ICRA 2023 Papers — Page 3
IEEE International Conference on Robotics and Automation · 1341 papers
Bounded Compensation with Friction Estimation for Accurate Motion Tracking and Compliant Behavior of Industrial Manipulators
Dongwoo Ko, Keehoon Kim
Robotic Intelligence
🎯 What it does: Propose a control structure that does not require additional sensors for precise tracking and compliance behavior in industrial robotic arms, achieving compensation through friction estimation.
Bridging the Domain Gap for Multi-Agent Perception
Runsheng Xu, Jiaqi Ma
Object DetectionDomain AdaptationAutonomous DrivingTransformerPoint Cloud
🎯 What it does: Propose a lightweight framework to bridge domain gaps in multi-agent perception, compatible with existing systems while maintaining confidentiality.
Buoyancy enabled autonomous underwater construction with cement blocks
Samuel E. Lensgraf, Alberto Quattrini Li
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: Developed the first free-floating autonomous underwater construction system, utilizing compressed air for active balance and battery-driven thrusters, capable of effectively transporting cement blocks and constructing structures with up to 12 components, totaling 100 kg (75 kg underwater) in experiments.
Burst Stimulation for Enhanced Locomotion Control of Terrestrial Cyborg Insects
Huu Duoc Nguyen, T. Vo-Doan
Robotic IntelligenceBiomedical Data
🎯 What it does: Using Burst Stimulation (burst stimulation) to electrically stimulate native hybrid insects—beetles—to enhance their maneuverability control performance.
CabiNet: Scaling Neural Collision Detection for Object Rearrangement with Procedural Scene Generation
Adithyavairavan Murali, D. Fox
Data SynthesisRobotic IntelligencePoint Cloud
🎯 What it does: Proposed the CabiNet model, trained using synthetic partial point clouds from over 650K procedurally generated cluttered scenes, which can quickly predict collision scenarios of objects' SE(3) poses under single-view depth observations, and achieve collision-free grasping and placement by combining the MPPI planner and SDF-generated path points.
Cable Routing and Assembly using Tactile-driven Motion Primitives
Achu Wilson, Wenzhen Yuan
Robotic IntelligenceVision-Language-Action ModelMultimodality
🎯 What it does: Integrating tactile-driven low-level motion control with high-level visual task parsing, achieving cable routing and assembly on a reconfigurable task board.
CAHIR: Co-Attentive Hierarchical Image Representations for Visual Place Recognition
Guohao Peng, Danwei W. Wang
RecognitionKnowledge DistillationRepresentation LearningTransformerImageBenchmark
🎯 What it does: Proposed the CAHIR framework, unifying attention-shared global and local descriptors into a single encoding pipeline for visual place recognition.
CalibDepth: Unifying Depth Map Representation for Iterative LiDAR-Camera Online Calibration
Jiangtong Zhu, Pu Zhang
Depth EstimationAutonomous DrivingImagePoint Cloud
🎯 What it does: Propose CalibDepth, which uses depth maps as a unified representation for images and LiDAR point clouds, and introduces a monocular depth estimation subnetwork to assist online calibration; treat online calibration as a sequence prediction problem, and optimize results using global and local losses.
Calibration and Uncertainty Characterization for Ultra-Wideband Two-Way-Ranging Measurements
Mohamed Fouad Shalaby, J. L. Ny
Autonomous DrivingSimultaneous Localization and MappingTabular
🎯 What it does: Proposed a UWB bidirectional ranging protocol and developed a scalable antenna delay calibration method, while modeling error and uncertainty as a function of received signal power, and applying it to positioning using an extended Kalman filter.
Can Machines Garden? Systematically Comparing the AlphaGarden vs. Professional Horticulturalists
S. Adebola (University of California Berkeley), Ken Goldberg (University of California Berkeley)
Agriculture Related
🎯 What it does: Compared the performance of the AlphaGarden automated multi-species planting system with that of UC Berkeley professional horticulturists under the same seed layout; both were tested in the same laboratory environment for a 60-day cycle experiment, measuring canopy coverage, plant diversity, and water consumption;
Can We Use Diffusion Probabilistic Models for 3D Motion Prediction?
Hyemin Ahn, Dongheui Lee
Pose EstimationDiffusion modelTime SeriesSequential
🎯 What it does: The study utilizes diffusion probability models to predict future 3D human motion based on prior observations.
CAROM Air - Vehicle Localization and Traffic Scene Reconstruction from Aerial Videos
Duo Lu, Yezhou Yang
Object TrackingVideoBenchmark
🎯 What it does: Automatically extract vehicle trajectory data from aerial videos captured by consumer-grade drones to reconstruct traffic scenes and enable precise reproduction.
