ICRA 2023 Papers — Page 2
IEEE International Conference on Robotics and Automation · 1341 papers
Achieving Extensive Trajectory Variation in Impulsive Robotic Systems
Luis Viornery, S. Bergbreiter
Robotic IntelligencePhysics Related
🎯 What it does: Developed, constructed, and tested an ejection mechanism inspired by grasshopper legs, achieving a wide range of projectile trajectories through force control of the spring constraint device, along with detailed system modeling and experimental verification;
Active Inference for Autonomous Decision-Making with Contextual Multi-Armed Bandits
Shohei Wakayama, N. Ahmed
OptimizationReinforcement Learning
🎯 What it does: This paper applies active inference methods to contextual multi-armed bandit problems and proposes variational and Laplace approximation methods for computing expected free energy.
Active Metric-Semantic Mapping by Multiple Aerial Robots
Xu Liu, Vijay R. Kumar
OptimizationRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Proposes a method for multiple heterogeneous drones to collaboratively actively construct metric-semantic maps using sparse object models (basic shapes + semantic labels), actively exploring to minimize uncertainty in semantic classification and geometric modeling.
Active Predictive Coding: Brain-Inspired Reinforcement Learning for Sparse Reward Robotic Control Problems
Alexander Ororbia, A. Mali
Robotic IntelligenceReinforcement Learning
🎯 What it does: Proposed a no-backpropagation robot control method based on neural generative encoding, designing an active predictive coding (ActPC) agent composed entirely of predictive processing circuits.
Active Probing and Influencing Human Behaviors Via Autonomous Agents
Shuang Wang, J. Dolan
Autonomous DrivingOptimizationAgentic AI
🎯 What it does: Proposes an online optimization-driven active probing method to help autonomous agents proactively acquire information about human models and interactively influence based on this information.
Active Reward Learning from Online Preferences
Vivek Myers, Dorsa Sadigh
Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement LearningText
🎯 What it does: Design and online presentation of easy-to-answer pairwise action preference queries to learn rewards from human experts and rapidly adapt robot policies.
Actuator Capabilities Aware Limitation for TDPA Passivity Controller Action
Francesco Porcini, A. Frisoli
OptimizationRobotic Intelligence
🎯 What it does: Proposed a new clamping strategy to limit the action of the reactive controller in the time-domain passivity approach (TDPA), and verified its effectiveness in experiments.
ADAPT: A 3 Degrees of Freedom Reconfigurable Force Balanced Parallel Manipulator for Aerial Applications
Kartik Suryavanshi, J. Herder
Robotic IntelligencePhysics Related
🎯 What it does: Proposed and implemented ADAPT—a reconfigurable 3-degree-of-freedom force-balanced parallel manipulator for spatial motion and interaction beneath drones; modeled its kinematic model and validated its motion and force characteristics through MSC ADAMS dynamic simulation and physical experiments.
ADAPT: Action-aware Driving Caption Transformer
Bu Jin, Jingjing Liu
Autonomous DrivingExplainability and InterpretabilityTransformerVision-Language-Action ModelVideo
🎯 What it does: Proposes ADAPT, an end-to-end Transformer-based architecture that jointly trains driving captioning and vehicle control prediction tasks through shared video representations, providing user-friendly natural language narratives and reasoning for each decision step.
Adaptive and Explainable Deployment of Navigation Skills via Hierarchical Deep Reinforcement Learning
Kyowoon Lee, Jaesik Choi
Autonomous DrivingExplainability and InterpretabilityReinforcement Learning
🎯 What it does: Propose a hierarchical deep reinforcement learning framework that learns multiple low-level navigation strategies and adapts and deploys them through high-level strategies.
Adaptive approximation of dynamics gradients via interpolation to speed up trajectory optimisation
D. Russell, M. Dogar
Optimization
🎯 What it does: Accelerate the optimization process by using interpolation to approximate dynamics gradients in trajectory optimization, thereby reducing the frequent computation of finite differences.
Adaptive based Assist-as-needed control strategy for Ankle movement assistance
R. Jradi, S. Mohammed
Robotic IntelligenceBiomedical Data
🎯 What it does: Propose an adaptive active disturbance suppression controller to provide on-demand assistive motion for the ankle joint during the gait phase
Adaptive Heading for Perception-Aware Trajectory Following
J. S. Willners, Y. Pétillot
OptimizationRobotic IntelligenceSimultaneous Localization and Mapping
🎯 What it does: Proposed and implemented an adaptive heading method that adjusts the robot's heading to enhance feature tracking in mapped environments, thereby improving localization accuracy, particularly suitable for GPS-denied environments such as underwater or cave environments;
Adaptive Keyframe Generation based LiDAR Inertial Odometry for Complex Underground Environments
Boseong Kim, Ali-akbar Agha-mohammadi
Pose EstimationOptimizationSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes a LiDAR-Inertial Odometry (LIO) algorithm based on adaptive keyframe generation for achieving fast and accurate state estimation in complex underground environments.
