ICRA 2024 Papers — Page 4
IEEE International Conference on Robotics and Automation · 1760 papers
Characterizing Physical Adversarial Attacks on Robot Motion Planners
Wenxi Wu, Martim Brandão
Adversarial AttackPhysics Related
🎯 What it does: Proposed and experimentally verified physical adversarial attacks on robot motion planners, namely 'planner failure' and 'blindspot' attacks, demonstrating that subtle modifications to the physical environment can cause planner failure or collisions.
Chasing Day and Night: Towards Robust and Efficient All-Day Object Detection Guided by an Event Camera
Jiahang Cao, Lin Wang
Object DetectionSpiking Neural NetworkOptical FlowImageMultimodality
🎯 What it does: Propose the EOLO framework to achieve RGB and event camera fusion for object detection under all-day lighting conditions.
ChatAdp: ChatGPT-powered Adaptation System for Human-Robot Interaction
Zhidong Su, Weihua Sheng
Robotic IntelligenceTransformerLarge Language ModelText
🎯 What it does: Propose a ChatGPT-based adaptation system called ChatAdp to reduce the amount of user feedback in human-robot interaction while achieving good adaptation results.
Choosing the Right Tool for the Job: Online Decision Making over SLAM Algorithms
Samer B. Nashed, S. Zilberstein
Robotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Propose and implement multiple low-cost SLAM frontends running in parallel, with online selection of the most promising feature set
Circular Field Motion Planning for Highly-Dynamic Multi-Robot Systems with Application to Robot Soccer
Fabrice Zeug, M. A. Müller
Robotic Intelligence
🎯 What it does: Propose a new reactive circular field motion planner specifically designed for simultaneously controlling multiple omnidirectional robots in robot soccer competitions, achieving safe and efficient motion planning.
CITR: A Coordinate-Invariant Task Representation for Robotic Manipulation
Peter So, Sami Haddadin
ClassificationRepresentation LearningRobotic IntelligenceImage
🎯 What it does: Propose a coordinate-invariant feature space transformation based on the pairwise inner product of input measurements, and use this to construct task fingerprint image representations for task classification.
CLIP-Loc: Multi-modal Landmark Association for Global Localization in Object-based Maps
Shigemichi Matsuzaki, Kazuhiro Shintani
Computational EfficiencyVision Language ModelImageTextMultimodality
🎯 What it does: Propose a multi-modal data association method that utilizes natural language descriptions of landmarks and aligns camera images with a Vision Language Model to achieve global localization based on object maps;
CLIPUNetr: Assisting Human-robot Interface for Uncalibrated Visual Servoing Control with CLIP-driven Referring Expression Segmentation
Chen Jiang, Martin Jagersand
Robotic IntelligenceConvolutional Neural NetworkVision Language ModelImageTextMultimodality
🎯 What it does: Proposed a CLIP-based referential expression segmentation network called CLIPUNetr for robot-human interaction in uncalibrated visual servo control.
Close the Sim2real Gap via Physically-based Structured Light Synthetic Data Simulation
Kaixin Bai, Jianwei Zhang
Object DetectionSegmentationData SynthesisDepth EstimationDomain AdaptationRobotic IntelligenceImagePhysics Related
🎯 What it does: Proposed and implemented a physics-driven structured light simulation system capable of generating synthetic data containing both RGB and physically accurate depth maps, and based on this, created an RGBD dataset specifically for industrial robot grasping scenarios; subsequently evaluated this dataset on multiple tasks including object detection, instance segmentation, and embedding simulated visual perception into industrial robot grasping tasks;
Closing the Perception-Action Loop for Semantically Safe Navigation in Semi-Static Environments
Jingxing Qian, Angela P. Schoellig
Autonomous DrivingOptimizationRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Developed a closed-loop perception-action pipeline that integrates online-built dense maps, object-level semantics, and consistency estimation into control barrier functions (CBF) for safe navigation.
Closing the Visual Sim-to-Real Gap with Object-Composable NeRFs
Nikhil Mishra, Xi Chen
Data SynthesisDomain AdaptationNeural Radiance FieldMultimodality
🎯 What it does: Proposed a real-time simulation data synthesis pipeline based on Composable Object Volume NeRF (COV-NeRF), which can extract objects from real images, synthesize new scenes, and generate realistic renderings along with various 2D/3D supervisory data.
CloudGripper: An Open Source Cloud Robotics Testbed for Robotic Manipulation Research, Benchmarking and Data Collection at Scale
M. Zahid, Florian T. Pokorny
Robotic IntelligenceImageVideoBenchmark
🎯 What it does: Introduce and implement an open-source cloud robotics test platform composed of 32 small robotic arm work units, assess the repeatability of robotic arms, analyze the throughput and latency of the local network API, and collect the CloudGripper-Rope-100 dataset.