Carrying the uncarriable: a deformation-agnostic and human-cooperative framework for unwieldy objects using multiple robots
Doganay Sirintuna, Arash Ajoudani
Robotic Intelligence
🎯 What it does: Proposes a multi-robot collaborative carrying framework that is agnostic to object deformability, allowing humans to fully control the carrying trajectory while robots share the load based on object size and weight.
Category-Level Global Camera Pose Estimation with Multi-Hypothesis Point Cloud Correspondences
Jun-Jee Chao, Volkan Isler
Pose EstimationOptimizationPoint Cloud
🎯 What it does: Propose an optimization method that retains all possible correspondences and achieves category-level global camera pose estimation through iterative updates; meanwhile, design a novel point feature descriptor to measure the similarity of local point cloud regions.
Category-level Shape Estimation for Densely Cluttered Objects
Zhenyu Wu, Haibin Yan
SegmentationImagePoint Cloud
🎯 What it does: Proposes a category-level shape estimation method for densely stacked objects, utilizing multi-view RGB-D images for point cloud reconstruction and achieving high-precision instance segmentation and shape recovery through feature fusion.
Causal Inference for De-biasing Motion Estimation from Robotic Observational Data
Junhong Xu, Lantao Liu
Robotic IntelligenceReinforcement LearningTime Series
🎯 What it does: Proposed a method utilizing causal inference frameworks (IPW, DR) for unbiased estimation of parameters in a robot's stochastic motion model, and developed an approximate policy iteration algorithm based on the debiased state transition function.
Cautious Planning with Incremental Symbolic Perception: Designing Verified Reactive Driving Maneuvers
Disha Kamale, C. Vasile
Autonomous Driving
🎯 What it does: Propose utilizing incrementally refined symbolic knowledge for provable reactive control synthesis in autonomous driving, combining abstract models of motion control and information acquisition, and defining traffic rules using assume-guarantee specifications.
CDFI: Cross Domain Feature Interaction for Robust Bronchi Lumen Detection
Jiasheng Xu, Yun Gu
SegmentationConvolutional Neural NetworkBiomedical Data
🎯 What it does: Proposed a Cross-Domain Feature Interaction (CDFI) network to extract bronchial lumen structural features and provide visual noise feature hints; the Quadruple Feature Constraints (QFC) module and Guided Feature Fusion (GFF) module are used to extract structural and artifact features and adaptively fuse them for samples of different image qualities.
CEAFFOD: Cross-Ensemble Attention-based Feature Fusion Architecture Towards a Robust and Real-time UAV-based Object Detection in Complex Scenarios
Ahmed Elhagry, G. Masi
Object DetectionConvolutional Neural NetworkImage
🎯 What it does: Proposed a novel single-stage detection architecture designed for real-time, lightweight object detection in UAV environments.
Center Feature Fusion: Selective Multi-Sensor Fusion of Center-based Objects
P. Jacobson, Ming Wu
Object DetectionAutonomous DrivingConvolutional Neural NetworkImagePoint Cloud
🎯 What it does: Proposes a Center Feature Fusion (CFF) method that utilizes center detection networks from both camera and LiDAR streams to locate object positions, and projects and fuses only the pixel features related to the target into the Bird's Eye View (BEV) space;
CenterLineDet: CenterLine Graph Detection for Road Lanes with Vehicle-mounted Sensors by Transformer for HD Map Generation
Zhenhua Xu, Lujia Wang
Autonomous DrivingTransformerImageMultimodalityPoint Cloud
🎯 What it does: Proposed a method called CenterLineDet for automatically detecting road lane centerlines using vehicle-mounted sensors (six cameras and a LiDAR) to achieve high-definition map generation.
Cerberus: Low-Drift Visual-Inertial-Leg Odometry For Agile Locomotion
Shuozhi Yang, Zachary Manchester
Robotic IntelligenceSimultaneous Localization and MappingImage
🎯 What it does: Proposed an open-source visual-inertial-leg odometry (VILO) state estimator called Cerberus, which can accurately estimate position in real-time across various terrains using stereo cameras, IMU, joint encoders, and contact sensors.
CFVS: Coarse-to-Fine Visual Servoing for 6-DoF Object-Agnostic Peg-In-Hole Assembly
B. Lu, Winston H. Hsu
Pose EstimationRobotic Intelligence
🎯 What it does: Propose a coarse-to-fine visual servoing (CFVS) method based on three-dimensional visual feedback for achieving 6 degrees of freedom plug-in hole assembly.