Adaptive Optimal Electrical Resistance Tomography for Large-Area Tactile Sensing
Wendong Zheng, Wuqiang Yang
OptimizationComputed Tomography
🎯 What it does: Proposed an adaptive optimal resistance imaging driving strategy suitable for large-area tactile perception, which can adaptively select current injection and voltage measurement modes based on the initially detected contact area, thereby optimizing tactile stimulation and enhancing perception performance.
Adaptive Risk-Tendency: Nano Drone Navigation in Cluttered Environments with Distributional Reinforcement Learning
Cheng Liu, G. D. Croon
Reinforcement Learning
🎯 What it does: Propose a distributed reinforcement learning framework to generate adaptive risk-averse strategies, studying navigation of nano quadrotor robots in unknown crowded environments.
Adaptive Sampling-based Particle Filter for Visual-inertial Gimbal in the Wild
Xueyang Kang, P. Vandewalle
Object TrackingPose EstimationConvolutional Neural NetworkImageMultimodality
🎯 What it does: Propose a tracking and fusion algorithm based on computer vision, specifically designed for a 3D-printed gimbal system flying in natural environments, utilizing the skyline and ground plane as references to achieve robust camera orientation control.
Adaptive-SpikeNet: Event-based Optical Flow Estimation using Spiking Neural Networks with Learnable Neuronal Dynamics
A. Kosta, Kaushik Roy
Spiking Neural NetworkOptical Flow
🎯 What it does: Using a Spiking Neural Network (SNN) with fully spiking, learnable neural dynamics for optical flow estimation in event cameras.
AdaSfM: From Coarse Global to Fine Incremental Adaptive Structure from Motion
Yu Chen, G. Lee
Pose EstimationAutonomous DrivingSimultaneous Localization and MappingImage
🎯 What it does: Propose AdaSfM, a coarse-to-fine adaptive Structure from Motion method, which first uses IMU and wheel speedometer for rough global SfM to enhance view graph reliability, then divides the view graph into sub-scenes for parallel local incremental SfM refinement, and finally aligns all local reconstructions to the global coordinate system through a threshold adaptive strategy.
Advanced Skills through Multiple Adversarial Motion Priors in Reinforcement Learning
Eric Vollenweider, Marco Hutter
Robotic IntelligenceReinforcement LearningGenerative Adversarial Network
🎯 What it does: Propose an improved adversarial motion prior reinforcement learning method that supports multiple switchable motion styles and simultaneously learns multiple skills on a robot.
AeriaLPiPS: A Local Planner for Aerial Vehicles with Geometric Collision Checking
Justin S. Smith, P. Vela
Autonomous DrivingRobotic Intelligence
🎯 What it does: Proposed a perception-space-based aerial local planner called AeriaLPiPS, which achieves safe navigation of MAVs with non-ideal geometry using egocan and 3D Gap methods
Agile and Versatile Robot Locomotion via Kernel-based Residual Learning
Milo Carroll, Zhibin Li
Robotic IntelligenceReinforcement Learning
🎯 What it does: Developed a kernel-based residual learning framework for quadrupedal robot locomotion.
AI-Based Multi-Object Relative State Estimation with Self-Calibration Capabilities
Thomas Jantos, J. Steinbrener
Pose EstimationRobotic IntelligenceImageMultimodality
🎯 What it does: Propose a method that fuses AI-based image pose estimator with inertial measurement unit (IMU) data for 6-degree-of-freedom relative state estimation of multiple objects by mobile robots.
AIMY: An Open-source Table Tennis Ball Launcher for Versatile and High-fidelity Trajectory Generation
Alexander Dittrich, Dieter Büchler
Robotic IntelligenceReinforcement Learning
🎯 What it does: Developed a three-wheel open-hardware, open-source table tennis ball launcher capable of fully controlling ball speed, spin, launch direction, and timing through Ethernet or Wi-Fi interfaces, achieving repeatable, controllable diverse ball trajectories.
AirTrack: Onboard Deep Learning Framework for Long-Range Aircraft Detection and Tracking
Sourish Ghosh, S. Scherer
Object DetectionObject TrackingVideo
🎯 What it does: Developed the AirTrack framework, which implements a visual system capable of real-time long-distance aircraft detection and tracking on lightweight drones.