CNS: Correspondence Encoded Neural Image Servo Policy
An-Jen Chen, Rong Xiong
Pose EstimationRobotic IntelligenceGraph Neural NetworkGraph
🎯 What it does: Encode key points and their correspondences into a graph structure, employing a graph neural network as a controller to achieve image steering strategies; simultaneously introduce photorealistic data generation, feature clustering, and distance decoupling techniques to enhance efficiency, accuracy, and generalization capabilities.
Co-Axial Slender Tubular robot (CAST): Towards Robotized Operation for Transorbital Neurosurgery with Minimal Invasiveness
Shuai Wang, Hongbin Liu
Robotic Intelligence
🎯 What it does: Developed a dual-segment slender tubular surgical robot arm (CAST) for transorbital neurosurgery.
Co-Design Optimisation of Morphing Topology and Control of Winged Drones
Fabio Bergonti, D. Floreano
OptimizationPhysics Related
🎯 What it does: Proposes a co-design optimization method for morphing UAVs to generate conceptual designs including topology, actuation, deformation strategies, and controller parameters.
Co-Designing Manipulation Systems Using Task-Relevant Constraints
Apoorv Vaish, Oliver Brock
OptimizationRobotic Intelligence
🎯 What it does: Structuralize the collaborative design space by utilizing environmental constraints, construct a collaborative Jacobian matrix, and perform joint optimization of hardware and control strategies for robotic arms, grippers, and multi-fingered hands based on this framework.
Co-learning Planning and Control Policies Constrained by Differentiable Logic Specifications
Zikang Xiong, Suresh Jagannathan
Robotic IntelligenceReinforcement LearningImage
🎯 What it does: Propose a novel reinforcement learning method that jointly learns planning and control strategies, constrained by differentiable logic specifications, to solve high-dimensional robot navigation tasks.
COAST: COnstraints And STreams for Task and Motion Planning
Brandon Vu, Jeannette Bohg
Robotic Intelligence
🎯 What it does: Proposed the COAST algorithm for Task and Motion Planning (TAMP), which combines streaming motion planning with constrained task planning to address long-term robotic tasks.
CoBRA: A Composable Benchmark for Robotics Applications
M. Mayer (Technical University of Munich), M. Althoff (Technical University of Munich)
OptimizationRobotic IntelligencePoint CloudBenchmark
🎯 What it does: Created a composable robotics application benchmark suite, providing unified formats for robot, environment, and task descriptions, along with tasks such as machining and welding, as well as synthetic and real 3D scan data.
CoBT: Collaborative Programming of Behaviour Trees from One Demonstration for Robot Manipulation
Aayush Jain, M. Leva
Robotic Intelligence
🎯 What it does: CoBT is a collaborative programming framework based on demonstrations that generates reactive and modular behavior trees to enable robotic manipulation.
CoFRIDA: Self-Supervised Fine-Tuning for Human-Robot Co-Painting
Peter Schaldenbrand, Jean Oh
Robotic IntelligenceTransformerSupervised Fine-TuningVision Language ModelDiffusion modelImageText
🎯 What it does: Developed the CoFRIDA framework to enable human-robot collaborative painting, allowing robots to modify and interact on existing canvases; proposed a self-supervised fine-tuning method to enable pre-trained text-image models to understand robot constraints and perform realistic edits;
COIN-LIO: Complementary Intensity-Augmented LiDAR Inertial Odometry
Patrick Pfreundschuh, Olov Andersson
Pose EstimationAutonomous DrivingOptimizationSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposed the COIN-LIO LiDAR-inertial odometry system, which combines LiDAR intensity information with geometric point cloud registration to enhance robustness in geometric degenerate environments (e.g., tunnels, flat fields).
ColAG: A Collaborative Air-Ground Framework for Perception-Limited UGVs’ Navigation
Zhehan Li, Yanjun Cao
Autonomous DrivingOptimizationSimultaneous Localization and Mapping
🎯 What it does: Proposed a collaborative aerial-ground framework named ColAG to address the autonomous navigation problem of blind UGVs in unknown obstacle environments;
Colibri5: Real-Time Monocular 5-DoF Trocar Pose Tracking for Robot-Assisted Vitreoretinal Surgery
Shervin Dehghani, M. Ali Nasseri
Pose EstimationRobotic IntelligenceVideo
🎯 What it does: Developed a real-time markerless 3D surgical instrument trocar pose tracking method using a monocular camera
Collaborative Decision-Making Using Spatiotemporal Graphs in Connected Autonomy
Peng Gao, Ming C. Lin
ClassificationAutonomous DrivingGraph Neural NetworkGraph
🎯 What it does: Proposes an efficient method that converts connected vehicle observation sequences into a spatiotemporal graph and employs a spatiotemporal graph neural network based on heterogeneous graph learning for collaborative decision-making, aiming to control the ego vehicle's safe driving in accident-prone scenarios.