Chance-Constrained Motion Planning with Event-Triggered Estimation
Anne Theurkauf, Morteza Lahijanian
Autonomous DrivingOptimizationComputational Efficiency
🎯 What it does: Proposes a fast and efficient sample-based motion planner integrating event-triggered estimation for motion and communication planning under uncertainty and limited information from remote sensor networks.
Characterisation of Antagonistically Actuated, Stiffness-Controllable Joint-Link Units for Cobots
Wenlong Gaozhang, H. Wurdemann
Robotic Intelligence
🎯 What it does: Studied the impact of adversarial-driven, adjustable stiffness joint-link units (JLU) on the performance of collaborative robots, and compared them with traditional rigid units.
Chronos and CRS: Design of a miniature car-like robot and a software framework for single and multi-agent robotics and control
Andrea Carron, M. Zeilinger
Robotic Intelligence
🎯 What it does: Proposed Chronos (a 1/28 scale vehicle robot model) and CRS (an open-source software framework for control and robotics), and implemented various advanced control, estimation, and multi-agent coordination algorithms.
CIOT: Constraint-Enhanced Inertial-Odometric Tracking for Articulated Dump Trucks in GNSS-Denied Mining Environments
David Benz, H. Vallery
Autonomous DrivingSimultaneous Localization and MappingTime Series
🎯 What it does: Proposed and evaluated a novel constraint-enhanced inertial-odometry tracking filter that integrates IMU, GNSS, and wheel speed sensors for positioning articulated loaders in GNSS-restricted deep pit mining environments.
CLIO: a Novel Robotic Solution for Exploration and Rescue Missions in Hostile Mountain Environments
Michele Focchi, L. Palopoli
OptimizationRobotic Intelligence
🎯 What it does: Propose a rope-assisted climbing robot capable of walking on vertical slopes and carrying heavy loads
Cloth Funnels: Canonicalized-Alignment for Multi-Purpose Garment Manipulation
Alper Canberk, Shuran Song
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed and implemented the 'canonicalized-alignment' task, which converts arbitrarily configured clothing into a predefined deformable configuration and aligns it to a suitable rigid pose, providing a compact and highly visualizable state for subsequent clothing manipulation.
Clothes Grasping and Unfolding Based on RGB-D Semantic Segmentation
Xin-Hai Zhu, Yixing Gao
Object DetectionSegmentationRobotic IntelligenceConvolutional Neural NetworkGenerative Adversarial NetworkImageMultimodality
🎯 What it does: Proposed a bidirectional fractal cross-fusion network (BiFCNet) based on RGB-D semantic segmentation for identifying graspable regions on clothing, combined with custom data augmentation and grasp point selection strategies to achieve clothing grasping and unfolding;
CMG-Net: An End-to-End Contact-based Multi-Finger Dexterous Grasping Network
M. Wei, Jian Tang
Pose EstimationRobotic IntelligencePoint Cloud
🎯 What it does: Propose a grasp representation based on contact between multi-fingered robot hands and objects, and construct an end-to-end network CMG-Net, which can efficiently predict multi-finger grasp poses and hand configurations for unknown objects in cluttered environments from single-frame point clouds.
CNN-based Visual Servoing for Simultaneous Positioning and Flattening of Soft Fabric Parts
Fuyuki Tokuda, K. Kosuge
Robotic IntelligenceConvolutional Neural NetworkImageMultimodality
🎯 What it does: Propose a CNN-based visual servoing method to simultaneously locate and flatten soft fabrics placed on a table using a dual-robot arm system.
Coarse-to-Fine Point Cloud Registration with SE(3)-Equivariant Representations
Chengxuan Lin, Winston H. Hsu
Pose EstimationPoint Cloud
🎯 What it does: A coarse-to-fine stage point cloud registration method was designed, first aligning global features to reduce pose differences, then refining by matching local features; simultaneously, an SE(3)-equivariant feature extractor was proposed to generate pose-preserving and pose-invariant features.
Coaxial Modular Aerial System and the Reconfiguration Applications
José Baca, Pablo Rangel
🎯 What it does: This paper proposes a modular aerial system based on coaxial motors (CMAS), which achieves flight control by realizing two-degree-of-freedom mass center transfer within the module, and enables rapid assembly and disassembly with other modules and metal surfaces through magnetic connectors;
Code as Policies: Language Model Programs for Embodied Control
Jacky Liang, Andy Zeng
Robotic IntelligenceLarge Language ModelPrompt EngineeringTextBenchmark
🎯 What it does: Leveraging large language models trained on code completion to convert natural language commands into robot policy code, supporting reactive control and vision-based path planning.