ALAN: Autonomously Exploring Robotic Agents in the Real World
R. Mendonca, Deepak Pathak
Robotic IntelligenceReinforcement LearningAgentic AI
🎯 What it does: Propose ALAN, a robot agent capable of autonomously exploring and completing tasks in the real world with minimal training and interaction time; provides environment-centric exploration signals by measuring environmental changes (capturing object motion while ignoring the robot's own position changes), and agent-centric exploration signals by maximizing uncertainty in predicted environmental changes.
Aligning Human Preferences with Baseline Objectives in Reinforcement Learning
Daniel Marta, Iolanda Leite
Reinforcement Learning from Human FeedbackReinforcement Learning
🎯 What it does: Propose a method that combines a baseline reward function with subjective human preferences using a multi-objective setup, step-by-step training of the optimal strategy, and constructing a reward estimator through querying trajectory pairs to generate Pareto optimal policies incorporating human preferences.
Allowing Safe Contact in Robotic Goal-Reaching: Planning and Tracking in Operational and Null Spaces
Xinghao Zhu, M. Tomizuka
OptimizationRobotic Intelligence
🎯 What it does: This study explores the benefits of allowing safe contact in robot target reaching tasks, generates and tracks compliance reference signals in both operational and null spaces, proposes a hybrid solver combining sampling and gradient methods, and evaluates them in five different collision conditions in simulation and real environments.
AMSwarm: An Alternating Minimization Approach for Safe Motion Planning of Quadrotor Swarms in Cluttered Environments
V. K. Adajania, Angela P. Schoellig
OptimizationRobotic Intelligence
🎯 What it does: Proposes a scalable online algorithm for generating safe and kinematically feasible trajectories for quadrotor swarms in complex environments.
An Active Learning Based Robot Kinematic Calibration Framework Using Gaussian Processes
Ersin Daş, J. Burdick
OptimizationRobotic Intelligence
🎯 What it does: Propose a robot motion calibration framework based on Gaussian processes, which models the residual kinematic errors of the robotic arm using Gaussian processes and adaptively selects calibration measurement points via the Gaussian Process Upper Confidence Bound (GP-UCB) algorithm, thereby reducing the number of experiments and recalibration time.
An Analysis of Unified Manipulation with Robot Arms and Dexterous Hands via Optimization-based Motion Synthesis
Vatsal V. Patel, A. Dollar
OptimizationRobotic Intelligence
🎯 What it does: Evaluated and implemented unified control of a robotic arm and dexterous hand, synthesizing the configuration state sequence of the entire control system through a motion optimization framework.
An Architecture for Reactive Mobile Manipulation On-The-Move
Ben Burgess-Limerick, Peter Corke
Robotic Intelligence
🎯 What it does: Proposed a general architecture for reactive mobile manipulation during robot base movement, with implementation examples on real robots and performance evaluation.
An equivalent two section method for calculating the workspace of multi-segment continuum robots
Yeman Fan, Dikai Liu
Computational EfficiencyRobotic Intelligence
🎯 What it does: Proposed an Equivalent Two-Link (ETS) method for calculating the workspace of multi-segment continuous robots.
An Open Approach to Energy-Efficient Autonomous Mobile Robots
Liangkai Liu, Weisong Shi
OptimizationComputational EfficiencyRobotic Intelligence
🎯 What it does: Proposed a novel real-time energy prediction model and three path planning models for mobile robot obstacle avoidance.
An Optimal Open-Loop Strategy for Handling a Flexible Beam with a Robot Manipulator
Shamil Mamedov, J. Swevers
OptimizationRobotic Intelligence
🎯 What it does: A model-based open-loop optimal control strategy was developed to suppress vibrations during robot grasping of a flexible beam without requiring external sensors.
An Optimized Portable Cable-Driven Haptic Robot Enables Free Motion and Hard Contact
Changqi Zhang, Mingming Zhang
OptimizationRobotic Intelligence
🎯 What it does: Optimized design of a single-degree-of-freedom, portable cable-driven tactile robot, achieving a prototype with a large workspace and high force feedback.
An Underwater Jet-Propulsion Soft Robot with High Flexibility Driven by Water Hydraulics
Siqing Chen, Ben Lu
Robotic IntelligencePhysics Related
🎯 What it does: Designed and experimentally validated an underwater jet propulsion unit composed of 80% soft material, which can absorb and eject fluid by changing the cavity volume through pressure adjustment. Three such units were combined with buoyancy elements to form a tetrahedral soft robot (JSR), and experimental analysis was conducted on its thrust output, deformation, jet flow rate, and pressure response characteristics.
Analysing the Safety and Security of a UV-C Disinfection Robot
Desiana Nurchalifah, N. Hochgeschwender
Robotic IntelligenceGraph
🎯 What it does: For a UV-C disinfection robot performing ultraviolet disinfection tasks in a home environment, the authors adopt the STPA-Safesec system-level safety and security collaborative analysis method, modeling safety and security information as a knowledge graph, and achieving automatic retrieval of hazardous scenarios, analysis of gaps, and understanding of overall risks by querying the graph.
