ICRA 2023 Papers — Page 7
IEEE International Conference on Robotics and Automation · 1341 papers
Hardware-in-the-Loop Simulator with Low-Thrust Actuator for Free-Flying Robot's Omni-Directional Control
D. Hirano, T. Saito
Robotic Intelligence
🎯 What it does: Established a hardware-in-the-loop (HIL) simulator integrated with low-thrust thrusters to verify the Guidance, Navigation, and Control (GNC) system for omnidirectional control of free-floating robots
HAT: Head-Worn Assistive Teleoperation of Mobile Manipulators
Akhil Padmanabha, Zackory Erickson
Robotic IntelligenceTime Series
🎯 What it does: Proposed a head-mounted wearable interface based on inertial measurement, enabling individuals with severe motor disabilities to remotely control a mobile manipulator to perform daily tasks through head movements.
Hazard Analysis of Collaborative Automation Systems: A Two-layer Approach based on Supervisory Control and Simulation
Tom P. Huck, Torsten Kroger
Safty and Privacy
🎯 What it does: Propose a two-layer approach that combines formal methods with simulation, using supervisory control theory to synthesize behaviors leading to unsafe states and employing them as inputs for simulation analysis.
Heading Control of a Long-Endurance Insect-Scale Aerial Robot Powered by Soft Artificial Muscles
Y. Hsiao, Yufeng Chen
Robotic Intelligence
🎯 What it does: Designed and demonstrated a micro-scale, soft muscle-driven long-endurance hover flight robot that achieved heading control.
Heading for the Abyss: Control Strategies for Exploiting Swinging of a Descending Tethered Aerial Robot
Max Polzin, Josie Hughes
Robotic Intelligence
🎯 What it does: Propose a suspended aerial robot using a wired dual-thruster wing suit, achieving stable flight and swinging through PD control and switch control to expand the workspace and navigate in environments with overhangs.
Heterogeneous Coverage and Multi-Resource Allocation in Supply-Constrained Teams
Mela C. Coffey, Alyssa Pierson
Optimization
🎯 What it does: Proposes a Voronoi-based coverage control method, utilizing a position and time-varying density function to distribute heterogeneous robots across multiple demand areas, achieving resource allocation.
HFT: Lifting Perspective Representations via Hybrid Feature Transformation for BEV Perception
Jiayu Zou, Xingang Wang
Object DetectionSegmentationAutonomous DrivingImage
🎯 What it does: Proposed a Hybrid Feature Transformation (HFT) module to lift front-view features to a bird's-eye view (BEV) representation, and validated its effectiveness in BEV perception tasks.
Hierarchical Adaptive Loco-manipulation Control for Quadruped Robots
M. Sombolestan, Quan Nguyen
OptimizationRobotic Intelligence
🎯 What it does: Proposes a hierarchical adaptive control framework enabling quadruped robots to complete localization and manipulation tasks under unknown object mass, unknown friction coefficients, and unknown slopes, achieving contact force regulation and robot posture maintenance through adaptive operational control and unified MPC.
Hierarchical Approach for Joint Semantic, Plant Instance, and Leaf Instance Segmentation in the Agricultural Domain
Gianmarco Roggiolani, C. Stachniss
SegmentationConvolutional Neural NetworkImageAgriculture Related
🎯 What it does: Propose a single convolutional neural network to simultaneously perform semantic segmentation, plant instance segmentation, and leaf instance segmentation on farmland RGB images.
Hierarchical Graph Neural Networks for Proprioceptive 6D Pose Estimation of In-hand Objects
Alireza Rezazadeh, Nawid Jamali
Pose EstimationGraph Neural NetworkMultimodality
🎯 What it does: Designed a hierarchical graph neural network to integrate visual and tactile data for 6D pose estimation of objects in the hand, incorporating proprioceptive information from fingertip positions.
Hierarchical Intention Tracking for Robust Human-Robot Collaboration in Industrial Assembly Tasks
Zhe Huang, K. Driggs-Campbell
Robotic Intelligence
🎯 What it does: Designed and implemented a collaborative robot system that simultaneously tracks high-level and low-level human intentions for industrial assembly tasks, demonstrated on the UR5e robot.
Hierarchical Policy Blending as Inference for Reactive Robot Control
Kay Hansel, G. Chalvatzaki
OptimizationRobotic Intelligence
🎯 What it does: Proposed a hierarchical motion generation method that combines reactive strategies with planning through probabilistic inference and stochastic optimization, computing optimal weights to form a weighted product model, thereby generating feasible reactive motion plans.