CogniDaVinci: Towards Estimating Mental Workload Modulated by Visual Delays During Telerobotic Surgery - An EEG-based Analysis
Satyam Kumar, J. del R. Millán
ClassificationBiomedical Data
🎯 What it does: Record EEG data from 9 users performing the Peg Transfer task on the da Vinci Research Kit under three visual delay conditions, and extract spectral features to estimate mental workload and detect visual delays.
CoGrasp: 6-DoF Grasp Generation for Human-Robot Collaboration
Abhinav K. Keshari, A. H. Qureshi
GenerationRobotic IntelligenceReinforcement Learning from Human FeedbackConvolutional Neural Network
🎯 What it does: Propose the CoGrasp method, which uses a deep neural network to generate robot grasping poses that consider human preferences, achieving human-robot collaborative grasping of objects.
COLA: COarse LAbel pre-training for 3D semantic segmentation of sparse LiDAR datasets
Jules Sanchez, Franccois Goulette
SegmentationAutonomous DrivingSupervised Fine-TuningPoint Cloud
🎯 What it does: Proposes a pre-training task (COLA) using coarse labels for semantic segmentation on 3D sparse LiDAR datasets to improve performance when fine-tuning on small datasets.
Collaborative Control Based on Payload- leading for the Multi-quadrotor Transportation Systems
Yuan Ping, Jinjin Guo
Robotic Intelligence
🎯 What it does: A load-leading cooperative control method is proposed for multi-quadrotor UAV transportation systems, aiming to maintain the relative distance between the quadrotor and the load as constant as possible during transportation, thereby ensuring the attitude stability of the load.
Collaborative Robotic Biopsy with Trajectory Guidance and Needle Tip Force Feedback
R. Mieling, A. Schlaefer
Robotic IntelligenceBiomedical Data
🎯 What it does: Designed and evaluated a collaborative robotic biopsy system that combines trajectory guidance with tip force feedback to assist doctors in accurately placing needles.
Collaborative Scheduling with Adaptation to Failure for Heterogeneous Robot Teams
Peng Gao, Haotian Zhang
OptimizationRobotic IntelligenceGraph Neural Network
🎯 What it does: Proposed a method that combines deep bipartite graph matching with imitation learning for collaborative scheduling in heterogeneous robot teams, achieving adaptive substitution when robots fail.
Collision Detection and Contact Point Estimation Using Virtual Joint Torque Sensing Applied to a Cobot
Dario Zurlo, Alessandro De Luca
Anomaly DetectionRobotic Intelligence
🎯 What it does: Proposed a complete collision detection, isolation, and response scheme, which was implemented and tested on a 6-degree-of-freedom industrial collaborative robot.
Collision-aware In-hand 6D Object Pose Estimation using Multiple Vision-based Tactile Sensors
Ga Caddeo, Lorenzo Natale
Pose EstimationConvolutional Neural NetworkMultimodality
🎯 What it does: Studied how to use multi-modal visual-tactile sensors to estimate the 6D pose of an object in the hand, using geometric reasoning and convolutional neural networks (CNNs) to filter contact hypotheses, followed by gradient descent optimization of the pose and penalizing solutions that collide with the sensor, ultimately obtaining a pose consistent with actual contact.
Collision-free Coverage Path Planning for the Variable-speed Curvature-constrained Robot
Lin Li, Hengzhu Liu
OptimizationRobotic Intelligence
🎯 What it does: Designed a collision-risk-free, variable-speed Dubins path coverage path planning method (CFC) applicable to known environments with obstacles.
Combining Motion and Appearance for Robust Probabilistic Object Segmentation in Real Time
Vito Mengers, O. Brock
SegmentationPose EstimationRecurrent Neural NetworkOptical FlowPoint Cloud
🎯 What it does: Propose a real-time probabilistic object segmentation method based on motion and appearance, and achieve robust object segmentation by fusing the two features through an interconnected recursive estimator.
Combining Scene Coordinate Regression and Absolute Pose Regression for Visual Relocalization
Jiahao Ruan, Hong Zhang
Pose EstimationConvolutional Neural NetworkSimultaneous Localization and MappingImage
🎯 What it does: Proposed a joint network for scene coordinate regression and absolute pose regression to achieve single-image visual localization.
Communication-Critical Planning via Multi-Agent Trajectory Exchange
Nathan Glaser, Z. Kira
Autonomous DrivingOptimizationImageBenchmark
🎯 What it does: Proposed a learnable, cost-map-based multi-agent planning mechanism for joint perception and planning.
Comparison of Model-Based and Model-Free Reinforcement Learning for Real-World Dexterous Robotic Manipulation Tasks
David Valencia, Henry Williams
Robotic IntelligenceReinforcement LearningWorld Model
🎯 What it does: Evaluated the feasibility of model-based reinforcement learning (MBRL) in two real-world dexterous manipulation tasks, using low-cost robotic grippers to learn predictive models and control policies from scratch.