Analytical Approach to Inverse Kinematics of Single Section Mobile Continuum Manipulators
Audrey Hyacinthe Bouyom Boutchouang, R. Merzouki
Robotic Intelligence
🎯 What it does: Propose an analytical solution for inverse kinematics of a single-section mobile continuous manipulator, using this solution to determine the pose parameters of the mobile platform and the single-section continuous manipulator to achieve a given end-effector posture.
Analyzing Infrastructure LiDAR Placement with Realistic LiDAR Simulation Library
Xinyu Cai, Yikang Li
Data SynthesisAutonomous DrivingOptimizationPoint Cloud
🎯 What it does: This paper studies the infrastructure LiDAR localization problem, proposing an efficient process to find optimal installation positions in real simulation environments, and building a real LiDAR simulation library capable of mimicking the characteristics of various mainstream LiDARs. It uses generated high-fidelity point cloud data and multiple detection models to assess the perception accuracy of different localization schemes, further analyzing the correlation between perception performance and point cloud density and uniformity within regions of interest.
Anchoring Sagittal Plane Templates in a Spatial Quadruped
T. Greco, D. Koditschek
Robotic Intelligence
🎯 What it does: A new controller is proposed to enable spatial quadruped robots to perform stable motion around a sagittal plane template, achieving high dynamic gaits and translational movements.
Annotating Covert Hazardous Driving Scenarios Online: Utilizing Drivers' Electroencephalography (EEG) Signals
Chen Zheng, Jiangtao Gong
Autonomous DrivingTime SeriesBiomedical Data
🎯 What it does: Utilize driver EEG signals to perform online annotation of hidden and explicit dangers in driving scenarios, and conduct EEG experiments to compare differences between EEG and explicit danger reports.
Anomaly Detection For Robust Autonomous Navigation
Kefan Jin, Zhe Liu
Anomaly DetectionAutonomous DrivingImageVideo
🎯 What it does: Developed a semi-supervised anomaly detection module and proposed an end-to-end robust autonomous driving framework for detecting damaged data and extracting traffic scene features.
ANSEL Photobot: A Robot Event Photographer with Semantic Intelligence
D. Rivkin, F. Hogan
Robotic IntelligenceTransformerLarge Language ModelVision Language ModelVideoText
🎯 What it does: Generate a list of natural language photo descriptions for events using a large language model, and use a vision-language model to match the best photos from the robot's video stream, creating a photo archive.
Anthropomorphic robot hand using the principle of sweat and fingerprints of human hands
Donghyun Kim, Dongwon Yun
Robotic Intelligence
🎯 What it does: Studied the effects of fingerprints and sweat on hand friction, manufactured a human-like robotic hand with fingerprint structures, and conducted object grasping and friction change experiments by varying sweat amounts. For the first time, proposed and implemented a variable friction system based on fluid and microstructures, verifying its effectiveness in active friction control and grasping tests.
Anticipation and Delayed Estimation of Sagittal Plane Human Hip Moments using Deep Learning and a Robotic Hip Exoskeleton
Dean D. Molinaro, Aaron J. Young
Robotic IntelligenceConvolutional Neural NetworkTime Series
🎯 What it does: Training a TCN model using wearable sensor data to estimate the instantaneous torque of the human hip joint in the sagittal plane
Anticipatory Planning: Improving Long-Lived Planning by Estimating Expected Cost of Future Tasks
Roshan Dhakal, Gregory J. Stein
OptimizationRobotic IntelligenceGraph Neural NetworkWorld ModelBenchmark
🎯 What it does: Propose a prospective planning method that guides service robots' behavior in long-term planning scenarios by estimating the expected costs of future tasks, thereby reducing the overall planning cost.
Approximating Discontinuous Nash Equilibrial Values of Two-Player General-Sum Differential Games
Lei Zhang, Yi Ren
Autonomous DrivingOptimizationSupervised Fine-TuningReinforcement LearningStochastic Differential Equation
🎯 What it does: Two methods—hybrid method (combining supervised Nash equilibrium with HJI PDE) and value hardening method (progressively hardening reward-based HJI solution)—are studied to approximate non-continuous Nash equilibrium values in two-player general sum differential games, with their generalization and safety performance compared in vehicle interaction simulations.
Approximation Algorithms for Robot Tours in Random Fields with Guaranteed Estimation Accuracy
Shamak Dutta, Stephen L. Smith
OptimizationRobotic Intelligence
🎯 What it does: Studied the robot sample placement and shortest cruise problems for mapping stable random fields, providing an approximate algorithm in convex environments, improving previous theoretical and experimental results, and refuting previous claims about the lower bounds of sample placement.