Collaborative Dynamic 3D Scene Graphs for Automated Driving
Elias Greve, Abhinav Valada
Autonomous DrivingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Propose a collaborative urban scene graph (CURB-SG), which constructs large-scale maps using multi-agent panoramic LiDAR data, integrates collaborative SLAM, lane graphs, and multi-layer scene graphs to achieve high-level reasoning and efficient querying for autonomous driving scenarios.
Collaborative Manipulation of Deformable Objects with Predictive Obstacle Avoidance
Burak Aksoy, John T. Wen
Robotic Intelligence
🎯 What it does: Designed and implemented a collaborative robot obstacle avoidance controller that combines position-based dynamics for deformable object motion prediction with control barrier functions, enabling real-time motion adjustment in multi-robot environments to avoid collisions and prevent excessive stretching of deformable objects.
Collision Avoidance and Navigation for a Quadrotor Swarm Using End-to-end Deep Reinforcement Learning
Zhehui Huang, G. Sukhatme
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed an end-to-end deep reinforcement learning method to control a multirotor drone fleet for collision avoidance and navigation in obstacle-filled environments.
Collision Detection and Avoidance for Black Box Multi-Robot Navigation
Sara Ayoubi, Antonio Massaro
Robotic IntelligenceRecurrent Neural Network
🎯 What it does: Propose the CODAK system to achieve decentralized collision detection and avoidance for heterogeneous industrial robots in black-box environments.
Colmap-PCD: An Open-source Tool for Fine Image-to-point cloud Registration
Chunge Bai, Xiang Gao
Pose EstimationSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: Proposed a low-cost reconstruction pipeline that utilizes pre-established LiDAR point cloud maps as fixed constraints to finely register images to point cloud maps, achieving high-precision image-point cloud alignment.
ColonMapper: topological mapping and localization for colonoscopy
Javier Morlana, J. Montiel
TransformerSimultaneous Localization and MappingBiomedical Data
🎯 What it does: Proposed a topological mapping and localization system capable of handling shape and illumination variations in real human colonoscopies.
CoLRIO: LiDAR-Ranging-Inertial Centralized State Estimation for Robotic Swarms
Shipeng Zhong, Ming Liu
Robotic IntelligenceSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes a centralized LiDAR-Ranging-Inertial state estimation system that enables robot swarms to collaborate in GPS-denied environments.
Combining Coordination and Independent Coverage in MultiRobot Graph Patrolling
Carlos Diaz Alvarenga, Stefano Carpin
OptimizationRobotic Intelligence
🎯 What it does: Propose a multi-robot patrolling method that partitions graph vertices into shared regions and robot-exclusive regions, solving the maximum weighted boredom minimization problem, and providing exact solutions and heuristic algorithms.
Commonsense Spatial Knowledge-aware 3-D Human Motion and Object Interaction Prediction
Sang Uk Lee
Pose EstimationTransformerPoint Cloud
🎯 What it does: Proposed a novel 3D human motion and object interaction prediction model that can perceive common-sense knowledge about human-object interactions, achieving this through joint prediction of human joint motion and human-object interactions; simultaneously combining the two prediction results to explicitly enforce common-sense relationships, such as predicting that when the right hand is expected to contact an object after 1 second, the distance between the right hand and the object should also be predicted as small.
Communicating Intent as Behaviour Trees for Decentralised Multi-Robot Coordination
Rhett Hull, Graeme Best
Robotic Intelligence
🎯 What it does: Proposed a decentralized multi-robot coordination algorithm that utilizes rich intent information representation to encode and communicate each robot's intent.
Communication-Aware Map Compression for Online Path-Planning
Evangelos Psomiadis, Panagiotis Tsiotras
CompressionOptimizationRobotic Intelligence
🎯 What it does: Proposes a communication-aware map compression framework for online path planning of mobile robots, which can dynamically select the optimal compression level based on the robot's path to balance map resolution and communication cost.
Comparison of Rating Scale and Pairwise Comparison Methods for Measuring Human Co-worker Subjective Impression of Robot during Physical Human-Robot Collaboration
Qiao Wang, Chin-Teng Lin
Robotic Intelligence
🎯 What it does: An experiment comparing RS (Rating Scale) and PC (Pairwise Comparison) methods for measuring human colleagues' subjective impressions of robots in physical human-robot collaboration (pHRC).