Completely Rational $\text{SO}(n)$ Orthonormalization
Jin Wu, Ming Liu
OptimizationComputational Efficiency
🎯 What it does: Propose a completely rational universal iterative formula to solve SO(n) orthogonalization (high-dimensional closest rotation problem) and develop the corresponding SO(n) neural network.
Compliant Finger Joint with Controlled Variable Stiffness based on Twisted Strings Actuation
M. Dragusanu, M. Malvezzi
Robotic Intelligence
🎯 What it does: Propose a passive elastic joint that achieves adjustable stiffness through the control of preload using a torsion spring actuator (TSA)
Compliant microgripper using soft polymer actuator
Jung-Hwan Youn, Ki-Uk Kyung
Robotic Intelligence
🎯 What it does: Designed and fabricated a soft polymer-driven, synthesizable micro-gripper capable of achieving rapid linear displacement and grasping operations at the millimeter scale;
Computational Design of 3D-Printable Compliant Mechanisms with Bio-Inspired Sliding Joints
Felipe Velasquez, Stelian Coros
Optimization
🎯 What it does: Developed a computational design tool that integrates forward and inverse simulation for designing 3D-printable flexible mechanisms, enabling biomimetic sliding joints to generate complex motions
Computational Design of Closed-Chain Linkages: Hopping Robot Driven by Morphological Computation
K.V. Nasonov, S. Kolyubin
OptimizationRobotic Intelligence
🎯 What it does: Proposed a closed-chain linkage design method based on morphological computation, synthesized and fabricated a jumping robot using low-performance servos, variable-length links, and elastic distribution
Computational Methods to Support Prototyping of an Adaptive Robot Joystick Controller for Children with Upper Limb Impairments
Mélanie Jouaiti, K. Dautenhahn
Robotic IntelligenceTime Series
🎯 What it does: Developed an adaptive robotic joystick controller to help children with upper limb impairments play games, and determine whether recalibration is needed by monitoring motion statistics.
Computational Modeling in System with Non-Circular Timing Pulleys
Renzo Caballero, Eric Feron
Pose EstimationPhysics Related
🎯 What it does: Analyze and model belt drive systems with non-circular timing pulleys, validate the model experimentally using a 3D printer, extend the model to more complex systems involving multiple pulleys, slip, and non-ideal tension, and finally provide two simulation examples of non-circular timing pulleys.
Computational Tradeoff in Minimum Obstacle Displacement Planning for Robot Navigation
Antony Thomas, M. Robba
OptimizationComputational EfficiencyRobotic Intelligence
🎯 What it does: Proposes an approximate solution method for the minimum obstacle displacement (MOD) planning problem in mobile robot navigation.
Concentration of Measure Phenomenon and its Implications for Sample-based Planning Algorithms in Very-High Dimensional Configuration Spaces
J. Esposito
Robotic Intelligence
🎯 What it does: Explored the concentration phenomenon of random sampling points in extremely high-dimensional configuration spaces, and analyzed their impact on RRT and PRM, two sampling-based planning algorithms.
Concept Design of a New XY Compliant Parallel Manipulator With Spatial Configuration
Z. Lyu, Qingsong Xu
Robotic Intelligence
🎯 What it does: Proposed a conceptual design of an XY compliant parallel manipulator (CPM) with spatial layout, and verified its performance through theoretical modeling, finite element simulation, and experimental testing.
ConDA: Unsupervised Domain Adaptation for LiDAR Segmentation via Regularized Domain Concatenation
Lingdong Kong, Venice Erin Liong
SegmentationDomain AdaptationPoint Cloud
🎯 What it does: Proposes ConDA, a stitching-based LiDAR segmentation unsupervised domain adaptation framework that transfers source domain knowledge by constructing an intermediate domain and performing self-training.
Conditional GANs for Sonar Image Filtering with Applications to Underwater Occupancy Mapping
Tianxiang Lin, M. Kaess
RestorationGenerative Adversarial NetworkImage
🎯 What it does: Train a model using conditional GAN to denoise sonar images and apply it to estimate underwater occupancy maps
Conflict-constrained Multi-agent Reinforcement Learning Method for Parking Trajectory Planning
Siyuan Chen, Wenjie Song
Autonomous DrivingReinforcement Learning
🎯 What it does: Proposes a distributed multi-agent reinforcement learning method for coordinating multiple vehicles in an automated parking system.