Hierarchical Whole-body Control of the cable-Suspended Aerial Manipulator endowed with Winch-based Actuation
Y. Sarkisov, K. Kondak
Robotic Intelligence
🎯 What it does: A winch-based suspension cable tension control scheme was designed and verified, utilizing three winch-controlled suspension cables to generate torque to reduce the gravitational torque of the robotic arm, and employing a hierarchical full-body controller to coordinate the end-effector pose maintenance and system center of gravity adjustment.
High Resolution Point Clouds from mmWave Radar
Akarsh Prabhakara, Anthony G. Rowe
GenerationPoint Cloud
🎯 What it does: Propose RadarHD, an end-to-end neural network that converts low-resolution millimeter-wave radar inputs into high-resolution LiDAR-like point clouds.
High-Speed High-Accuracy Spatial Curve Tracking Using Motion Primitives in Industrial Robots
Honglu He, J. Wen
OptimizationRobotic Intelligence
🎯 What it does: A systematic approach is proposed to optimize the motion parameters of industrial robots using motion primitives to achieve high-speed, high-precision spatial curve tracking.
High-Speed Scooping: An Implementation through Stiffness Control and Direct-Drive Actuation
K. Mak, Jungwon Seo
Robotic Intelligence
🎯 What it does: Proposed and implemented a high-speed extraction technique, achieving appropriate dynamic impact interactions between the robot, objects, and the environment under error and uncertainty; the implementation is based on stiffness control and a custom-made dual-finger direct drive gripper.
HMAAC: Hierarchical Multi-Agent Actor-Critic for Aerial Search with Explicit Coordination Modeling
Chuanneng Sun, D. Pompili
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposes the Hierarchical Multi-Agent Actor-Critic (HMAAC) framework for control and collaboration in UAV search and rescue tasks
Holistic Graph-based Motion Prediction
Daniel Grimm, J. M. Zöllner
Autonomous DrivingGraph Neural NetworkGraph
🎯 What it does: Proposes a motion prediction method for autonomous vehicles based on heterogeneous panoramic graphs.
Holistic view of Inverse Optimal Control by introducing projections on singularity curves
J. Colombel, F. Charpillet
Optimization
🎯 What it does: Proposed the Projected Inverse Optimal Control (PIOC) method to unify the understanding of inverse optimal control through projection.
Holo-Dex: Teaching Dexterity with Immersive Mixed Reality
Sridhar Pandian Arunachalam, Lerrel Pinto
Pose EstimationRepresentation LearningRobotic Intelligence
🎯 What it does: Developed a mixed reality framework called Holo-Dex, which enables teachers to remotely control robots in an immersive environment and collect demonstration data using commercial VR headsets and high-precision gesture estimators. Subsequently, various grasping and manipulation skills are learned through feature learning and non-parametric imitation.
Household Clothing Set and Benchmarks for Characterising End-Effector Cloth Manipulation
A. Clark, Austin Gregg-Smith
Robotic IntelligenceBenchmark
🎯 What it does: Constructed a home clothing benchmark dataset and designed four grasping and manipulation evaluation metrics to assess the performance of different robotic end-effectors in clothing grasping and folding tasks.
How Does It Feel? Self-Supervised Costmap Learning for Off-Road Vehicle Traversability
Mateo Guaman Castro, S. Scherer
Autonomous DrivingRobotic Intelligence
🎯 What it does: A self-supervised approach combines environmental perception information with terrain interaction feedback to learn the prediction of traversability cost maps for off-road vehicles, incorporating robot speed into the prediction pipeline.
HREyes: Design, Development, and Evaluation of a Novel Method for AUVs to Communicate Information and Gaze Direction*
Michael Fulton, Junaed Sattar
Robotic Intelligence
🎯 What it does: Designed, developed, and evaluated an optical biometric simulation communication device named HREyes for communication between AUVs and humans, incorporating active lucemes that explicitly convey information through animations and eye lucemes that express gaze direction by mimicking human eyes.
Human Non-Compliance with Robot Spatial Ownership Communicated via Augmented Reality: Implications for Human-Robot Teaming Safety
Christine T. Chang, Bradley Hayes
Safty and PrivacyRobotic IntelligenceVision-Language-Action ModelImage
🎯 What it does: Studied the use of augmented reality head-mounted displays to render colored grids on the ground for visual communication of spatial ownership in noisy environments with autonomous quadrotor robots, and explored participants' behavior and attitudes in this scenario.