Aquarium: A Fully Differentiable Fluid-Structure Interaction Solver for Robotics Applications
Jeonghoon Lee, Zachary Manchester
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: Proposes Aquarium, a fully differentiable 2D fluid-structure interaction solver capable of stable simulations, precise coupling of fluid and robot physics, and full differentiation of fluid and robot states and parameters;
Are All Point Clouds Suitable for Completion? Weakly Supervised Quality Evaluation Network for Point Cloud Completion
Jieqi Shi, S. Shen
Autonomous DrivingPoint Cloud
🎯 What it does: This paper proposes a weakly supervised quality assessment network to score the quality of point clouds before point cloud completion.
ARiADNE: A Reinforcement learning approach using Attention-based Deep Networks for Exploration
Yuhong Cao, Guillaume Sartoretti
Robotic IntelligenceTransformerReinforcement LearningImage
🎯 What it does: Proposes ARiADNE, an attention-based deep reinforcement learning method for real-time, non-greedy autonomous exploration path planning.
ARMBench: An Object-centric Benchmark Dataset for Robotic Manipulation
Chaitanya Mitash, M. Nambi
Robotic IntelligenceImageBenchmark
🎯 What it does: Proposed a large-scale, object-oriented benchmark dataset for warehouse robot operations called ARMBench, and defined related benchmark tasks;
Asking for Help: Failure Prediction in Behavioral Cloning through Value Approximation
Cem Gokmen, Mohi Khansari
Anomaly DetectionRobotic IntelligenceSequential
🎯 What it does: Proposed and implemented BCVA to predict failure in behavioral cloning, applied to the door unlocking task.
Asynchronous State Estimation of Simultaneous Ego-motion Estimation and Multiple Object Tracking for LiDAR-Inertial Odometry
Yu-Kai Lin, C. Wang
Object TrackingPose EstimationAutonomous DrivingOptimizationSimultaneous Localization and MappingPoint Cloud
🎯 What it does: Proposes an optimization-based LiDAR-Inertial Odometry method called LIO-SEGMOT, which can simultaneously estimate the vehicle's motion and track multiple targets in dynamic environments, while achieving asynchronous state updates.
ASystem for Generalized 3D Multi-Object Search
Kaiyu Zheng, Stefanie Tellex
Object DetectionRetrievalRobotic IntelligencePoint Cloud
🎯 What it does: Proposed and implemented the GenMOS system, achieving general-purpose 3D multi-object search that can perform independent search across different robots and environments;
Atomic-level Tracking and Analyzing of Quantum-dot Motion Steered by an Electrostatic Field Positioned by a Nanorobotic Manipulation Tip
Zhi Qu, Lixin Dong
Object TrackingImagePhysics Related
🎯 What it does: Studied the use of deep learning techniques to track the motion of quantum dots driven by the electric field at the probe tip at the single-atom level.
ATTACH Dataset: Annotated Two-Handed Assembly Actions for Human Action Understanding
Dustin Aganian, H. Groß
ClassificationRecognitionPose EstimationVideoMultimodalityBenchmark
🎯 What it does: Designed and released the ATTACH dataset, containing 51.6 hours of bimanual assembly actions, using three cameras to capture color and depth images, along with 3D skeletons estimated by the Azure Kinect Body Tracking SDK, with fine-grained actions annotated separately for each hand; subsequently conducted benchmark experiments on existing action recognition and detection methods using this dataset.
Augmented Reality-Assisted Robot Learning Framework for Minimally Invasive Surgery Task
Junling Fu, E. Momi
Robotic IntelligenceBiomedical Data
🎯 What it does: Propose an augmented reality-assisted robotic learning framework for minimally invasive surgery tasks.
Auto-Assembly: a framework for automated robotic assembly directly from CAD
Sergei Zobov, Komal Vendidandi
Robotic Intelligence
🎯 What it does: Proposed and implemented the Auto-Assembly framework, achieving direct robot assembly from CAD design files, and completed rapid assembly of modular components on an assembly unit composed of two robots, demonstrating flexibility for different input designs.
AutoBag: Learning to Open Plastic Bags and Insert Objects
L. Chen, Ken Goldberg
Robotic Intelligence
🎯 What it does: Propose a self-supervised learning framework that uses a dual-arm robot to gradually open bags and insert objects by identifying plastic bag handles and edges without using UV markers or lighting.
AutoCharge: Autonomous Charging for Perpetual Quadrotor Missions
Alessandro Saviolo, Giuseppe Loianno
Robotic Intelligence
🎯 What it does: Proposed an autonomous drone charging system called AutoCharge, combining a portable ground station and lightweight charging cables to achieve continuous endurance.