Complementary Random Masking for RGB-Thermal Semantic Segmentation
Ukcheol Shin, In-So Kweon
SegmentationKnowledge DistillationMultimodality
🎯 What it does: Proposes a complementary random masking strategy for RGB-thermal images and introduces a self-distillation loss between clean and masked input modalities;
Complementing Onboard Sensors with Satellite Maps: A New Perspective for HD Map Construction
Wenjie Gao, Nanning Zheng
Object DetectionSegmentationAutonomous DrivingTransformerSimultaneous Localization and MappingImagePoint Cloud
🎯 What it does: This paper proposes to utilize satellite maps to supplement vehicle-mounted sensors for high-definition (HD) map construction, and releases a corresponding supplementary dataset; meanwhile, a hierarchical fusion module is designed, which uses a mask generator and mask cross-attention to refine vehicle-mounted features at the feature level, and an alignment module to correct coordinate differences at the BEV level; this module can be seamlessly integrated into three existing HD map construction methods; experimental validation on the augmented nuScenes dataset shows significant performance improvements in HD map semantic segmentation and instance detection tasks.
Compliant Robotic Gripper with Integrated Ripeness Sensing for Blackberry Harvesting
Arvyn De, Ai-Ping Hu
Robotic IntelligenceAgriculture Related
🎯 What it does: Developed a flexible robotic gripper with maturity sensing functionality for non-destructive blackberry harvesting.
Composable Interaction Primitives: A Structured Policy Class for Efficiently Learning Sustained-Contact Manipulation Skills
Ben Abbatematteo, G. Konidaris
Computational EfficiencyRobotic IntelligenceReinforcement LearningWorld Model
🎯 What it does: Proposes a Composable Interaction Primitives (CIP) strategy class for efficiently learning operational skills involving continuous contact, and verifies its sequential composition and zero-shot execution capabilities in both simulation and real robots
COMPOSER: Scalable and Robust Modular Policies for Snake Robots
Yuyou Zhang, Ding Zhao
Robotic IntelligenceTransformerReinforcement Learning
🎯 What it does: Propose a modularizable control strategy called COMPOSER based on multi-agent reinforcement learning, leveraging the ultra-redundant and high-dimensional characteristics of serpentine robots. Each segment is treated as an independent agent, with collaborative control achieved through self-attention mechanisms and high-level imagination policies.
Composing Pre-Trained Object-Centric Representations for Robotics From "What" and "Where" Foundation Models
Junyao Shi, D. Jayaraman
Representation LearningRobotic IntelligenceVision-Language-Action ModelImage
🎯 What it does: Propose the POCR framework, which leverages segmentation results from pre-trained models to stably locate entities in scenes ("where" information), and applies other pre-trained models to generate vector descriptions for each entity ("what" information), thereby constructing object-centric representations for robot control.
Compositional Servoing by Recombining Demonstrations
Max Argus, Thomas Brox
Robotic Intelligence
🎯 What it does: Propose a framework that treats the visual servoing task as a graph traversal, achieving multi-task capability through splitting and recombining demonstrations.
Computation-Aware Multi-object Search in 3D Space using Submodular Tree
Yan-Shuo Li, Kuo-Shih Tseng
OptimizationComputational Efficiency
🎯 What it does: Proposes a computation-aware multi-target search algorithm called CASMO, which considers computational time constraints when searching for targets in 3D environments.
ConBaT: Control Barrier Transformer for Safe Robot Learning from Demonstrations
Yue Meng, Ashish Kapoor
OptimizationSafty and PrivacyRobotic IntelligenceTransformer
🎯 What it does: Propose a self-supervised Transformer model called ConBaT based on control barrier functions for learning safe robot behaviors from demonstrations; the model recursively predicts safe actions using a causal Transformer and employs a lightweight online optimization during deployment to ensure future states fall into the learned safe set;
ConceptGraphs: Open-Vocabulary 3D Scene Graphs for Perception and Planning
Qiao Gu, L. Paull
Autonomous DrivingRepresentation LearningData-Centric LearningGraph Neural NetworkVision Language ModelImageGraph
🎯 What it does: Proposed an open-vocabulary graph structure 3D scene representation method called ConceptGraphs.
Conditionally Combining Robot Skills using Large Language Models
K. Zentner, G. Sukhatme
Robotic IntelligenceTransformerLarge Language ModelWorld ModelText
🎯 What it does: Proposed Language-World, an extension of the Meta-World, enabling large language models to manipulate in simulated robot environments through semi-structured natural language queries and script-based natural language descriptions, and introduced the Plan Conditioned Behavioral Cloning (PCBC) method to fine-tune high-level plans through end-to-end example demonstrations.
Conflict Area Prediction for Boosting Search-Based Multi-Agent Pathfinding Algorithms
Jaesung Ryu, Kyungjae Lee
Optimization
🎯 What it does: Propose combining learning-based prediction of conflict regions with Conflict-Based Search (CBS) to enhance the efficiency of multi-agent path planning.
Conflict-Based Model Predictive Control for Scalable Multi-Robot Motion Planning
Ardalan Tajbakhsh, Aaron M. Johnson
OptimizationRobotic Intelligence
🎯 What it does: Proposed a scalable multi-robot motion planning algorithm CB-MPC, which combines a high-level conflict tree and low-level MPC to efficiently resolve conflicts among robots in continuous space, taking into account each robot's kinematics, dynamics, and drive constraints.