Human-Guided Planning for Complex Manipulation Tasks Using the Screw Geometry of Motion
Dasharadhan Mahalingam, N. Chakraborty
Robotic IntelligenceReinforcement Learning from Human Feedback
🎯 What it does: Proposes a method for complex manipulation task planning that utilizes human demonstrations and the helical geometry of motion
Humans Need Augmented Feedback to Physically Track Non-Biological Robot Movements
Mahdiar Edraki, D. Sternad
Robotic Intelligence
🎯 What it does: This study evaluates whether humans can reduce interaction force with a robot moving along an elliptical trajectory through physical tracking by subjects, under biological and non-biological speed profiles, after multiple days of practice with or without real-time visual force error feedback.
Hummingbird-bat hybrid wing by 3-D printing*
Tomoya Fujii, Hiroto Tanaka
Robotic IntelligenceMesh
🎯 What it does: Proposed a humanoid bird-bat hybrid wing (HBH) design, achieving torsional flexibility by adding a twisting arm at the leading edge and using a stretchable membrane, and manufacturing the three-dimensional shape via 3D printing. Subsequently, the lift and deformation performance were evaluated using an electric flapping mechanism.
Hybrid SUSD-Based Task Allocation for Heterogeneous Multi-Robot Teams
Shengkang Chen, Fumin Zhang
OptimizationRobotic Intelligence
🎯 What it does: Proposed a hybrid task allocation algorithm that improves task allocation through the SUSD (Speeding-Up and Slowing-Down) method based on the initial solution;
Identification of a Generalized Base Inertial Parameter Set of Robotic Manipulators Considering Mounting Configurations
Mario Tröbinger, Sami Haddadin
Robotic Intelligence
🎯 What it does: Proposes a generic base inertial parameter identification framework applicable to any robot installation configuration
Identifying Contact Distance Uncertainty in Whisker Sensing with Tapered, Flexible Whiskers
T. A. Kent, S. Bergbreiter
Robotic Intelligence
🎯 What it does: A new tactile sensing system with conical flexible whiskers is proposed, and the radial contact distance of the whiskers is estimated using Gradient Moment (GM) algorithm and Moment-Force (MF) algorithm. The GM algorithm achieves an error below 4% of the whisker length in most cases, and a Z-Disimilarity metric is introduced to quantify the uncertainty of the contact distance. Finally, the two algorithms are combined to obtain a more robust estimation of the contact distance.
Image Masking for Robust Self-Supervised Monocular Depth Estimation
Hemang Chawla, Bahram Zonooz
Depth EstimationImage
🎯 What it does: Propose the MIMDepth method, applying masked image modeling (MIM) to self-supervised monocular depth estimation.
Image Segmentation for Continuum Robots from a Kinematic Prior
Connor Watson, Tania. K. Morimoto
SegmentationOptimizationRobotic IntelligenceImage
🎯 What it does: Proposed a continuous robot image segmentation method based on kinematic priors, achieving robust segmentation without requiring training data or labels
Image-based Pose Estimation and Shape Reconstruction for Robot Manipulators and Soft, Continuum Robots via Differentiable Rendering
Jingpei Lu, Michael C. Yip
Pose EstimationRobotic IntelligenceNeural Radiance FieldImage
🎯 What it does: Achieved robot pose estimation and shape reconstruction based on camera images, utilizing a differentiable renderer and geometric shape primitives without requiring precise meshes.
Image-Based Visual Servoing Switchable Leader-follower Control of Heterogeneous Multi-agent Underwater Robot System
Kanzhong Yao, S. Watson
Robotic IntelligenceImage
🎯 What it does: Proposed an image-based visual servoing (IBVS) leader-follower control system applicable to heterogeneous underwater robot teams.
Image-to-Image Translation for Autonomous Driving from Coarsely-Aligned Image Pairs
Youya Xia, Mark E. Campbell
Image TranslationAutonomous DrivingImage
🎯 What it does: Propose a training framework that utilizes roughly registered image pairs for image translation, converting images captured under adverse weather conditions into clear weather images to enhance downstream task performance.
Immersive Demonstrations are the Key to Imitation Learning
Kelin Li, Nicolás Rojas
Robotic IntelligenceReinforcement Learning from Human Feedback
🎯 What it does: Explored the impact of pick-and-place task demonstrations using a feedback glove and robotic arm on imitation learning under three conditions: no force feedback, fingertip force feedback, and fingertip + palm force feedback
ImmFusion: Robust mmWave-RGB Fusion for 3D Human Body Reconstruction in All Weather Conditions
Anjun Chen, Qi Ye
GenerationTransformerImageMultimodalityPoint Cloud
🎯 What it does: Fusing millimeter-wave radar and RGB images for 3D human reconstruction under all-weather conditions
iMODE:Real-Time Incremental Monocular Dense Mapping Using Neural Field
H. Matsuki, A. Davison
Depth EstimationNeural Radiance FieldSimultaneous Localization and MappingImage
🎯 What it does: Propose a real-time dense and semantic neural field mapping system that uses only monocular image input.