Automated Action Evaluation for Robotic Imitation Learning via Siamese Neural Networks
Xiang Chang, Qiang Shen
Robotic IntelligenceReinforcement LearningVideo
🎯 What it does: Proposed a few-shot robot imitation learning algorithm based on a third-person perspective, and developed a Siamese neural network evaluation system called BODA to automatically assess multi-stage task completion, replacing sparse human rewards.
Automatic Cell Rotation Method Based on Deep Reinforcement Learning
Huiying Gong, Mingzhu Sun
Robotic IntelligenceReinforcement LearningBiomedical Data
🎯 What it does: This paper first applies deep reinforcement learning to the cell rotation task, constructs a cell rotation simulation environment, designs a reward function, and trains a deep Q-learning (DQL) agent to achieve microcatheter trajectory planning, completing cell rotation while significantly reducing mechanical damage.
Automatic Generation of Robot Facial Expressions with Preferences
Bing Tang, Feng Wu
GenerationRobotic Intelligence
🎯 What it does: Designed a physical robot with a human-like appearance and developed an automatically generated facial expression framework based on the MAP-Elites algorithm
Automating Vascular Shunt Insertion with the dVRK Surgical Robot
K. Dharmarajan, Ken Goldberg
Robotic IntelligenceImageBiomedical Data
🎯 What it does: Proposed an automated pipeline for vascular catheter insertion using the da Vinci Research Kit (dVRK).
Autonomous Control for Orographic Soaring of Fixed-Wing UAVs
Tom Suys, B. Remes
Robotic Intelligence
🎯 What it does: Propose a novel controller that enables fixed-wing UAVs to autonomously soar in mountainous wind fields and extend their endurance.
Autonomous Drifting with 3 Minutes of Data via Learned Tire Models
Franck Djeumou, Avinash Balachandran
Autonomous DrivingOrdinary Differential Equation
🎯 What it does: Proposed a tire force model based on neural ordinary differential equations and neural-ExpTanh parameterization, serving as an alternative to the analytical brush tire model in existing nonlinear model predictive control frameworks, achieving high-performance autonomous drifting on various trajectories;
Autonomous Drone Racing: Time-Optimal Spatial Iterative Learning Control within a Virtual Tube
Shuli Lv, Q. Quan
Autonomous DrivingOptimization
🎯 What it does: Proposed an efficient learning method that mimics the training experience of top drivers, enabling online trajectory learning to complete drone racing as quickly as possible.
Autonomous Endoscope Control Algorithm with Visibility and Joint Limits Avoidance Constraints for da Vinci Research Kit Robot
R. Moccia, F. Ficuciello
OptimizationRobotic Intelligence
🎯 What it does: Proposed an autonomous endoscope control method that enables the camera to track surgical instruments on the patient-side manipulator.
Autonomous Intelligent Navigation for Flexible Endoscopy Using Monocular Depth Guidance and 3-D Shape Planning
Yiang Lu, Yunhui Liu
Depth EstimationRobotic IntelligenceTransformerBiomedical Data
🎯 What it does: Propose a data-driven, vision-shape fusion framework without prior system models and global environmental knowledge for autonomous intelligent navigation of flexible endoscopes
Autonomous Needle Navigation in Retinal Microsurgery: Evaluation in ex vivo Porcine Eyes
Peiyao Zhang, Marin Kobilarov
Robotic IntelligenceBiomedical Data
🎯 What it does: Developed an autonomous needle path navigation strategy for retinal microsurgery, aiming to achieve precise control, reduce surgery time, and enhance safety.
Autonomous Task Planning for Heterogeneous Multi-Agent Systems
Anatoli A. Tziola, S. Loizou
Optimization
🎯 What it does: Proposed a multi-agent system automatic task planning method based on non-deterministic finite automata (NFA) and ε-transitions.
Autonomous Underwater Docking using Flow State Estimation and Model Predictive Control
Rakesh Vivekanandan, Geoffrey A. Hollinger
OptimizationRobotic Intelligence
🎯 What it does: Proposed a navigation framework that combines flow state estimation with model predictive control (MPC) to enable autonomous underwater drones to dock with wave energy converters under varying sea conditions.
Autotuning Symbolic Optimization Fabrics for Trajectory Generation
Max Spahn, Javier Alonso-Mora
OptimizationHyperparameter SearchRobotic Intelligence
🎯 What it does: Propose a trajectory generation method using Bayesian optimization for automatic parameter tuning, and apply it to non-Riemannian geometry-based optimal fabric.
