IROS 2023 Papers — Page 11
IEEE/RSJ International Conference on Intelligent Robots and Systems · 1195 papers
Sharing the Control of Robot Swarms Among Multiple Human Operators: A User Study
Genki Miyauchi, Roderich Groß
Robotic Intelligence
🎯 What it does: Evaluate the feasibility of multi-operator shared robot swarm control and the impact of different communication methods on performance and human factors.
Sim-to-Real Vision-Depth Fusion CNNs for Robust Pose Estimation Aboard Autonomous Nano-quadcopters
Luca Crupi, D. Palossi
Pose EstimationConvolutional Neural NetworkImagePoint Cloud
🎯 What it does: Investigated the use of a CNN that fuses depth and image data for human pose estimation on an automated nano quadrotor drone.
Simultaneous Action and Grasp Feasibility Prediction for Task and Motion Planning Through Multi-Task Learning
Smail Ait Bouhsain, Thierry Siméon
Robotic Intelligence
🎯 What it does: Through multi-task learning neural networks, AGFP-Net predicts action feasibility and the feasibility of multiple grasp types, using this information as heuristic input for symbolic planners to guide them toward geometrically feasible solutions, significantly reducing geometric planning time; meanwhile, an improved feasibility-driven TAMP algorithm is proposed, capable of handling more complex problems and incompletely specified goals.
Simultaneous Survey and Inspection with Autonomous Underwater Vehicles
J. McMahon, E. Plaku
OptimizationRobotic Intelligence
🎯 What it does: Proposes a planning method that combines sample-based motion planning with time-constrained road network constraints, equipped with a real-time execution framework for synchronous survey and inspection tasks of multiple AUVs.
Single Channel Soft Robotic Actuator Leveraging Switchable Strain-Limiting Structures for Deep-Sea Suction Sampling
Jan Peters, Annika Raatz
Robotic Intelligence
🎯 What it does: Design and manufacture a single-channel soft robotic actuator capable of bending in six directions and absorbing process forces, embedding low-melting-point alloy (LMPA) as a switchable strain-limiting structure to achieve 40° six-way planar bending and 30% elongation, while incorporating a locking function to improve energy efficiency.
Skill Generalization with Verbs
Rachel Ma, G. Konidaris
Robotic IntelligenceReinforcement LearningText
🎯 What it does: Proposes a method to generalize robot manipulation skills using verbs; trains a probabilistic classifier to determine whether a given object trajectory can be described by a specific verb, and performs policy search on object dynamics to generate trajectories that conform to the verb description; subsequently uses the generated trajectories as input for motion planning to execute tasks on new objects.
SkiROS2: A Skill-Based Robot Control Platform for ROS
Matthias Mayr, Volker Krüger
Robotic IntelligenceWorld Model
🎯 What it does: Introduces SkiROS2, a ROS-based skill-based robotic control platform, proposes a hierarchical hybrid control structure for automatic task planning and reactive execution, and performs world state reasoning through a knowledge base.
Skirting Line Estimation Using Sparse to Dense Deformation
Daniel Perez Banuelos, A. Alempijevic
SegmentationOptical FlowImage
🎯 What it does: Automatically extract the boundary line of wool fibers (i.e., the separation line between clean and contaminated wool) using RGB images, achieved through a sparse-to-dense deformation method;
SLAM and Shape Estimation for Soft Robots
Mohammadtaghi Karimi, Matthew Spenko
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Developed a soft robot mobile SLAM technology using onboard local sensors, achieving autonomous localization and map construction on planar boundary-constrained swarm robots.
Sliding Touch-Based Exploration for Modeling Unknown Object Shape with Multi-Fingered Hands
Yiting Chen, Yasemin Bekiroglu
OptimizationRobotic IntelligenceImagePoint Cloud
🎯 What it does: Propose a multi-finger sliding touch strategy utilizing a multi-finger hand equipped with tactile sensors and only local depth camera views, achieving efficient exploration and reconstruction of unknown three-dimensional object shapes through Bayesian optimization, and introducing a single leader-multi follower strategy to realize smooth local surface perception with multiple fingers.
SMART-Degradation: A Dataset for LiDAR Degradation Evaluation in Rain
Chen Zhang, Daniela Rus
Data SynthesisAutonomous DrivingSimultaneous Localization and MappingPoint CloudBenchmark
🎯 What it does: Released a natural driving dataset and toolbox containing LiDAR scan pairs under rainy and sunny conditions, providing mapping, localization, and scan synthesis functions to promote research on LiDAR degradation in rainy weather.
SMART-Rain: A Degradation Evaluation Dataset for Autonomous Driving in Rain
Chen Zhang, Daniela Rus
Object DetectionAutonomous DrivingImageMultimodalityPoint CloudBenchmark
🎯 What it does: Proposes a novel dataset specifically for rainy-weather autonomous driving, along with three benchmark evaluation tasks for rainfall intensity estimation, LiDAR degradation estimation, and 2D object detection.