Implementation and Optimization of Grasping Learning with Dual-modal Soft Gripper
Lei Zhao, Bin Fang
Robotic IntelligenceReinforcement LearningImage
🎯 What it does: Design a dual-mode soft gripper and propose an operational framework based on deep reinforcement learning, supporting enveloping grasp and pinching grasp modes.
Implicit Neural Field Guidance for Teleoperated Robot-assisted Surgery
Heng Zhang, Ai-Guon Song
OptimizationRobotic IntelligenceNeural Radiance FieldBiomedical Data
🎯 What it does: Propose a new framework that utilizes implicit neural fields to guide teleoperated surgical robots, avoiding collisions between the robot and human tissue caused by input inaccuracies.
Improved Benthic Classification using Resolution Scaling and SymmNet Unsupervised Domain Adaptation
Heather J. Doig, S. Williams
ClassificationDomain AdaptationImage
🎯 What it does: Proposes a framework to enhance the performance of benthic morphology classifiers on survey images with different training data.
Improved Event-Based Dense Depth Estimation via Optical Flow Compensation
Dian-xi Shi, Yi Zhang
Depth EstimationConvolutional Neural NetworkOptical Flow
🎯 What it does: Propose a dense depth estimation framework based on event cameras called Mixed-EF2DNet.
Improving robot navigation in crowded environments using intrinsic rewards
Diego Martinez-Baselga, L. Montano
Robotic IntelligenceReinforcement Learning
🎯 What it does: Propose using intrinsic rewards to balance a robot's exploration and exploitation in crowded environments, and explore based on state uncertainty.
Improving the Generalizability of Trajectory Prediction Models with Frenét-Based Domain Normalization
Luyao Ye, Jianping Wang
Domain AdaptationAutonomous DrivingBenchmark
🎯 What it does: Propose a domain normalization strategy based on Frenet coordinates to enhance the generalization ability of trajectory prediction models across different domains.
Improving the Performance of Local Bundle Adjustment for Visual-Inertial SLAM with Efficient Use of GPU Resources
Shishir Gopinath, Steven Y. Ko
OptimizationComputational EfficiencySimultaneous Localization and MappingVideo
🎯 What it does: Developed a GPU-based block solver to accelerate local bundle adjustment in visual-inertial SLAM, integrating it as a replaceable block solver into ORB-SLAM3.
Improving Video Super-Resolution with Long-Term Self-Exemplars
Guotao Meng, Qifeng Chen
Super ResolutionVideo
🎯 What it does: Propose a post-processing method that integrates similar patches (self-examples) from remote frames into video super-resolution by leveraging long-term cross-scale aggregation and multi-reference alignment modules
In-Hand Manipulation in Power Grasp: Design of an Adaptive Robot Hand with Active Surfaces
Yilin Cai, Shenli Yuan
Robotic Intelligence
🎯 What it does: Developed and tested the BACH compliant manipulator, achieving grasping and internal manipulation under power grip.
In-Mouth Robotic Bite Transfer with Visual and Haptic Sensing
Lorenzo Shaikewitz, Dorsa Sadigh
Robotic IntelligenceMultimodality
🎯 What it does: Demonstrated a semi-autonomous robotic platform capable of safely, comfortably, and effectively transferring bite-sized food directly from utensils into the human mouth.
In-situ Mechanical Calibration for Vision-based Tactile Sensors
Can Zhao, Daolin Ma
Physics Related
🎯 What it does: Proposes a method for in-situ calibration of mechanical parameters (Young's modulus and Poisson's ratio) of visual tactile sensors
Increasing Admittance of Industrial Robots By Velocity Feedback Inner-Loop Shaping
Kangwagye Samuel, Sehoon Oh
Robotic Intelligence
🎯 What it does: Utilizing a velocity feedback internal loop shape shaping technique to enable industrial robots with low intrinsic impedance to achieve higher adhesion through control.
Incremental Few-Shot Object Detection via Simple Fine-Tuning Approach
Taehyean Choi, Jong-Hwan Kim
Object DetectionConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: Proposes a simple fine-tuning method called Incremental Two-stage Fine-tuning Approach (iTFA) for incremental few-shot object detection, which separates the RoI feature extractor and classifier into base class and new class branches after base class training, and only uses a small number of new class samples to fine-tune the new class branch.