Avatarm: an Avatar With Manipulation Capabilities for the Physical Metaverse
A. Villani, D. Prattichizzo
Robotic IntelligenceVision-Language-Action Model
🎯 What it does: Proposed a new interface called 'Avatarm,' which integrates hidden robotic arms into virtual avatars, enabling users to manipulate physical objects in the metaverse and addressing the limitations of digital-physical interaction;
AvoidBench: A high-fidelity vision-based obstacle avoidance benchmarking suite for multi-rotors
Hang Yu, C. de Wagter
Autonomous DrivingRobotic IntelligenceBenchmark
🎯 What it does: Proposes AvoidBench, a high-fidelity benchmark suite for evaluating multirotor visual obstacle avoidance algorithms, integrating RotorS dynamics with Unity3D virtual environments to provide performance and environmental metrics.
AZTR: Aerial Video Action Recognition with Auto Zoom and Temporal Reasoning
Xijun Wang, Dinesh Manocha
RecognitionVideo
🎯 What it does: Propose a method for action recognition in aerial videos captured by drones, combining adaptive scaling and spatiotemporal reasoning
Balancing Efficiency and Unpredictability in Multi-robot Patrolling: A MARL-Based Approach
Lingxiao Guo, Jianping He
Robotic IntelligenceGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: Propose a multi-robot patrolling scheme based on graph deep learning and collaborative multi-agent reinforcement learning, using an autoregressive mechanism to output actions.
BAMF-SLAM: Bundle Adjusted Multi-Fisheye Visual-Inertial SLAM Using Recurrent Field Transforms
Wei Zhang, N. Haala
Pose EstimationDepth EstimationAutonomous DrivingOptimizationSimultaneous Localization and MappingImageMultimodality
🎯 What it does: Proposed and implemented a visual-inertial SLAM system called BAMF-SLAM based on multi-fisheye cameras and inertial measurements, achieving accurate and robust state estimation through bundle adjustment and recursive field transformation.
Bayesian deep learning for affordance segmentation in images
Lorenzo Mur-Labadia, J. J. Guerrero
RecognitionObject DetectionSegmentationConvolutional Neural NetworkImage
🎯 What it does: Proposed a Bayesian deep network-based image operability segmentation method that can quantify randomness and uncertainty at the spatial level.
Bayesian inference of fog visibility from LiDAR point clouds and correlation with probabilities of detection
Karl Montalban, Simon Lacroix
Object DetectionAutonomous DrivingExplainability and InterpretabilityPoint Cloud
🎯 What it does: Infer optical visibility from LiDAR point clouds using Bayesian inference and Markov Chain Monte Carlo (MCMC) sampling, model visibility estimation as a classification problem based on distance distribution, and evaluate the impact of visibility on standard object detection probabilities.
Benchmarking Potential Based Rewards for Learning Humanoid Locomotion
Seungmin Jeon, Sangbae Kim
Robotic IntelligenceReinforcement LearningBenchmark
🎯 What it does: Benchmarking standard reward shaping and potential-based reward shaping (PBRS) in humanoid robot walking tasks
Benchmarking Reinforcement Learning Techniques for Autonomous Navigation
Zifan Xu, P. Stone
Autonomous DrivingReinforcement LearningBenchmark
🎯 What it does: This paper proposes and implements an open-source large-scale navigation benchmark, using this benchmark to systematically evaluate four categories of deep reinforcement learning techniques (memory-based neural network architectures, safety-guaranteed reinforcement learning, model-based reinforcement learning, and domain randomization) in real environments, to explore their performance on four key requirements in autonomous navigation (uncertainty reasoning, safety, limited trial-and-error learning, and generalization ability).
BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation
Zhijian Liu, Song Han
Object DetectionSegmentationAutonomous DrivingOptimizationImageMultimodalityPoint Cloud
🎯 What it does: Propose the BEVFusion framework, unifying multi-modal features into a shared bird's-eye view (BEV) to achieve multi-task multi-sensor fusion.
Bi-Manual Manipulation of Multi-Component Garments towards Robot-Assisted Dressing
Stelios Kotsovolis, Y. Demiris
Robotic IntelligenceMultimodality
🎯 What it does: Proposes a robot-assisted dressing strategy for multi-component clothing, including a decision-making process based on hierarchical task representation and environmental conditions in a glove dressing scenario, as well as a dual-arm control strategy combining mixed position, visual, and force feedback.
Bidirectional Generalised Rigid Point Set Registration
Ang Zhang, M. Q. Meng
Pose EstimationOptimizationPoint Cloud
🎯 What it does: Propose Bi-GRPSR and achieve bidirectional rigid point set registration through the Bi-AGCPD algorithm.
Big data approach for synthesizing a spatial linkage mechanism
N. Yim, Yoon Young Kim
Data SynthesisOptimization
🎯 What it does: Propose a two-step big data method to synthesize spatial linkage mechanisms, first determining the topology and then using gradient optimization to determine the dimensions.