Congestion Prediction for Large Fleets of Mobile Robots
Ge Yu, Michael T. Wolf
Robotic IntelligenceConvolutional Neural NetworkRecurrent Neural NetworkTime SeriesSequential
🎯 What it does: Propose to use deep learning to predict future space-time congestion delays in large-scale multi-robot systems, and apply the prediction results to path planning and travel time estimation.
Constant Distance and Orientation Following of an Unknown Surface with a Cable-Driven Parallel Robot
T. Rousseau, F. Chaumette
OptimizationRobotic IntelligenceImage
🎯 What it does: Proposed and verified a constant distance and orientation tracking control scheme for unknown surfaces
Constraint Manifolds for Robotic Inference and Planning
Yetong Zhang, F. Dellaert
OptimizationRobotic Intelligence
🎯 What it does: Proposed and implemented a framework that converts arbitrary nonlinear equality-constrained optimization problems into unconstrained manifold optimization problems, and evaluated it on various constraint reasoning and planning tasks.
Contact Based Turning Gait of a Novel Legged-Wheeled Quadruped
Alper Yeldan, G. Soh
Robotic IntelligenceVideo
🎯 What it does: Studied the motion and turning of the four-legged wheeled robot QuadRunner with semicircular wheel design, proposed a dual-frequency gait planning method to control the duty cycle of the gait period and generate unique turning gait patterns, and validated them through experiments and high-speed cameras.
Contact Force Control with Continuously Compliant Robotic Legs
Robin Bendfeld, C. Remy
Robotic Intelligence
🎯 What it does: Designed a novel structure for a deformable robot leg and proposed a model-based controller to stably and precisely regulate ground contact forces during support phases, which was subsequently validated through bench experiments on a modified ScarlETH leg.
Contact Optimization for Non-Prehensile Loco-Manipulation via Hierarchical Model Predictive Control
Alberto Rigo, Quan Nguyen
OptimizationRobotic Intelligence
🎯 What it does: Proposed a two-level model predictive control (MPC) framework to optimize contact forces and contact positions for quadruped robots in non-grasping transport tasks, and adjust interaction forces through robot locomotion.
Contact-Based Pose Estimation of Workpieces for Robotic Setups
Yitaek Kim, Christoffer Sloth
Pose EstimationOptimizationRobotic IntelligencePoint CloudTime Series
🎯 What it does: Proposes a method for contact-based workpiece pose estimation using collaborative robots, which obtains position data along arbitrary paths on the workpiece and estimates surface normals based on contact forces;
Context-aware robot control using gesture episodes
Petr Vanc, K. Štěpánová
Robotic IntelligenceMultimodality
🎯 What it does: Proposed a robot control method that infers user intent by utilizing gesture segments, context, and common sense.
Contextual Multi-Objective Path Planning
Anna Nickelson, W. Smart
Autonomous DrivingOptimization
🎯 What it does: Proposes the Contextual Multi-Objective Path Planning (CMOPP) algorithm, which separates path planning and path cost estimation into two steps to achieve context-based multi-objective path planning.
Contingency-Aware Task Assignment and Scheduling for Human-Robot Teams
N. Dhanaraj, Sandeep K. S. Gupta
OptimizationRobotic IntelligenceReinforcement LearningTabular
🎯 What it does: Studying task allocation and scheduling in human-robot teams within high-mix low-volume environments, aiming to efficiently complete complex tasks such as satellite assembly.
Continuity-Aware Latent Interframe Information Mining for Reliable UAV Tracking
Changhong Fu, Chongjun Liu
Object TrackingConvolutional Neural NetworkTransformerVideo
🎯 What it does: Proposes a continuity-aware latent cross-frame information mining framework called ClimRT, aimed at enhancing the reliability of drone tracking.
Continuous Prediction of Leg Kinematics during Walking using Inertial Sensors, Smart Glasses, and Embedded Computing
Oleksii Tsepa, Alex Mihailidis
Pose EstimationComputational EfficiencyMultimodalityTime SeriesBiomedical Data
🎯 What it does: Developed the KIFNet system, which uses a lightweight and efficient deep learning model to continuously predict leg kinematics during walking, achieving data fusion through inertial sensors, smart glasses, and embedded computing.
Continuous-Time Gaussian Process Motion-Compensation for Event-Vision Pattern Tracking with Distance Fields
C. Gentil, Teresa Vidal-Calleja
Object Tracking
🎯 What it does: Proposes a motion compensation method based on continuous-time Gaussian processes for pattern tracking in event cameras.
Continuous-Time LiDAR-Inertial-Vehicle Odometry Method with Lateral Acceleration Constraint
Bin He, Yu Zhang
Autonomous DrivingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Propose a continuous-time LiDAR-IMU-vehicle measurement fusion odometry method, incorporating lateral acceleration constraints and online vehicle correction factor estimation.