Smooth Stride Length Change of Rat Robot with a Compliant Actuated Spine Based on CPG Controller
Yuhong Huang, A. Knoll
Robotic Intelligence
🎯 What it does: Investigate the gait of rodent-like robots with a flexible spine, and propose a CPG model based on spinal flexion and a novel circular limit cycle oscillator
Social Triangles and Aggressive Lines: Multi-Robot Formations Impact Navigation and Approach
Alexandra Bacula, Heather Knight
Robotic Intelligence
🎯 What it does: Investigated differences in audience perception of social cues and sense of safety when three robots navigate through public spaces and approach humans under four geometric formations (wedge, V-shape, vertical line, horizontal line), and evaluated the robustness of path planning and user experience through two experiments.
Soft Cap for Vine Robots
Cem Suulker, K. Althoefer
Robotic Intelligence
🎯 What it does: Designed and tested a soft, fully fabric-made cylindrical top cap that can easily slide onto the front end of a vine-like robot to maintain loads such as sensors.
Soft Optical Sensor and Haptic Feedback System for Remote and Robot-Assisted Palpation
Arincheyan Gerald, Sheila Russo
Robotic IntelligenceBiomedical Data
🎯 What it does: Designed and implemented a system integrating a soft optical sensor with a wearable tactile glove for tumor detection during robotic palpation.
Soft Robot Shape Estimation: A Load-Agnostic Geometric Method
Christian Sorensen, Marc D. Killpack
Robotic Intelligence
🎯 What it does: Proposed a novel kinematic representation method for soft continuum robots, achieving full shape estimation using a pure geometric solution.
Soft, Modular, Shape-Changing Displays with Hyperelastic Bubble Arrays
Matthew R. Devlin, Amirhossein H. Memar
🎯 What it does: This paper designs and implements a flexible deformation display skin based on a modular super-elastic bubble array for tactile display and wearable devices.
SoftGPT: Learn Goal-Oriented Soft Object Manipulation Skills by Generative Pre-Trained Heterogeneous Graph Transformer
Junjia Liu, Fei Chen
Robotic IntelligenceGraph Neural NetworkTransformerLarge Language ModelGraphChain-of-Thought
🎯 What it does: This paper proposes a pre-trained soft object manipulation skill learning model called SoftGPT, which utilizes 3D heterogeneous graph representations and a GPT-based dynamic generation model for pre-training. For each downstream task, goal-oriented policies are trained, and during policy decision-making, SoftGPT generates action consequences to form a robot's 'thinking process,' providing roll-out trajectories to facilitate policy learning.
SOLL-E: A Module Transport and Placement Robot for Autonomous Assembly of Discrete Lattice Structures
In-Won Park, Kenneth C. Cheung
Robotic Intelligence
🎯 What it does: Designed and developed a transport and placement robot capable of autonomous assembly of discrete lattice structure blocks;
Sonar2Depth: Acoustic-Based 3D Reconstruction Using cGANs
Nael Jaber, Frank Kirchner
Image TranslationDepth EstimationGenerative Adversarial NetworkAudio
🎯 What it does: Using conditional generative adversarial networks (cGAN) to convert acoustic images into images containing depth information, achieving acoustic 3D reconstruction.
Soy: An Efficient MILP Solver for Piecewise-Affine Systems
Haoze Wu, Clark W. Barrett
OptimizationComputational EfficiencyBenchmark
🎯 What it does: Implemented an efficient MILP solver named Soy, specifically designed for solving one-hot constraints in PWA systems.
Sparse Dense Fusion for 3D Object Detection
Yulu Gao, Hongyang Li
Object DetectionAutonomous DrivingTransformerImagePoint CloudBenchmark
🎯 What it does: This paper proposes a sparse-dense fusion framework (SD-Fusion) to improve the fusion of cameras and LiDAR, thereby enhancing 3D object detection performance.
Spatial Reasoning via Deep Vision Models for Robotic Sequential Manipulation
Hongyou Zhou, Ozgur S. Oguz
Robotic IntelligenceConvolutional Neural NetworkTransformerWorld ModelImage
🎯 What it does: Propose using deep vision models (Vision Transformer and ResNet) as heuristic methods for decision-making in robot sequential manipulation tasks.
Spatio-Temporal Attention Network for Persistent Monitoring of Multiple Mobile Targets
Yizhuo Wang, Guillaume Sartoretti
Robotic IntelligenceTransformerReinforcement Learning
🎯 What it does: This study proposes a deep reinforcement learning method based on an attention network for persistent monitoring of moving targets, where the robot adaptively plans paths to revisit all targets and update their position estimates.