Conformal Decision Theory: Safe Autonomous Decisions from Imperfect Predictions
Jordan Lekeufack, Jitendra Malik
Safty and Privacy
🎯 What it does: Proposes Conformal Decision Theory, which enables safe autonomous decision-making in scenarios where machine learning predictions are imperfect.
Conformal Policy Learning for Sensorimotor Control under Distribution Shifts
Huang Huang, Jitendra Malik
Robotic IntelligenceReinforcement Learning
🎯 What it does: Designed strategies capable of receiving conformal quantiles as input to detect and respond to changes in observation distributions in sensor action control (i.e., conformal policy learning).
Constant-time Motion Planning with Anytime Refinement for Manipulation
Itamar Mishani, M. Likhachev
OptimizationRobotic Intelligence
🎯 What it does: A framework is proposed that combines constant-time motion planning (CTMP) with arbitrary improvement methods to rapidly generate an initial motion plan within a user-specified time threshold and iteratively optimize the plan during the remaining planning time to approach the optimal solution.
Constrained Bimanual Planning with Analytic Inverse Kinematics
Thomas Cohn, Russ Tedrake
OptimizationRobotic Intelligence
🎯 What it does: Studies how to perform constrained dual-arm planning on dual robots by parameterizing the configuration space using analytical inverse kinematics to satisfy nonlinear constraints on the end-effector's fixed transformation.
Constrained Hierarchical Monte Carlo Belief-State Planning
Arec L. Jamgochian, M. Kochenderfer
OptimizationRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a COBeTS algorithm based on hierarchical options for belief tree search in online planning for constrained partially observable Markov decision processes (CPOMDP), achieving scalability in large or continuous robotic problems.
Constrained Nonlinear Disturbance Observer for Robotic Systems
Ji Han, Min Jun Kim
OptimizationRobotic Intelligence
🎯 What it does: Proposed the Constrained Nonlinear Robust Internal Loop Compensator (C-NRIC) framework, and designed a motion controller within this framework that can respond to unknown contacts and precisely track desired trajectories during free motion, demonstrating its practicality;
Constrained Passive Interaction Control: Leveraging Passivity and Safety for Robot Manipulators
Zhiquan Zhang, Nadia Figueroa
Safty and PrivacyRobotic Intelligence
🎯 What it does: Proposes a restricted passive interaction control architecture that enables the robot to maintain passivity only when feasible, while satisfying multiple safety constraints such as kinematic limits, self-collision, external collision, and singularities.
Contact Energy Based Hindsight Experience Prioritization
Erdi Sayar, A. Knoll
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposes a contact energy-based priority sampling method (CEBP) to enhance the learning efficiency of Hindsight Experience Replay (HER) in multi-target robotic manipulation tasks, focusing on sampling trajectories with rich contact interactions;
Containerized Vertical Farming Using Cobots
Dasharadhan Mahalingam, IV Ramakrishnan
Robotic IntelligenceReinforcement LearningImageAgriculture Related
🎯 What it does: Automate two key agricultural tasks in containerized vertical farms—seedling transplanting and leafy vegetable harvesting—using collaborative robots (cobots) without requiring task-specific programming.
Continual Driving Policy Optimization with Closed-Loop Individualized Curricula
Haoyi Niu, Jianming Hu
Autonomous Driving
🎯 What it does: Developed a closed-loop individualized curricula framework called Closed-Loop Individualized Curricula (CLIC) for closed-loop optimization of autonomous driving policies in a large-scale pre-collected scenario library.
Continuous Adaptation in Person Re-identification for Robotic Assistance
Federico Rollo, Arash Ajoudani
RecognitionObject TrackingDomain AdaptationRobotic Intelligence
🎯 What it does: Proposes a continuous visual adaptation person re-identification module, enabling robots to identify and follow specific targets in crowded environments even when the target's appearance changes or partial/full occlusion occurs; verifies its effectiveness in laboratory environments and human-robot interaction scenarios.
Continuous Robotic Tracking of Dynamic Targets in Complex Environments Based on Detectability
Zhihao Wang, Haoyao Chen
Object TrackingRobotic Intelligence
🎯 What it does: This paper studies the continuous tracking of single, double, and triple targets in complex environments, analyzes the interaction mechanisms between the robot, environment, and targets, proposes detectability as a general metric, and designs a motion planning framework based on model predictive control (MPC) to achieve continuous and robust tracking of dynamic targets.
Continuously Estimate and Control Prosthetic Grip Force by an Optical Waveguide Sensor
Linhang Ju, Wuxiang Zhang
Robotic IntelligenceBiomedical Data
🎯 What it does: Utilize photonic waveguide sensors embedded with carbon fiber to collect human arm muscle deformation information, achieving continuous control of prosthetic grip strength.