Contour Context: Abstract Structural Distribution for 3D LiDAR Loop Detection and Metric Pose Estimation
Binqian Jiang, S. Shen
Pose EstimationRetrievalSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes Contour Context, a simple and efficient topological loop detection and 3-DoF metric pose estimation pipeline based on a hierarchical distribution of bird's-eye view (BEV) levels generated from 3D LiDAR projections.
Control of Shape Memory Alloy Actuator via Electrostatic Capacitive Sensor for Meso-scale Mirror Tilting System
Baekgyeom Kim, Je-sung Koh
Robotic Intelligence
🎯 What it does: This paper develops a shape memory alloy (SMA) artificial muscle actuator with capacitive displacement sensing functionality and integrates it into a micro-mirror system capable of tilting up to 20°;
Controllable Mechanical-domain Energy Accumulators
Sung Y. Kim, D. Braun
Physics Related
🎯 What it does: Designed a lockable compression spring that utilizes a small chuck clutch to achieve passive locking
Controlling an Underactuated AUV as an Inverted Pendulum using Nonlinear Model Predictive Control and Behavior Trees
S. Bhat, Ivan Stenius
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: A control method was researched and implemented to enable an underactuated AUV to maintain a near 90-degree pitch angle at a specific depth, similar to an inverted pendulum. First, an optimal control strategy was offline obtained using nonlinear model predictive control (NMPC) in the high-fidelity Simulink simulation environment. Then, this strategy was combined with a behavior tree (BT) to construct a real-time hybrid controller, which was deployed on the AUV system. The control scheme's reproducibility in inverted pendulum operations was verified through Simulink, Stonefish-ROS simulation, and field experiments on the SAM AUV.
Convolutional Bayesian Kernel Inference for 3D Semantic Mapping
Joey Wilson, Maani Ghaffari
Autonomous DrivingConvolutional Neural NetworkPoint Cloud
🎯 What it does: Proposed the Con-vBKI layer and applied it to real-time 3D semantic mapping
Cooperative Driving in Mixed Traffic of Manned and Unmanned Vehicles based on Human Driving Behavior Understanding
Jiaxing Lu, H. Bai
Autonomous DrivingTime SeriesSequential
🎯 What it does: Propose an HMM-based method to identify driver actions (acceleration, braking, lane changing), and use a probabilistic model to predict the acceleration of human-driven vehicles, achieving safe cooperative driving in mixed traffic.
Coordinate Calibration of a Dual-Arm Robot System by Visual Tool Tracking
Junlei Hu, P. Valdastri
Robotic Intelligence
🎯 What it does: Propose a coordinate calibration method for dual-arm robot systems based on Kronecker product, achieving decoupling of translation and rotation through visual tool tracking.
Cost-Aware Evaluation and Model Scaling for LiDAR-Based 3D Object Detection
Xiaofang Wang, Kris M. Kitani
Object DetectionComputational EfficiencyConvolutional Neural NetworkPoint CloudBenchmark
🎯 What it does: Conduct cost-aware LiDAR 3D object detection evaluation, focusing on analyzing performance under different computational costs by extending the SECOND network, and comparing with state-of-the-art methods such as Voxel R-CNN and PV-RCNN++.
Counter-Hypothetical Particle Filters for Single Object Pose Tracking
Elizabeth A. Olson, O. C. Jenkins
Object TrackingPose Estimation
🎯 What it does: Propose a method that dynamically estimates the reactivation rate of particles in a particle filter using the Counter-Hypothetical likelihood function for single-target 6D pose tracking.
Coupled, closed-system fluidic actuators for use in wearable rehabilitation devices
J. Greig, E. Chadwick
🎯 What it does: Proposed a closed-coupled soft actuator capable of achieving higher bending torque through a single air pump, and designed an experimental setup to quantitatively compare positive pressure, vacuum, and combined actuators.
Covariance Steering for Uncertain Contact-rich Systems
Y. Shirai, A. Raghunathan
OptimizationRobotic Intelligence
🎯 What it does: Proposed a covariance regulation method for uncertain contact systems, utilizing stochastic completion constraints in chance-constrained optimization, and designing a contact-aware closed-loop controller by propagating the matrix over time through particle filtering.
COVINS-G: A Generic Back-end for Collaborative Visual-Inertial SLAM
Manthan Patel, M. Chli
Pose EstimationSimultaneous Localization and MappingImage
🎯 What it does: Provides a generic backend compatible with any VIO frontend to realize a collaborative visual-inertial SLAM system;
CPnP: Consistent Pose Estimator for Perspective-n-Point Problem with Bias Elimination
Guangyang Zeng, Junfeng Wu
Pose Estimation
🎯 What it does: Proposed a PnP consistent estimator named CPnP, which constructs linear equations through measurement model modification and variable elimination, derives a closed-form least squares solution, obtains a consistent estimate by eliminating bias, and further optimizes it using Gauss-Newton iteration.