SpeedFormer: Learning Speed Profiles with Upper and Lower Boundary Constraints Based on Transformer
Kyuhwan Yeon, Seong-Gyun Jeong
Autonomous DrivingOptimizationTransformerSupervised Fine-TuningTime Series
🎯 What it does: Proposed a Transformer-based network to predict the coefficients of a fifth-degree polynomial, thereby generating speed profiles for autonomous vehicles;
Spiking Reinforcement Learning with Memory Ability for Mapless Navigation
Bo Yang, Huajin Tang
Autonomous DrivingRecurrent Neural NetworkSpiking Neural NetworkReinforcement Learning
🎯 What it does: Propose a multi-critic DDPG with spiking memory framework (MC-DDPGSM) for mapless navigation in dynamic and partially observable environments.
SpinDOE: A Ball Spin Estimation Method for Table Tennis Robot
T. Gossard, A. Zell
Pose EstimationRobotic IntelligenceConvolutional Neural NetworkVideo
🎯 What it does: Proposed and implemented a method for estimating the rotation of a billiard ball based on point array pose estimation.
SPONGE: Sequence Planning with Deformable-ON-Rigid Contact Prediction from Geometric Features
Tran Nguyen Le, Ville Kyrki
OptimizationRobotic IntelligencePoint Cloud
🎯 What it does: Proposed SPONGE, a sequence planning pipeline based on a deep learning contact prediction model, for predicting interactions between flexible and rigid objects and generating dishwashing trajectories.
SSC-RS: Elevate LiDAR Semantic Scene Completion with Representation Separation and BEV Fusion
Jianbiao Mei, Yong Liu
SegmentationAutonomous DrivingRepresentation LearningConvolutional Neural NetworkPoint Cloud
🎯 What it does: Propose a network called SSC-RS for outdoor LiDAR semantic scene completion, combining representation separation and BEV fusion techniques.
SSGM: Spatial Semantic Graph Matching for Loop Closure Detection in Indoor Environments
Yujie Tang, Yufeng Yue
Graph Neural NetworkSimultaneous Localization and MappingGraph
🎯 What it does: Proposed a spatial semantic graph matching method (SSGM) for loop closure detection in indoor environments, achieving loop closure detection by aligning semantic graphs and evaluating spatial distribution similarity.
Stable Real-Time Feedback Control of a Pneumatic Soft Robot
Sean Even, Yasemin Ozkan-Aydin
OptimizationRobotic Intelligence
🎯 What it does: Achieved real-time implementation of an infinite-dimensional feedback controller based on PDEs under finite actuators, and realized feasible soft robot control by real-time adjustment of feedback gain through convex quadratic programming (QP)
Stackelberg Meta-Learning for Strategic Guidance in Multi-Robot Trajectory Planning
Yuhan Zhao, Quanyan Zhu
Robotic IntelligenceMeta Learning
🎯 What it does: This paper proposes a meta-learning framework based on the Stackelberg game for collaborative guidance of leader-follower robots in multi-trajectory planning.
Staged Contact Optimization: Combining Contact-Implicit and Multi-Phase Hybrid Trajectory Optimization
M. Turski, Aaron M. Johnson
OptimizationRobotic Intelligence
🎯 What it does: Propose a Staged Contact Optimization (SCO) algorithm that combines contact-implicit optimization (CIO) with multi-stage hybrid trajectory optimization (HTO) for legged robot trajectory optimization;
Stair Climbing Using the Angular Momentum Linear Inverted Pendulum Model and Model Predictive Control
O. Dosunmu-Ogunbi, J. Grizzle
Physics Related
🎯 What it does: Studied an improved ALIP model and combined virtual constraint control with model predictive control to achieve stable stair climbing gait;
State- Based Control for an Actuated Reciprocal Gait Orthosis
Simon Eckstein, C. D. Remy
OptimizationRobotic Intelligence
🎯 What it does: Extended IRGO by incorporating a single actuator and achieving stable mutual gait using a Hybrid Zero Dynamics controller
Statistical Characterization of Position-Dependent Behavior Using Frequency-Aware B-Spline
Y. Al-Rawashdeh, M. Janaideh
Autonomous DrivingOptimizationTime Series
🎯 What it does: A frequency-aware B-spline basis function is studied to generate reference motion trajectories that satisfy motion system constraints, and data is collected through frequency-related random walks to statistically analyze position-related localization errors.
Step Toward Deploying the Torque-Controlled Robot TALOS on Industrial Operations
Côme Perrot, O. Stasse
OptimizationRobotic Intelligence
🎯 What it does: In industrial manufacturing environments, using the torque-controlled bipedal robot TALOS, it is demonstrated that tools can be accurately inserted into aircraft structural holes with millimeter-level precision through whole-body model predictive control (WBMPC).