Contrastive Initial State Buffer for Reinforcement Learning
Nico Messikommer, Davide Scaramuzza
Reinforcement LearningContrastive Learning
🎯 What it does: Proposed the Contrastive Initial State Buffer (CISB), which selects states from past experiences to initialize the agent, thereby guiding it to explore more informative states.
Contrastive Learning for Enhancing Robust Scene Transfer in Vision-based Agile Flight
Jiaxu Xing, Davide Scaramuzza
Domain AdaptationRepresentation LearningRobotic IntelligenceContrastive LearningImage
🎯 What it does: Propose an adaptive multi-contrastive learning strategy for visual representation learning, enabling zero-shot scene transfer and validated in agile flight tasks of vision-based quadrotors.
Contrastive Learning-Based Attribute Extraction Method for Enhanced Terrain Classification
Xiao Liu, Haoyao Chen
ClassificationRobotic IntelligenceContrastive Learning
🎯 What it does: Automatically extract terrain attributes from tactile data generated by the interaction between robot legs and terrain using contrastive learning for terrain classification.
Control-Barrier-Aided Teleoperation with Visual-Inertial SLAM for Safe MAV Navigation in Complex Environments
Siqi Zhou, Angela P. Schoellig
Autonomous DrivingSafty and PrivacyRobotic IntelligenceSimultaneous Localization and MappingImageMultimodalityPoint Cloud
🎯 What it does: Developed a safety filter based on Control Barrier Functions (CBF), integrated with visual-inertial SLAM and dense 3D occupancy maps, enabling safe navigation of non-expert-operated MAVs in complex environments.
Controlling FES of arm movements using physics-informed reinforcement learning via co-kriging adjustment
Nat Wannawas, A. Faisal
Robotic IntelligenceReinforcement LearningTime SeriesBiomedical DataPhysics Related
🎯 What it does: Control of functional electrical stimulation to achieve point-to-point reaching with a two-degree-of-freedom planar arm through physics-informed reinforcement learning and co-kriging adjustment
Convergent iLQR for Safe Trajectory Planning and Control of Legged Robots
James Zhu, Aaron M. Johnson
OptimizationRobotic Intelligence
🎯 What it does: Designed a trajectory planning and control framework that integrates worst-case perturbation analysis into iLQR to achieve safe dynamic trajectory planning and control for legged robots in hybrid dynamic environments.
Convolutional Vision Transformer as a Path Following Controller for Omnidirectional Robots
S. Hiremath, N. Bajçinca
OptimizationRobotic IntelligenceConvolutional Neural NetworkTransformerImage
🎯 What it does: Proposed a path tracking controller for omni-directional robots based on deep neural networks, where the controller decomposes the given reference path into multiple shorter sub-paths over a fixed prediction horizon.
Cook2LTL: Translating Cooking Recipes to LTL Formulae using Large Language Models
A. Mavrogiannis, Y. Aloimonos
Robotic IntelligenceTransformerLarge Language ModelText
🎯 What it does: Propose a system called Cook2LTL that translates cooking recipes into linear temporal logic (LTL) formulas and normalizes high-level cooking actions into atomic actions executable by kitchen robots.
Cooperative vs. Teleoperation Control of the Steady Hand Eye Robot with Adaptive Sclera Force Control: A Comparative Study
Mojtaba Esfandiari, I. Iordachita
Robotic IntelligenceBiomedical Data
🎯 What it does: Implemented a teleoperation control mode integrated with adaptive scleral force control on the Steady-Hand Eye Robot, and compared it with the collaborative mode.
Coordinated Landing Control for Cross-Domain UAV-USV Fleets Using Heterogeneous-Feature Matching
Jianing Ding, Bin-Bin Hu
OptimizationRobotic Intelligence
🎯 What it does: Proposed a multi-UAV-multi-USV cooperative landing control algorithm based on heterogeneous feature matching.
CoPAL: Corrective Planning of Robot Actions with Large Language Models
F. Joublin, M. Gienger
Robotic IntelligenceTransformerLarge Language Model
🎯 What it does: Proposed a multi-level cognitive robotic planning system architecture and introduced a new replanning strategy to handle physical, logical, and semantic errors.
CopperTag: A Real-Time Occlusion-Resilient Fiducial Marker
Xu Bian, Donglai Ran
Pose EstimationRobotic IntelligenceImageBenchmark
🎯 What it does: Proposes a real-time occlusion-tolerant tag named CopperTag, specifically designed for tasks such as navigation, parking, and grasping in industrial environments.
Coupled Active Perception and Manipulation Planning for a Mobile Manipulator in Precision Agriculture Applications
Shuangyun Xie, Dezhen Song
OptimizationRobotic IntelligenceAgriculture Related
🎯 What it does: Proposed a coupled active perception and manipulation (CAPM) critical state planning algorithm for mobile manipulators, jointly considering perception uncertainty and operational constraints to generate energy-efficient trajectories.