Induced Vertex Motion As a Performance Measure for Surgery in Confined Spaces
Neel Shihora, N. Simaan
OptimizationRobotic Intelligence
🎯 What it does: Proposes a performance metric to evaluate non-target movements generated along the length of surgical robots in confined spaces, and compares design schemes using this metric
Informable Multi-Objective and Multi-Directional RRT* System for Robot Path Planning
Jiunn-Kai Huang, J. Grizzle
OptimizationRobotic Intelligence
🎯 What it does: Proposed a real-time iterative system for simultaneously solving multi-objective path planning problems and determining the destination visit order.
Information-theoretic Abstraction of Semantic Octree Models for Integrated Perception and Planning
Daniel T. Larsson, P. Tsiotras
CompressionAutonomous DrivingPoint Cloud
🎯 What it does: Construct and compress a semantic environment representation based on point cloud data, generating a multi-resolution semantic octree model.
Infrared Image Captioning with Wearable Device
Chenjun Gao, Huaping Liu
GenerationVision Language ModelImage
🎯 What it does: Proposed an image caption generation framework based on infrared images and integrated it into wearable devices
Infrastructure-based End-to-End Learning and Prevention of Driver Failure
Noam Buckman, Daniela Rus
Anomaly DetectionAutonomous DrivingRecurrent Neural NetworkTime SeriesSequential
🎯 What it does: Developed FailureNet, an end-to-end recurrent neural network (RNN) for detecting hazardous drivers or autonomous system failures when approaching intersections in a small-scale urban model, and issuing warnings to relative traffic at intersections.
Input-Output Boundedness of a Magnetically-Actuated Helical Device
Leendert-Jan W. Ligtenberg, Islam S. M. Khalil
Robotic IntelligencePhysics Related
🎯 What it does: Study the input-output boundedness of a helical device driven by a single rotating magnet at low Reynolds numbers. Numerical and experimental analyses are conducted on drive frequency, inclination angle, lateral velocity, and magnetic field strength. The boundedness of the device's state is proven in open-loop circular and linear motion, and bounded propulsion against gravity is demonstrated without adjusting the angle of attack.
Instance-wise Grasp Synthesis for Robotic Grasping
Yucheng Xu, Zhibin Li
GenerationRobotic IntelligenceImage
🎯 What it does: Proposed the Single-Stage Grasp Synthesis Network (SSG), which can generate instance masks and grasp configurations for each object in one go;
Integrated vector field and backstepping control for quadcopters
Arthur H. D. Nunes, L. Pimenta
Robotic Intelligence
🎯 What it does: A scheme integrating vector fields and backstepping integral control is designed for quadcopter path tracking and obstacle avoidance, capable of counteracting constant uncertainties.
Intention Aware Robot Crowd Navigation with Attention-Based Interaction Graph
Shuijing Liu, K. Driggs-Campbell
Robotic IntelligenceRecurrent Neural NetworkGraph Neural NetworkReinforcement Learning
🎯 What it does: This paper proposes a multi-robot navigation framework based on recursive graph neural networks and attention mechanisms, which can capture heterogeneous interactions in space and time, infer intentions by predicting the future trajectories of dynamic agents, and guide robots to achieve safe and intention-aware navigation in densely interactive crowds;
Interacting with Multi-Robot Systems via Mixed Reality
Florian Kennel-Maushart, Stelian Coros
Robotic Intelligence
🎯 What it does: Implemented a mixed reality interface capable of interacting with multiple mobile robots simultaneously, and conducted user studies on mobile robots with the Microsoft HoloLens 2 headset or tablet devices.
Interaction-Aware Trajectory Planning for Autonomous Vehicles with Analytic Integration of Neural Networks into Model Predictive Control
P. Gupta, S. Bae
Autonomous DrivingOptimization
🎯 What it does: Designed an interactive perception motion planner that enables autonomous vehicles (AVs) to interact with surrounding vehicles and perform complex maneuvers in a locally optimal manner, using a neural network to predict interactive trajectories and analytically integrate with model predictive control (MPC).
Interactive Object Segmentation in 3D Point Clouds
Theodora Kontogianni, K. Schindler
SegmentationPoint Cloud
🎯 What it does: Proposes an interactive 3D point cloud instance segmentation method, where users can directly click on targets or the background in the point cloud to iteratively refine the segmentation results.
Intermittent diffusion-based path planning for heterogeneous groups of mobile sensors in cluttered environments
Christina Frederick, Frank Crosby
OptimizationRobotic IntelligenceDiffusion model
🎯 What it does: A task-oriented path planning and collision avoidance method is proposed for deploying heterogeneous omnidirectional mobile sensors in cluttered environments.