Stereo Visual Odometry with Deep Learning-Based Point and Line Feature Matching Using an Attention Graph Neural Network
Shenbagaraj Kannapiran, Gaurav Pandey
Pose EstimationAutonomous DrivingGraph Neural NetworkSimultaneous Localization and Mapping
🎯 What it does: Proposes a stereo visual odometry (StereoVO) technique based on point-line features, utilizing an attention graph neural network to achieve robust feature matching, applicable in low visibility weather conditions such as fog, haze, rain, and snow, as well as dynamic lighting environments like nighttime/glare.
STL: Surprisingly Tricky Logic (for System Validation)
H. Siu, Makai Mann
Explainability and Interpretability
🎯 What it does: Conducted a human experiment to evaluate the accuracy of people's judgments on whether Signal Temporal Logic (STL) constraints can ensure safety and complete tasks in a grid world label capture task.
Streaming Motion Forecasting for Autonomous Driving
Ziqi Pang, Yu-Xiong Wang
Autonomous DrivingTime SeriesSequentialBenchmark
🎯 What it does: Propose a continuous streaming trajectory prediction benchmark and introduce a meta-algorithm called 'Predictive Streamer' compatible with existing prediction models to handle occluded agents and maintain temporal consistency.
Stroke-Based Rendering and Planning for Robotic Performance of Artistic Drawing
Ivaylo Ilinkin, Young J. Kim
SegmentationGenerationDepth EstimationRobotic IntelligenceImage
🎯 What it does: Propose a robot painting system based on stroke-based rendering (SBR) that can generate paintings similar to the input image with an artistic style, and imitate the planning process of human artists through a reasonable brush order
Structure from Action: Learning Interactions for 3D Articulated Object Structure Discovery
Neil Nie, Shuran Song
Reinforcement LearningVision-Language-Action Model
🎯 What it does: Proposes the 'Structure from Action' (SfA) framework, which infers 3D part geometry and joint parameters of unseen movable objects through a series of inferred interaction sequences.
Subtask Aware End-to-End Learning for Visual Room Rearrangement
Youngho Kim, Jong-Hwan Kim
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed and implemented a vision-based room rearrangement agent based on subtask prediction, achieving subtask-aware end-to-end learning through OSPNet and SAPNet, with the agent trained in simulation.
SUIT: Learning Significance-Guided Information for 3D Temporal Detection
Zheyuan Zhou, Li Zhang
Object DetectionAutonomous DrivingPoint Cloud
🎯 What it does: Propose the SUIT method, which achieves temporal fusion of sparse features through a significant sampling mechanism and geometric transformation learning for 3D LiDAR object detection.
Sunram 7: An MR Safe Robotic System for Breast Biopsy
Harsh Ranjan, Stefano Stramigioli
Robotic IntelligenceBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Developed an MR-safe breast biopsy robot Sunram 7 based on pneumatic stepper motors.
Superpixel Transformers for Efficient Semantic Segmentation
Alex Zhu, Henrik Kretzschmar
SegmentationTransformerImage
🎯 What it does: Propose a Transformer model based on superpixels for efficient semantic segmentation. First, the pixel space is reduced to a superpixel space through local cross-attention, then multi-head self-attention is applied in the superpixel space, and finally, the superpixel class predictions are projected back to the pixel space.
Surface Navigation of Alginate Artificial Cells in Mucus Solutions
L. Rogowski, Min Jun Kim
Robotic Intelligence
🎯 What it does: Artificial cells were prepared by crosslinking sodium alginate with magnetic nanoparticles, and their propulsion capability was evaluated in mucus solution. The study investigated the effects of simplified gastric fluids, artificial cell properties, and magnetic field characteristics on surface movement, while comparing automatic feedback control with manual open-loop operation.
Swarm of One: Bottom-Up Emergence of Stable Robot Bodies from Identical Cells
T. Smith, Yu Gu
Robotic Intelligence
🎯 What it does: Propose a bottom-up robotic morphology design method, construct the Loopy 'single-body multi-body' multi-morphology robot test platform, and achieve spontaneous formation of symmetric protrusions in robot bodies from homogeneous cells through self-organization and morphological computation.
Swashplateless-Elevon Actuation for a Dual-Rotor Tail-Sitter VTOL UAV
Nan Chen, Fu Zhang
Robotic Intelligence
🎯 What it does: Proposed and verified the screwless elevator actuation (SEA) for dual-rotor tail-sitting VTOL UAVs, and conducted comparative experiments with the conventional elevator actuation (CEA).
Switching Head-Tail Funnel UNITER for Dual Referring Expression Comprehension with Fetch-and-Carry Tasks
Ryosuke Korekata, Komei Sugiura
Robotic IntelligenceVision Language ModelMultimodality
🎯 What it does: Propose a model based on Switching Head-Tail Funnel UNITER to solve target recognition and destination localization for service robots' grasping and carrying tasks.
