Joyce Chai

Robotic IntelligenceLarge Language ModelAgentic AIVision-Language-Action ModelText

🎯 What it does: Propose a framework named ORION that enables robots to identify and navigate to user-specific target objects through natural language dialogue in unknown environments, addressing the issue that traditional ZSON only focuses on general objects and lacks interaction.

Through the Real World Haze Scenes: Navigating the Synthetic-to-Real Gap in Challenging Image Dehazing

Shijie Chen, Tao Chen

RestorationDomain AdaptationAuto EncoderGenerative Adversarial NetworkImage

🎯 What it does: Proposes a method that combines low-level features with deep features, utilizes a pre-trained vector quantized GAN to generate detail patches, employs a decoder with normalization modules, enhances feature matching through controllable operations, and finally improves visibility of real-world foggy images by dehazing and enhancing via gamma correction and multi-exposure image fusion.

Tight Fusion of Odometry and Kinematic Constraints for Multiple Aerial Vehicles in Physical Interconnection

Yingjun Fan, Yiqun Dong

OptimizationRobotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Propose a distributed multi-robot visual-inertial-rangefinder odometry system, fusing velocity and attitude constraints between the robots to improve localization accuracy.

Tight Motion Planning by Riemannian Optimization for Sliding and Rolling with Finite Number of Contact Points

Dror Livnat, Dan Halperin

OptimizationRobotic Intelligence

🎯 What it does: Proposed a compact motion planning method using Riemannian optimization to achieve sliding and rolling, capable of navigating through narrow passages in configuration space and smoothly switching between free and semi-free regions.

Tightly Coupled Range Inertial Localization on a 3D Prior Map Based on Sliding Window Factor Graph Optimization

Kenji Koide, A. Banno

OptimizationSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes a range inertial localization algorithm based on 3D prior maps, tightly coupling scan-to-scan and scan-to-map point cloud registration factors with IMU factors on a sliding window factor graph.

Tightly-Coupled LiDAR-Visual-Inertial SLAM and Large-Scale Volumetric Occupancy Mapping

Simon Boche, Stefan Leutenegger

Autonomous DrivingOptimizationSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes a fully tight-coupled LiDAR-visual-inertial SLAM system and 3D mapping framework, achieving scalability for large-scale environments through a local subgraph strategy.

Time-Optimal Gate-Traversing Planner for Autonomous Drone Racing

Chao Qin, Hugh H. T. Liu

OptimizationRobotic Intelligence

🎯 What it does: Designed a time-optimal gate crossing planner that accurately models various gate configurations to achieve faster trajectories

TinyMPC: Model-Predictive Control on Resource-Constrained Microcontrollers

Anoushka Alavilli, Zachary Manchester

OptimizationComputational EfficiencyBenchmark

🎯 What it does: Developed TinyMPC, a high-speed model predictive control (MPC) solver designed for microcontrollers.

Tip-Clutching Winch for High Tensile Force Application with Soft Growing Robots

O. G. Osele, H. Asada

Robotic Intelligence

🎯 What it does: A top clutch pulley was designed, allowing the vine-like soft growing robot to actively and reversibly attach and carry loads without being restricted by the tip device.

TiV-ODE: A Neural ODE-based Approach for Controllable Video Generation From Text-Image Pairs

Yucheng Xu, Zhibin Li

GenerationImageVideoTextOrdinary Differential Equation

🎯 What it does: Proposed a framework called TiV-ODE for generating controllable videos from text-image pairs.

ToP-ToM: Trust-aware Robot Policy with Theory of Mind

Chuang Yu, Angelo Cangelosi

Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: In multi-agent collaborative scenarios, the study investigates trustable robot strategies based on Theory of Mind (ToM), exploring how robots can infer human trust beliefs (true and false beliefs) to balance improving team performance and avoiding trust collapse.

Topological Exploration using Segmented Map with Keyframe Contribution in Subterranean Environments

Boseong Kim, D. H. Shim

Simultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposes a method that utilizes 3D dense maps to generate segmented exploration regions (SER) and employs a global perspective to generate a topological exploration approach, suitable for large underground environments.

Torque Transmission in Double-Tendon Sheath Driven Actuators for Application in Exoskeletons

Daniel Pérez-Suay, Sami Haddadin

Robotic Intelligence

🎯 What it does: Studied the torque transmission of dual-line cable-driven actuators in exoskeletons, explored the effects of pre-tension and cable shape on friction and torque efficiency, and constructed a self-made test platform integrating actuator units, pulleys, and a novel pre-tension mechanism for experimental testing.

Touch-Based Manipulation with Multi-Fingered Robot using Off-policy RL and Temporal Contrastive Learning

N. Morihira, Takayuki Osa

Representation LearningRobotic IntelligenceReinforcement LearningContrastive Learning

🎯 What it does: Proposed a time contrastive learning-based offline reinforcement learning method for manipulation tasks of multi-fingered robots under tactile perception.

Toward a framework integrating augmented reality and virtual fixtures for safer robot-assisted lymphadenectomy

Ziyang Chen, E. Momi

Safty and PrivacyRobotic IntelligenceBiomedical Data

🎯 What it does: Proposed and implemented a framework that integrates augmented reality (AR) with virtual devices for robot-assisted lymph node dissection, utilizing AR to visualize hidden lymph nodes and force feedback to avoid collisions with blood vessels.

Toward Accurate Camera-based 3D Object Detection via Cascade Depth Estimation and Calibration

Chaoqun Wang, Ruimao Zhang

Object DetectionDepth EstimationTransformerImageBenchmark

🎯 What it does: Proposes a cascaded depth estimation and calibration framework to enhance the accuracy of camera-based 3D object detection.

Toward Automated Programming for Robotic Assembly Using ChatGPT

Annabella Macaluso, Sachin Chitta

Robotic IntelligenceAI Code AssistantTransformerLarge Language Model

🎯 What it does: Using ChatGPT to automate robotic assembly programming by decomposing complex tasks to generate control code, simulating execution, and debugging errors.

Toward Grounded Commonsense Reasoning

Minae Kwon, Dorsa Sadigh

Robotic IntelligenceTransformerLarge Language ModelVision Language ModelImageBenchmark

🎯 What it does: Propose a framework that leverages large language models (LLM) and vision-language models (VLM) to enable robots to actively perceive the environment, achieving real-world common-sense reasoning to complete desktop cleaning tasks.