Interpretable and Flexible Target-Conditioned Neural Planners For Autonomous Vehicles
Haolan Liu, Liangjun Zhang
Autonomous DrivingExplainability and InterpretabilityPoint Cloud
🎯 What it does: Proposed an interpretable neural planner that represents multi-object planning schemes through regression heatmaps;
Intuitive Robot Integration via Virtual Reality Workspaces
Minh Q. Tram, William J. Beksi
Robotic IntelligenceWorld Model
🎯 What it does: Developed a purely virtual robot system simulation framework that utilizes an immersive virtual reality workspace to achieve natural human-robot interaction, supporting robot programming and task training.
Intuitive Telemanipulation of Hyper-Redundant Snake Robots within Locomotion and Reorientation using Task-Priority Inverse Kinematics
Tim-Lukas Habich, Svenja Spindeldreier
OptimizationRobotic Intelligence
🎯 What it does: A unified teleoperation strategy is proposed, utilizing a shape fitting method to achieve the following guidance motion and posture adjustment of the snake robot while minimizing shape changes as much as possible.
Inverse Perspective Mapping-Based Neural Occupancy Grid Map for Visual Parking
Xiangru Mu, Tong Qin
Autonomous DrivingSupervised Fine-TuningSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: Propose a visual occupancy grid mapping method based on inverse perspective mapping (IPM), which uses LiDAR pseudo labels to supervise the visual network for detecting obstacle occupancy boundaries, and fuses visual occupancy information with vehicle motion data to construct a multi-frame local occupancy grid map.
Inverse Reinforcement Learning Framework for Transferring Task Sequencing Policies from Humans to Robots in Manufacturing Applications
O. Manyar, Sandeep K. S. Gupta
Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement Learning
🎯 What it does: Proposed an inverse reinforcement learning framework to address task scheduling problems for robots in complex manufacturing processes, capable of adapting to process changes and sorting requirements for new components.
Inverted Landing in a Small Aerial Robot via Deep Reinforcement Learning for Triggering and Control of Rotational Maneuvers
Bryan Habas, Bo Cheng
Domain AdaptationRobotic IntelligenceReinforcement Learning
🎯 What it does: Learns an optimal control strategy for achieving inverted landing under any approach conditions using deep reinforcement learning and physical simulation;
It Takes Two: Learning to Plan for Human-Robot Cooperative Carrying
Eley Ng, Monroe Kennedy
Robotic IntelligenceRecurrent Neural NetworkSequential
🎯 What it does: Proposed a real-time planning method based on VRNN for predicting feasible trajectories in human-robot collaborative lifting tasks.
Joint Camera Intrinsic and LiDAR-Camera Extrinsic Calibration
Guohang Yan, Yikang Li
Autonomous DrivingOptimizationImagePoint Cloud
🎯 What it does: Proposed a target-based joint calibration method for camera intrinsics and LiDAR-camera extrinsic parameters, designed a novel calibration board with four circular holes surrounding a chessboard, and constructed a cost function under reprojection constraints to solve camera intrinsics, distortion coefficients, and LiDAR-camera extrinsic parameters.
Joint Segmentation and Grasp Pose Detection with Multi-Modal Feature Fusion Network
Xiaozhen Liu, Jiaqi Zhao
SegmentationPose EstimationRobotic IntelligenceImageMultimodalityPoint Cloud
🎯 What it does: Propose a multimodal feature fusion network for simultaneously performing object segmentation and grasp pose detection.
Joint Semi-Supervised and Active Learning via 3D Consistency for 3D Object Detection
Sihwan Hwang, Dongsuk Kum
Object DetectionAutonomous DrivingPoint Cloud
🎯 What it does: Proposes a 3D object detection method combining semi-supervised learning and active learning, utilizing 3D consistency constraints and self-supervised learning to reduce bounding box uncertainty, and performing active annotation for occluded or distant objects with remaining high uncertainty.
Just Round: Quantized Observation Spaces Enable Memory Efficient Learning of Dynamic Locomotion
Lev Grossman, B. Plancher
Computational EfficiencyRobotic IntelligenceReinforcement Learning
🎯 What it does: Reducing memory consumption in DRL training through observation space quantization
Keypoint-GraspNet: Keypoint-based 6-DoF Grasp Generation from the Monocular RGB-D input
Yiye Chen, P. Vela
Data SynthesisPose EstimationConvolutional Neural NetworkImage
🎯 What it does: This paper proposes Keypoint-GraspNet, which directly detects gripper keypoints using monocular RGB-D images and recovers their 6-DoF SE(3) pose via the PnP algorithm to generate grasp poses.