Symbolic State Space Optimization for Long Horizon Mobile Manipulation Planning
Xiaohan Zhang, Shiqi Zhang
OptimizationRobotic Intelligence
🎯 What it does: Propose the Symbolic State Space Optimization (S3O) method, which automatically computes abstract positions and their 2D geometric foundations to generate task-motion plans for long-term mobile manipulation planning.
Symmetry-Based Modeling and Hybrid Orientation-Force Control of Wearable Cutaneous Haptic Device
Somang Lee, Dongjun Lee
Robotic Intelligence
🎯 What it does: Proposed a modeling method based on symmetry and a hybrid pose-force control framework for wearable skin-based tactile devices (CHD), achieving precise control of three degrees of freedom contact force at the fingertip and robustness to user differences.
System Identification and Control of Front-Steered Ackermann Vehicles Through Differentiable Physics
Burak M. Gonultas, Volkan Isler
Autonomous DrivingOptimizationPhysics Related
🎯 What it does: Developed a differentiable physics simulator (DPS) for system identification and control of front-wheel steering Ackermann vehicles, and obtained vehicle parameters through gradient learning; subsequently, implemented a feedback controller on a real F1TENTH vehicle to achieve stable lane keeping.
T-Top, an Open Source Tabletop Robot with Advanced Onboard Audio, Vision and Deep Learning Capabilities
Marc-Antoine Maheux, François Michaud
Robotic IntelligenceImageMultimodalityAudio
🎯 What it does: This paper proposes and open-releases the desktop robot T-Top, equipped with advanced audio and visual processing and deep learning networks, aiming to provide the elderly with richer interactive modes.
T-UDA: Temporal Unsupervised Domain Adaptation in Sequential Point Clouds
Awet Haileslassie Gebrehiwot, Tomáš Svoboda
SegmentationDomain AdaptationPoint Cloud
🎯 What it does: Proposed an unsupervised domain adaptation framework combining temporal point cloud geometric consistency with the mean teacher method to enhance the robustness of 3D semantic segmentation.
T2FPV: Dataset and Method for Correcting First-Person View Errors in Pedestrian Trajectory Prediction
Ben Stoler, Jean Oh
Data SynthesisAutonomous DrivingVideoBenchmark
🎯 What it does: Proposed the T2FPV method to generate high-fidelity first-person view (FPV) datasets from real-world top-down trajectory data, designed the CoFE module for end-to-end correction of FPV errors in trajectory prediction algorithms, and released the T2FPV-ETH dataset along with software tools.
Target Attribute Perception Based UAV Real-Time Task Planning in Dynamic Environments
Jinhong He, Chao Cai
OptimizationRobotic Intelligence
🎯 What it does: A comprehensive solution is proposed to enable drones to autonomously fly in complex dynamic environments. The solution employs deep learning for 3D dynamic environment perception, models dynamic targets, and integrates them into a static grid occupancy map. A gradient field is constructed based on attribute information, and an adaptive planning method based on gradient optimization is designed. The method automatically adjusts planning frequency, uses manually constructed gradients instead of signed distance fields, and integrates the solution into a custom quadrotor for field testing.
Task and Configuration Space Compliance of Continuum Robots via Lie Group and Modal Shape Formulations
A. Orekhov, N. Simaan
Robotic Intelligence
🎯 What it does: Derive analytic compliance formulations for continuum robots modeled as Kirchhoff rods, using modal approximations and Lie group integration, and validate them experimentally on a tendon-actuated segment.
Task and Motion Planning with Large Language Models for Object Rearrangement
Yan Ding, Shiqi Zhang
Robotic IntelligenceLarge Language ModelPrompt EngineeringText
🎯 What it does: Proposed and implemented LLM-GROP for multi-object rearrangement tasks, converting natural language commands into human-consistent object rearrangement plans.
Task Assignment, Scheduling, and Motion Planning for Automated Warehouses for Million Product Workloads
Christopher Leet, Pierluigi Nuzzo
OptimizationRobotic Intelligence
🎯 What it does: Proposed and implemented a contract-based cyclic motion planning (CCMP) method to address task allocation, scheduling, and motion planning problems in large-scale automated warehouses.
Task Planning and Motion Control with Temporal Logic Specifications
Marcos S. Pereira, Bruno Vilhena Adorno
OptimizationRobotic Intelligence
🎯 What it does: Propose a task planning and motion control framework that can generate task plans satisfying linear temporal logic (LTL) specifications, and execute them through a task-space constraint motion controller and a local task planner, enabling free-flying robots to complete tasks while avoiding collisions and meeting control signal constraints.