Covariance Based Terrain Mapping for Autonomous Mobile Robots
Lennart Werner, R. Brockers
Robotic IntelligenceSimultaneous Localization and MappingImage
🎯 What it does: Developed a local robot-centric navigation map based on covariance cells, which can rapidly detect obstacles and hazardous slopes in unknown environments through stereo vision with ultra-short baseline and high elevation angle, achieving constant-time queries for height, obstacle presence, and slope.
CppFlow: Generative Inverse Kinematics for Efficient and Robust Cartesian Path Planning
Jeremy Morgan, G. Sukhatme
OptimizationRobotic Intelligence
🎯 What it does: Proposes CppFlow, an efficient and robust Cartesian path planner.
CrackNex: a Few-shot Low-light Crack Segmentation Model Based on Retinex Theory for UAV Inspections
Zhen Yao, Mooi Choo Choo Chuah
RestorationSegmentationMeta LearningImageBenchmark
🎯 What it does: Proposed the CrackNex framework based on Retinex theory, which learns a unified illumination-invariant representation using reflectance information and addresses the issue of insufficient training data by combining few-shot segmentation.
CrazySim: A Software-in-the-Loop Simulator for the Crazyflie Nano Quadrotor
Christian Llanes, Samuel Coogan
Robotic Intelligence
🎯 What it does: Developed a software-in-the-loop (SIL) simulation platform for the Crazyflie Nano quadrotor drone swarm, and validated its effectiveness through case studies.
CRITERIA: a New Benchmarking Paradigm for Evaluating Trajectory Prediction Models for Autonomous Driving
Changhe Chen, Amir Rasouli
Autonomous DrivingBenchmark
🎯 What it does: Proposed the CRITE-RIA evaluation method, which includes multi-level driving scenario extraction based on road structure, model performance, and data characteristics, as well as new unbiased metrics for diversity and feasibility. Extensive experiments on the Argoverse dataset were conducted on representative prediction models, demonstrating its ability to provide more accurate model rankings and behavioral feature descriptions.
Cross Domain Policy Transfer with Effect Cycle-Consistency
Ruiqi Zhu, O. Çeliktutan
Domain AdaptationRobotic IntelligenceReinforcement Learning
🎯 What it does: Proposes a method for transferring robot policies between different state and action spaces, learning mapping functions using unpaired data, and aligning transfer effects between two domains through effect cycle consistency.
Cross View Capture for Distributed Image Compression with Decoder Side Information
Yankai Yin, Chi Zhu
CompressionImage
🎯 What it does: Proposed a distributed image compression network called CVCDIC based on cross-perspective capture, which improves compression quality by leveraging decoder side information.
Cross-Cluster Shifting for Efficient and Effective 3D Object Detection in Autonomous Driving
Zhili Chen, Qifeng Chen
Object DetectionAutonomous DrivingPoint Cloud
🎯 What it does: Proposed a new point-based 3D detector called Shift-SSD, designed for precise 3D object detection in autonomous driving.
Cross-Modal Registration Using Adaptive Modeling in Infrastructure-based Vehicle Localization*
Fei Wang, Ming Yang
Object TrackingAutonomous DrivingSimultaneous Localization and MappingImageMultimodalityPoint Cloud
🎯 What it does: Propose an infrastructure-based cross-modal localization method that uses LiDAR for object modeling, employs cost-effective cameras for object tracking, and achieves real-time correspondence through image-point cloud registration.
Crosstalk-Free Impedance-Separating Array Measurement for Iontronic Tactile Sensors*
Funing Hou, Shijie Guo
🎯 What it does: A method for impedance separation without requiring complex analog components is proposed, and the Quadri-Terminal Impedance Network (QTIN) model is introduced to reduce crosstalk, the precise measurement range is measured, and a simple denoising method is provided.
CrossVideo: Self-supervised Cross-modal Contrastive Learning for Point Cloud Video Understanding
Yunze Liu, Li Yi
RecognitionContrastive LearningVideoPoint Cloud
🎯 What it does: Proposed a self-supervised cross-modal contrastive learning method called CrossVideo for point cloud video understanding.
Crossway Diffusion: Improving Diffusion-based Visuomotor Policy via Self-supervised Learning
Xiang Li, M. Ryoo
Robotic IntelligenceReinforcement LearningDiffusion modelImage
🎯 What it does: Propose Crossway Diffusion, enhancing diffusion-based visual-motor policy learning through a self-supervised state decoder and reconstruction objective.