KGNet: Knowledge-Guided Networks for Category-Level 6D Object Pose and Size Estimation
Qiwei Meng, Wei Song
Pose EstimationRobotic IntelligenceImagePoint CloudBenchmark
🎯 What it does: Propose the KGNet network for estimating the 6D pose and size of category-level unseen objects, supporting robotic grasping.
Kinematic Analysis and Design of a Novel (6+3)-DoF Parallel Robot with Fixed Actuators
Arda Yiğit, C. Gosselin
Robotic Intelligence
🎯 What it does: A kinematically redundant parallel robot with 6+3 degrees of freedom is proposed, featuring an RU/2-RUS leg structure, base-fixed motors, and direct drive for intuitive human-robot interaction. It achieves a maximized workspace with 2g acceleration in all directions, identifies all singularities, and provides a CAD model.
Kinodynamic Rapidly-exploring Random Forest for Rearrangement-Based Nonprehensile Manipulation
Kejia Ren, Kaiyu Hang
Robotic Intelligence
🎯 What it does: This paper proposes a forest-based motion dynamics planning framework for solving rearrangement-based non-grasping manipulation problems.
Knowledge Distillation for Feature Extraction in Underwater VSLAM
Jinghe Yang, Yen-Yu Pu
Data SynthesisKnowledge DistillationSimultaneous Localization and MappingImage
🎯 What it does: Propose a cross-modal knowledge distillation framework to train an underwater feature detection and matching network UFEN, and integrate it into ORB-SLAM3 to replace ORB features.
kollagen: A Collaborative SLAM Pose Graph Generator
Roberto C. Sundin, David Umsonst
GenerationData SynthesisSimultaneous Localization and MappingGraph
🎯 What it does: Proposed a collaborative SLAM pose graph generator named Kollagen, which can generate reproducible pose graph datasets based on user-defined parameters.
KRIS: A Novel Device for Kinesthetic Corrective Feedback during Robot Motion
Jorn Verheggen, Kim Baraka
Robotic Intelligence
🎯 What it does: Introduces a device called KRIS that can be mounted on a robot end-effector, capable of providing intuitive kinesthetic corrective feedback during human manipulation.
KubeROS: A Unified Platform for Automated and Scalable Deployment of ROS2-based Multi-Robot Applications
Yong-Zeng Zhang, B. Hein
Robotic Intelligence
🎯 What it does: Propose the KubeROS platform to achieve unified automated deployment of ROS2 robot applications across robots, edge devices, and the cloud;
L-C*: Visual-inertial Loose Coupling for Resilient and Lightweight Direct Visual Localization
Shuji Oishi, A. Banno
Pose EstimationSimultaneous Localization and MappingImageMultimodality
🎯 What it does: Proposed the L-C* framework, which achieves lightweight direct visual localization using a visual-inertial loosely coupled approach;
L2E: Lasers to Events for 6-DoF Extrinsic Calibration of Lidars and Event Cameras
Kevin Ta, L. Gool
Pose EstimationOptimizationPoint Cloud
🎯 What it does: Proposes a direct, time-decoupled extrinsic calibration method between event cameras and LiDAR.
LAPTNet-FPN: Multi-Scale LiDAR-Aided Projective Transform Network for Real Time Semantic Grid Prediction
M. Diaz-Zapata, C. Laugier
SegmentationAutonomous DrivingConvolutional Neural NetworkPoint Cloud
🎯 What it does: Proposed a multi-scale LiDAR-assisted projection transformation network for real-time semantic grid generation.
Large-Scale Radar Localization using Online Public Maps
Ziyang Hong, Sen Wang
Simultaneous Localization and Mapping
🎯 What it does: Propose using online public maps (e.g., OpenStreetMap) for large-scale radar localization without requiring prior mapping.
LATITUDE: Robotic Global Localization with Truncated Dynamic Low-pass Filter in City-scale NeRF
Z. Zhu, Guyue Zhou
Robotic IntelligenceNeural Radiance FieldSimultaneous Localization and MappingImage
🎯 What it does: Proposes the LATITUDE two-stage city-scale NeRF global localization framework, which includes a regressor trained on NeRF images to provide initial poses, and pose optimization achieved by minimizing the residual between observed and rendered images on the tangent plane combined with a truncated dynamic low-pass filter (TDLF).
LATTE: LAnguage Trajectory TransformEr
A. Bucker, Rogerio Bonatti
Robotic IntelligenceTransformerLarge Language ModelVision Language ModelImageText
🎯 What it does: Proposes a language-based framework that leverages pre-trained language models and visual models to encode user intent and target objects, generating trajectories applicable to different robot platforms.