Task-Oriented Grasp Prediction with Visual-Language Inputs
Chao Tang, Hong Zhang
Robotic IntelligenceVision Language ModelVision-Language-Action ModelMultimodality
🎯 What it does: Proposed a task-oriented grasping prediction method called GraspCLIP based on vision-language input, addressing the limitation of previous approaches that only focused on object localization, and enabling simultaneous object localization and task-oriented grasping.
Task-Oriented Grasping with Point Cloud Representation of Objects
Aditya Patankar, IV Ramakrishnan
Robotic IntelligenceConvolutional Neural NetworkSupervised Fine-TuningPoint Cloud
🎯 What it does: Proposed a task-oriented grasping method based on partial point clouds, utilizing an eye-in-hand camera configuration to achieve object grasping and subsequent operations.
Task2Morph: Differentiable Task-Inspired Framework for Contact-Aware Robot Design
Yishuai Cai, Wenjing Yang
OptimizationRobotic Intelligence
🎯 What it does: Proposes a differentiable task-inspired framework called Task2Morph for simultaneously optimizing robot morphology and controller while considering contact.
Team Coordination on Graphs with State-Dependent Edge Costs
Sara Oughourli, Daigo Shishika
OptimizationGraph
🎯 What it does: Propose handling team collaboration problems in a graph environment by introducing supporting actions to reduce the traversal cost of high-cost edges, thereby transforming multi-agent path planning into single-agent planning on a joint state graph.
Team Northeastern's Approach to ANA XPRIZE Avatar Final Testing: A Holistic Approach to Telepresence and Lessons Learned
Rui Luo, J. P. Whitney
Robotic Intelligence
🎯 What it does: Reports on the Avatar system developed by the Northeast team and their overall approach in the ANA Avatar XPRIZE Final test
Template Model Inspired Task Space Learning for Robust Bipedal Locomotion
Guillermo A. Castillo, Ayonga Hereid
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose a hierarchical framework that integrates a high-level planner based on reinforcement learning with a model-driven low-level controller to achieve online generation of task-space commands and trajectory tracking;
Temporal Logic-Based Intent Monitoring for Mobile Robots
Hansol Yoon, S. Sankaranarayanan
Robotic IntelligenceSequential
🎯 What it does: Proposes an intent monitoring framework based on temporal logic specifications, predicting and monitoring robot intentions through passive observation of their behavior.
TemporalStereo: Efficient Spatial-Temporal Stereo Matching Network
Youming Zhang, Stefano Mattoccia
Depth EstimationImageVideo
🎯 What it does: Proposes an efficient coarse-to-fine layered stereo matching network called TemporalStereo, which leverages past geometric and context information to improve matching accuracy.
Tension Jamming for Deployable Structures
Daniel Hasegawa, Robert D. Howe
Physics Related
🎯 What it does: This paper designs and implements a composite beam based on tension jamming for large-scale deployable structures, constructs a bridge prototype that can be assembled by drones in the air, and demonstrates its performance of being soft during transportation and rigid after deployment.
Terrain-Aware Kinodynamic Planning with Efficiently Adaptive State Lattices for Mobile Robot Navigation in Off-Road Environments
Eric R. Damm, Thomas M. Howard
Autonomous DrivingOptimizationRobotic IntelligenceOrdinary Differential Equation
🎯 What it does: Proposes a terrain-aware dynamic planning method called KEASL for off-ground environments, based on an efficiently adaptive state lattice for mobile robot navigation.
Test-Time Adaptation for Point Cloud Upsampling Using Meta-Learning
A. Hatem, Yang Wang
Meta LearningSupervised Fine-TuningPoint Cloud
🎯 What it does: Propose a meta-learning based test-time adaptation method for point cloud upsampling, which fine-tunes the model during the testing phase to enhance generalization performance.
The Audio-Visual BatVision Dataset for Research on Sight and Sound
Amandine Brunetto, F. Moutarde
Robotic IntelligenceImageMultimodalityAudio
🎯 What it does: Collected an audio-visual dataset containing robot-generated sounds, echoes, and synchronized RGB-D images.
The Bystander Affect Detection (BAD) Dataset for Failure Detection in HRI
Alexandra W. D. Bremers, Wendy Ju
Anomaly DetectionData-Centric LearningRobotic IntelligenceConvolutional Neural NetworkVideoBenchmark
🎯 What it does: Created and collected 2,452 webcam videos from 54 participants corresponding to 46 fault stimulus videos, to capture bystanders' implicit reactions to human and robot errors, and trained a deep learning model BADNet to predict task failure occurrences.
The Effects of Robot Motion on Comfort Dynamics of Novice Users in Close-Proximity Human-Robot Interaction
Pierce Howell, H. Ravichandar
Robotic Intelligence
🎯 What it does: This study investigates the impact of motion characteristics such as robot workspace overlap, end-effector speed, and motion readability on the comfort of first-time users and their habituation over time through user experiments.