CPSeg: Cluster-free Panoptic Segmentation of 3D LiDAR Point Clouds
Enxu Li, Bingbing Liu
SegmentationDepth EstimationAutonomous DrivingPoint Cloud
🎯 What it does: Developed a real-time end-to-end LiDAR point cloud panoptic segmentation network called CPSeg, capable of simultaneously performing semantic and instance segmentation;
Credible Online Dynamics Learning for Hybrid UAVs
David Rohr, R. Siegwart
Robotic Intelligence
🎯 What it does: Proposed a data-efficient, probabilistic semi-parametric dynamics modeling method for online filtering inference of nonlinear 6DoF dynamics of hybrid unmanned aerial vehicles (H-UAV).
Croche-Matic: a robot for crocheting 3D cylindrical geometry
Gabriella Perry, Nathan Melenbrink
Robotic Intelligence
🎯 What it does: Developed Croche-Matic, a radial crocheting robot capable of generating three-dimensional cylindrical geometries.
CropNav: a Framework for Autonomous Navigation in Real Farms
M. V. Gasparino, Girish Chowdhary
Autonomous DrivingPoint CloudAgriculture Related
🎯 What it does: A hybrid navigation system was developed that can automatically switch between multiple perception modes (such as LiDAR row following and waypoint path tracking) under crop canopies, and achieve navigation failure detection and automatic recovery, supporting full-field navigation both inside and outside the field.
Cross-Agent Relocalization for Decentralized Collaborative SLAM
Philipp Bänninger, M. Chli
OptimizationRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Proposes a strategy for efficiently sharing map regions built by different agents in a decentralized collaborative SLAM system to eliminate map redundancy and maintain estimation consistency.
Cross-domain Transfer Learning and State Inference for Soft Robots via a Semi-supervised Sequential Variational Bayes Framework
Shageenderan Sapai, S. Nurzaman
Domain AdaptationRobotic IntelligenceRecurrent Neural NetworkTime SeriesSequential
🎯 What it does: Propose a semi-supervised sequential variational Bayesian framework for transfer learning and state inference in soft robotic manipulators with missing state labels.
Cross-Modal Monocular Localization in Prior LiDAR Maps Utilizing Semantic Consistency
Chi Zhang, Ming Yang
Autonomous DrivingSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: Proposes a monocular visual localization system based on cross-modal registration, localized to a prior LiDAR map.
Cross-Modality Time-Variant Relation Learning for Generating Dynamic Scene Graphs
Jingyi Wang, Zhidong Deng
GenerationTransformerVision-Language-Action ModelVideoTextMultimodality
🎯 What it does: Proposed a time-varying relationship-aware Transformer (TR2) for generating dynamic scene graphs from videos and modeling temporal changes in cross-frame relationships
CrossDTR: Cross-view and Depth-guided Transformers for 3D Object Detection
Ching-Yu Tseng, Winston H. Hsu
Object DetectionDepth EstimationTransformerImageMultimodality
🎯 What it does: Proposes the CrossDTR method, which includes a lightweight depth predictor and a cross-perspective depth-guided Transformer for 3D object detection.
CueCAn: Cue-driven Contextual Attention for Identifying Missing Traffic Signs on Unconstrained Roads
Varun Gupta, Rohit Saluja
Object DetectionSegmentationConvolutional Neural NetworkVideoBenchmark
🎯 What it does: Proposes the CueCAn model and the MTSVD dataset for detecting missing traffic signs.
CUREE: A Curious Underwater Robot for Ecosystem Exploration
Yogesh A. Girdhar, T. Mooney
Object TrackingAutonomous DrivingRobotic IntelligenceImageMultimodalityAudio
🎯 What it does: Developed a curiosity-driven underwater robot platform named CUREE, integrating low-altitude visual surveys, acoustic scene investigations, habitat characterization, and animal following capabilities. It conducted two field deployments in coral reefs of the US Virgin Islands, verifying its ability to identify habitats preferred by snapping shrimp and follow groupers and jellyfish.
CuRobo: Parallelized Collision-Free Robot Motion Generation
Balakumar Sundaralingam, D. Fox
OptimizationRobotic Intelligence
🎯 What it does: To address the collision-agnostic motion generation problem for robotic arms, this paper proposes and implements a parallel optimization technique based on global motion optimization, encapsulated as a GPU-accelerated library CuRobo.