Learnable Tegotae-based Feedback in CPGs with Sparse Observation Produces Efficient and Adaptive Locomotion
Christopher Herneth, D. Owaki
Robotic Intelligence
🎯 What it does: Proposed a task-distributed, end-to-end trainable closed-loop CPG control strategy that extends and generalizes the Tegotae control method.
Learned Risk Metric Maps for Kinodynamic Systems
R. Allen (Massachusetts Institute of Technology), D. Rus (Massachusetts Institute of Technology)
Computational EfficiencyRobotic Intelligence
🎯 What it does: Proposed and implemented the LRMM model for real-time estimation of coherent risk measures in high-dimensional dynamical systems within partially observable environments.
LEARNEST: LEARNing Enhanced Model-based State ESTimation for Robots using Knowledge-based Neural Ordinary Differential Equations
K. Y. Chee, M. A. Hsieh
Robotic IntelligenceOrdinary Differential Equation
🎯 What it does: Proposed the LEARNEST framework, which leverages knowledge-based neural ordinary differential equations (KNODEs) to enhance the state estimation accuracy of robotic systems, and integrated them into two model-based state estimation algorithms: KNODE-MHE and KNODE-UKF; experimental validations were conducted on various robotic applications, including the Cart-pole, ground robot localization, and quadrotor state estimation.
Learning a Single Near-hover Position Controller for Vastly Different Quadcopters
Dingqi Zhang, Mark W. Mueller
Representation LearningRobotic Intelligence
🎯 What it does: Propose an adaptive near-hover position controller that uses neural networks to estimate potential representations of the robot and environment, and online adapts robot dynamics and unknown disturbances within the same framework; the controller, trained in simulation, is directly deployed on two quadrotors with significant differences in mass, size, motors, and propellers, and can rapidly adapt to sudden disturbances equivalent to one-third of the drone's mass.
Learning Agent-Aware Affordances for Closed-Loop Interaction with Articulated Objects
Giulio Schiavi, Jen Jen Chung
Robotic Intelligence
🎯 What it does: Propose a closed-loop control pipeline that integrates an affordance-based operational prior with sample-based whole-body control to enable interaction between mobile robots and articulated objects.
Learning Agile Flight Maneuvers: Deep SE(3) Motion Planning and Control for Quadrotors
Yixiao Wang, Lin Zhao
OptimizationRobotic IntelligenceReinforcement Learning
🎯 What it does: Developed a scheme based on deep reinforcement learning, utilizing parameterized deep neural networks to generate adaptive tracking references for model predictive controllers, including passage time and SE(3) passage attitude for quadrotors, and proposed a binary search algorithm to achieve real-time online adaptation for dynamic doorways.
Learning an Efficient Terrain Representation for Haptic Localization of a Legged Robot
D. Sójka, Piotr Skrzypczy'nski
Computational EfficiencyRobotic IntelligenceTransformerContrastive LearningSimultaneous Localization and Mapping
🎯 What it does: Propose an efficient terrain representation method for localizing quadruped robots through tactile information in extreme environments, combining transformer networks and triplet loss to learn embeddings, and integrating with Monte Carlo algorithms to achieve accurate and computationally efficient localization.
Learning and Blending Robot Hugging Behaviors in Time and Space
M. Drolet, H. B. Amor
Robotic Intelligence
🎯 What it does: Proposed a physical human-robot interaction algorithm based on imitation learning to predict and realize the robot's response behavior in complex hugging scenarios.
Learning Arm-Assisted Fall Damage Reduction and Recovery for Legged Mobile Manipulators
Yuntao Ma, Marco Hutter
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a learning-based wrist-assisted method for fall damage reduction and recovery
Learning Augmented, Multi-Robot Long-Horizon Navigation in Partially Mapped Environments
Abhish Khanal, Gregory J. Stein
Robotic IntelligenceWorld Model
🎯 What it does: Proposes a method for goal-oriented long-range navigation by multi-robot teams in structured unknown environments, leveraging statistical predictions of unknown spaces to achieve efficient and reliable navigation.
Learning Category-Level Manipulation Tasks from Point Clouds with Dynamic Graph CNNs
Junchi Liang, Abdeslam Boularias
Pose EstimationRobotic IntelligenceGraph Neural NetworkImagePoint Cloud
🎯 What it does: Uses a dynamic graph convolutional neural network (DG-CNN) to learn category-level grasping and manipulation from raw RGB-D videos without manual annotations, enabling task execution on new objects.