The Impact of Overall Optimization on Warehouse Automation
H. Yoshitake, P. Abbeel
OptimizationReinforcement Learning
🎯 What it does: Proposed a multi-agent reinforcement learning framework with centralized training and decentralized execution, applying a single shared critic in automated warehouses for robot collaborative task selection, and evaluated the performance improvement from global optimization.
The MyoPassivity Puzzle: How Does Muscle Fatigue Affect Energetic Behavior of the Human Upper-Limb During Physical Interaction with Robots?
S. Oliver, S. F. Atashzar
Robotic IntelligenceBiomedical Data
🎯 What it does: Investigate the effect of muscle fatigue on the energy absorption capacity (EoP) of the human wrist in physical human-robot interaction.
The Role of Absolute Positioning Error in Hand-Eye Calibration and Robotic Guidance Systems: An Analysis
Michal Chaluš, J. Liška
Robotic Intelligence
🎯 What it does: This paper describes the components of the robot guidance system (RGS) and proposes a calibration process model (MCP) for analyzing the impact of absolute positioning errors on hand-eye calibration, six-point calibration, and tool mutual transformation accuracy. Subsequently, the validity of this model is verified through simulation.
Thoracic Cartilage Ultrasound-CT Registration Using Dense Skeleton Graph
Zhongliang Jiang, N. Navab
Graph Neural NetworkPoint CloudBiomedical DataComputed TomographyUltrasound
🎯 What it does: Propose a non-rigid registration method based on a dense skeleton graph to align thoracic cartilage ultrasound with CT images, thereby transferring planning paths from a generic atlas to individual patients.
TidyBot: Personalized Robot Assistance with Large Language Models
Jimmy Wu, Thomas Funkhouser
Robotic IntelligenceLarge Language ModelVision-Language-Action ModelMultimodalityBenchmark
🎯 What it does: Propose a robot-assisted system that learns and generalizes user preferences through a few examples, used in household cleaning tasks to place objects at user-preferred locations.
Tight Collision Probability for UAV Motion Planning in Uncertain Environment
Tianyu Liu, Jia-Yu Pan
OptimizationRobotic Intelligence
🎯 What it does: Proposes a reliable UAV motion planning framework that incorporates various uncertainties into chance constraints, computes the collision probability between the robot's Gaussian distribution forward reachable sets and obstacle states, and provides a tight upper bound for this probability; generates motion primitives using approximate solutions and performs iterative optimization with exact solutions; verifies its reliability and effectiveness through simulations and real experiments.
Tightly-Coupled Visual- DVL- Inertial Odometry for Robot-Based Ice-Water Boundary Exploration
Lin Zhao, B. Loose
Robotic IntelligenceSimultaneous Localization and MappingMultimodality
🎯 What it does: Propose a tightly coupled multi-sensor fusion framework that uses visual, DVL, IMU, and pressure sensors to achieve robot localization in ice-water boundary exploration.
Tightly-Coupled Visual-DVL Fusion For Accurate Localization of Underwater Robots
Yupei Huang, Junzhi Yu
Robotic IntelligenceSimultaneous Localization and MappingImage
🎯 What it does: Proposes a tightly coupled Visual-Doppler Velocity Log (Visual-DVL) fusion method that directly integrates DVL speed measurements into visual odometry (VO) to achieve underwater robot localization.
Time to Danger, an Alternative to Passive Safety for the Locomotion of a Biped Robot in a Crowd
M. Ciocca, Thierry Fraichard
OptimizationRobotic Intelligence
🎯 What it does: Proposed and studied the concept of Time To Danger (TTD), and designed a novel gait planning strategy based on TTD, using recursive horizon model predictive control (MPC) to maximize balance maintenance and TTD.
Time-Optimal Control via Heaviside Step-Function Approximation
K. Pfeiffer, Quang Pham
Optimization
🎯 What it does: Proposed a nonlinear hierarchical least squares programming (NL-HLSP) for time-optimal control of nonlinear discrete dynamic systems.
Time-Optimal Path Tracking with ISO Safety Guarantees
Shohei Fujii, Quang Pham
OptimizationSafty and PrivacyRobotic Intelligence
🎯 What it does: Propose a safety control strategy based on time-optimal path parameterization (TOPP), ensuring that robots remain stationary during collisions under the ISO/TS 15066 SSM framework, validated in simulations where it outperforms existing methods, and introduce parallelized precomputation to achieve near real-time performance, finally demonstrated on a 6-degree-of-freedom industrial robot.