CTA-LO: Accurate and Robust LiDAR Odometry Using Continuous-Time Adaptive Estimation
Yuezhang Lv, Yang Jin
Autonomous DrivingSimultaneous Localization and MappingPoint Cloud
🎯 What it does: A high-precision and robust LiDAR odometry method based on continuous time estimation is proposed, along with constructing a point uncertainty model to quantify ranging errors; to address the issue of insufficient constraints caused by excessive variables in continuous time estimation, a marginalization method using B-spline local support is introduced; additionally, residual adaptive weighting and probabilistic point cloud maps based on point uncertainty are incorporated to further improve odometry accuracy.
CURL-MAP: Continuous Mapping and Positioning with CURL Representation†
Kaicheng Zhang, Sen Wang
OptimizationRepresentation LearningSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Leveraging CURL's continuous and ultra-compact representation, the CURL-MAP framework is proposed, achieving variable density 3D LiDAR map reconstruction and localization.
Curriculum-based Sensing Reduction in Simulation to Real-World Transfer for In-hand Manipulation
Lingfeng Tao, Xiaoli Zhang
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a curriculum-based perceptual feature reduction method to gradually reduce rich features from simulation environments to real-world accessible features, aiming to enhance the training efficiency and transfer performance of deep reinforcement learning in physical manipulation tasks, with real-world palm manipulation tasks validated on the Allegro robot hand.
CushSense: Soft, Stretchable, and Comfortable Tactile-Sensing Skin for Physical Human-Robot Interaction
Boxin Xu, T. Bhattacharjee
Robotic Intelligence
🎯 What it does: This paper proposes and implements CushSense—a flexible, stretchable tactile sensing skin based on fabric for physical human-robot interaction (pHRI) tasks, such as robotic caregiving; including sensor design, fabrication processes, property testing, and user experience studies.
CVAE-SM: A Conditional Variational Autoencoder with Style Modulation for Efficient Uncertainty Quantification
Amin Ullah, Fuxin Li
SegmentationComputational EfficiencyAuto EncoderImage
🎯 What it does: Proposed the CVAE-SM framework based on a conditional variational autoencoder (CVAE), enhancing uncertainty quantification in underwater image segmentation through style modulation.
CVFormer: Learning Circum-View Representation and Consistency for Vision-Based Occupancy Prediction via Transformers
Zhengqi Bai, Jiamao Li
Autonomous DrivingTransformer
🎯 What it does: Proposes a Transformer-based framework called CVFormer, which extracts 3D features from the surrounding environment using a 2D panoramic view from the vehicle's perspective to achieve fine-grained 3D semantic occupancy prediction.
Cycle-Correspondence Loss: Learning Dense View-Invariant Visual Features from Unlabeled and Unordered RGB Images
David B. Adrian, H. Neumann
Representation LearningRobotic IntelligenceContrastive LearningImage
🎯 What it does: Proposed and implemented Cycle-Correspondence Loss (CCL) to learn viewpoint-invariant dense visual descriptors, leveraging cycle consistency to detect pixel correspondences and enabling self-supervised training on unpaired RGB images.
D-LGP: Dynamic Logic-Geometric Program for Reactive Task and Motion Planning
Teng Xue, Sylvain Calinon
OptimizationRobotic IntelligenceBenchmark
🎯 What it does: Proposed Dynamic Logic Geometric Program (D-LGP), achieving efficient hybrid planning for task and motion planning (TAMP) problems by integrating dynamic tree search with global optimization.
DA-RAW: Domain Adaptive Object Detection for Real-World Adverse Weather Conditions
Minsik Jeon, Jihong Min
Object DetectionDomain AdaptationContrastive LearningImage
🎯 What it does: Proposed an unsupervised domain adaptation framework for target detection under adverse weather conditions such as rain and snow, achieving more robust detection by addressing the style gap and weather gap separately.
Data-Driven Latent Space Representation for Robust Bipedal Locomotion Learning
Guillermo A. Castillo, Ayonga Hereid
Robotic IntelligenceReinforcement LearningAuto Encoder
🎯 What it does: Proposes a framework that combines data-driven state representation with reinforcement learning (RL)-based walking strategies to learn robust bipedal gaits.
db-CBS: Discontinuity-Bounded Conflict-Based Search for Multi-Robot Kinodynamic Motion Planning
A. Moldagalieva, Wolfgang Honig
OptimizationRobotic Intelligence
🎯 What it does: Proposed a multi-robot dynamic motion planning method called db-CBS, combining conflict-based search (CBS) and discretized discontinuity-bounded A*, achieving collaborative planning for multiple robots through a three-tier strategy.
DCPT: Darkness Clue-Prompted Tracking in Nighttime UAVs
Jiawen Zhu, Huchuan Lu
Object TrackingTransformerPrompt EngineeringVideo
🎯 What it does: Proposes an architecture named Darkness Clue-Prompted Tracking (DCPT) for nighttime drone tracking, which replaces the traditional enhancement-then-tracking workflow with darkness clue prompts.