Bilateral asymmetric hip stiffness applied by a robotic hip exoskeleton elicits kinematic and kinetic adaptation
Banu Abdikadirova, M. Huber
Robotic IntelligenceBiomedical Data
🎯 What it does: By applying asymmetric stiffness perturbations to both hips during walking, the study investigates human adaptation to such perturbations and records changes in step length and ground reaction forces.
Bimanual Rope Manipulation Skill Synthesis through Context Dependent Correction Policy Learning from Human Demonstration
T. Akbulut, E. Oztop
Robotic IntelligenceReinforcement Learning from Human Feedback
🎯 What it does: Learned explicit correction strategies when expected state transitions between motion primitives were not achieved, and used conditional neural motion primitives (CNMPs) to generate context-based correction trajectories.
Biodegradable Origami Gripper Actuated with Gelatin Hydrogel for Aerial Sensor Attachment to Tree Branches
Christian Geckeler, S. Mintchev
Robotic IntelligenceAgriculture Related
🎯 What it does: Designed and verified a biodegradable origami gripper for attaching sensors to branches via aerial robots, which self-degrades upon exposure to moisture.
Bioinspired tearing manipulation with a robotic fish
Stanley J. Wang, Hannah S. Stuart
Robotic Intelligence
🎯 What it does: Designed and tested a fish-like robot named SunBot, which utilizes tail-driven 'head shaking' motion to perform tearing operations on fixed prey. The mechanical performance was measured in the laboratory under different tail speeds and motion ranges; a simplified dynamic model was established to explain the experimental results, and the tearing function was demonstrated in free swimming experiments.
Bipedal Robot Walking Control Using Human Whole-Body Dynamic Telelocomotion
Guillermo Colin, João Ramos
Robotic Intelligence
🎯 What it does: Proposed a dynamic remote walking control framework that achieves gait synchronization between a biped robot and a human operator by generating a virtual human walking model and applying forces to the human-robot system.
BITS: Bi-level Imitation for Traffic Simulation
Danfei Xu, M. Pavone
Data SynthesisAutonomous DrivingOptimizationSequential
🎯 What it does: A data-driven approach is used to generate traffic behaviors from real driving logs, employing a two-layer hierarchical structure: high-level intent inference and low-level driving behavior imitation, combined with a planning module to achieve stable long-term behaviors; simultaneously providing a unified data format tool and an interactive simulation environment; and proposing a behavior realism evaluation metric.
BO-ICP: Initialization of Iterative Closest Point Based on Bayesian Optimization
Harel Biggie, C. Heckman
OptimizationHyperparameter SearchPoint Cloud
🎯 What it does: Propose a Bayesian optimization-based iterative closest point (ICP) initialization method to find the critical initial transformation
BogieCopter: A Multi-Modal Aerial-Ground Vehicle for Long-Endurance Inspection Applications
T. Dias, M. Basiri
Robotic Intelligence
🎯 What it does: Designed and developed a novel hybrid aerial-ground vehicle with multimodal mobility capabilities, capable of flying, landing, and moving on air, flat surfaces, and inclined planes, with experimental evaluation of its flight, ground movement, wall-climbing capabilities, and energy consumption.
Boosting 3D Point Cloud Registration by Transferring Multi-modality Knowledge
Mingzhi Yuan, Manning Wang
Pose EstimationAutonomous DrivingKnowledge DistillationConvolutional Neural NetworkSupervised Fine-TuningMultimodalityPoint Cloud
🎯 What it does: Improved the accuracy of point cloud registration by transferring knowledge from pre-trained multi-modal models to a new point cloud descriptor neural network, using only single-modal point cloud data during inference.
Boosting Performance of a Baseline Visual Place Recognition Technique by Predicting the Maximally Complementary Technique
Connor Malone, Michael Milford
RetrievalImage
🎯 What it does: Propose a visual place recognition technique that predicts the most complementary methods and integrates them with existing baseline techniques to enhance overall performance.
Bootstrapping the Dynamic Gait Controller of the Soft Robot Arm
Rudolf J. Szadkowski, J. Faigl
Robotic Intelligence
🎯 What it does: Proposed a dynamic gait controller for soft robotic manipulators to improve performance in repetitive tasks.
Boundary Conditions in Geodesic Motion Planning for Manipulators
Mario Laux, A. Zell
OptimizationRobotic IntelligencePhysics Related
🎯 What it does: This paper proposes a method for introducing arbitrary velocity and acceleration boundary conditions into geometric trajectory planning, utilizing generalized coordinates to achieve optimal transition between boundary conditions and free motion, and can be combined with the time-scale algorithm to further enhance trajectory quality.