Time-Optimal Point-To-Point Motion Planning and Assembly Mode Change of Cuspidal Manipulators: Application to 3R and 6R Robots
Tobias Marauli, P. Wenger
OptimizationRobotic Intelligence
🎯 What it does: A singularity-free and time-optimal point-to-point trajectory planning method for cuspidal 3R and 6R robots is proposed, and it is applied to time-optimal singularity-free switching in assembly mode.
Timor Python: A Toolbox for Industrial Modular Robotics
Jonathan Külz, M. Althoff
OptimizationRobotic Intelligence
🎯 What it does: Proposed and implemented Timor, a Python toolbox for industrial modular robots (MRR), supporting model generation, task-based configuration optimization, URDF export, and seamless integration into existing simulation and optimization workflows.
TIMS: A Tactile Internet-Based Micromanipulation System with Haptic Guidance for Surgical Training
Jialin Lin, Dandan Zhang
Robotic IntelligenceBiomedical Data
🎯 What it does: Developed a micro-manipulation system TIMS based on the haptic internet for minimally invasive surgery training, providing real-time haptic feedback through wearable haptic displays and achieving data transmission via the ROS-Django framework.
TOP-UAV: Open-Source Time-Optimal Trajectory Planner for Point-Masses Under Acceleration and Velocity Constraints
Fabian Meyer, D. Sayah
Autonomous DrivingOptimization
🎯 What it does: Provides mathematical proofs for alternative methods, reveals the shortcomings of existing SOTA methods, and proposes an improved approach that better leverages UAV kinematics, which on average improves trajectory execution speed by about 14%.
Topology-Guided Perception-Aware Receding Horizon Trajectory Generation for UAVs
Gang Sun, Zhuang Yan
Autonomous DrivingOptimizationSimultaneous Localization and Mapping
🎯 What it does: Proposed a topology-guided perception-aware fallback field trajectory generation method aimed at real-time generation of perception-aware trajectories.
TopSpark: A Timestep Optimization Methodology for Energy-Efficient Spiking Neural Networks on Autonomous Mobile Agents
Rachmad Vidya Wicaksana Putra, Muhammad Shafique
Computational EfficiencySpiking Neural Network
🎯 What it does: Proposed the TopSpark method, which reduces computation time through adaptive time steps to achieve low-energy SNN processing during both training and inference phases.
Touch if it's Transparent! ACTOR: Active Tactile-Based Category-Level Transparent Object Reconstruction
P. Murali, Mohsen Kaboli
Pose EstimationRobotic Intelligence
🎯 What it does: Proposed an active tactile-based category-level transparent object reconstruction framework called AC-TOR, and achieved the corresponding pose estimation task.
Toward Closed-Loop Additive Manufacturing: Paradigm Shift in Fabrication, Inspection, and Repair
Manpreet Singh, Lu Li
Anomaly DetectionOptimizationRobotic IntelligencePoint Cloud
🎯 What it does: Proposed a hierarchical closed-loop additive manufacturing system that enhances the quality, tolerance, and reliability of 3D printed parts through real-time detection and online repair of geometric defects.
Toward Human-Like Social Robot Navigation: A Large-Scale, Multi-Modal, Social Human Navigation Dataset
Duc M. Nguyen, Xuesu Xiao (George Mason University)
Robotic IntelligenceMultimodality
🎯 What it does: Propose to utilize human natural social navigation data in public spaces to design a wearable panoramic multimodal sensor kit for data collection, and analyze and verify its usability in social robot navigation learning based on this dataset.
Toward Sub-Gram Helicopters: Designing a Miniaturized Flybar for Passive Stability
Kyle Johnson, Vikram Iyer
OptimizationHyperparameter Search
🎯 What it does: Designed and manufactured a 48mg miniature folded carbon fiber flyer bar, which couples the tilt of the flyer bar with the angle of attack of the rotor through flexible joints, thereby providing passive attitude stabilization for sub-gram helicopters.
Towards a Robust Adversarial Patch Attack Against Unmanned Aerial Vehicles Object Detection
Samridha Shrestha, Eduardo Viegas
Object DetectionAdversarial AttackImage
🎯 What it does: Propose a robust adversarial patch attack method for UAV target detection, considering factors such as UAV camera perspective, distance, and brightness variations;
Towards an Accurate Augmented-Reality-Assisted Orthopedic Surgical Robotic System Using Bidirectional Generalized Point Set Registration
Ang Zhang, M. Q. Meng
Robotic IntelligencePoint CloudBiomedical Data
🎯 What it does: Proposed an AR-assisted orthopedic surgery robot system based on a head-mounted display (HMD), which can overlay preoperative planning onto the patient's anatomy and provide surgical guidance.
Towards Automated Void Detection for Search and Rescue with 3D Perception
Ananya Bal, H. Choset
Anomaly DetectionImagePoint Cloud
🎯 What it does: Reconstruct 3D point clouds using multi-temporal aerial images and perform temporal stacking to detect voids in rubble heaps.