arXivSub Start free trial

ICRA 2024 Papers — Page 11

IEEE International Conference on Robotics and Automation · 1760 papers

MMAUD: A Comprehensive Multi-Modal Anti-UAV Dataset for Modern Miniature Drone Threats

Shenghai Yuan, Lihua Xie

ClassificationObject DetectionObject TrackingImageMultimodalityPoint CloudBenchmarkAudio

🎯 What it does: Proposed a comprehensive multi-modal anti-UAV dataset named MMAUD, covering stereo vision, LiDAR, radar, and acoustic arrays, and provides precise Leica-generated ground truth for UAV detection, type classification, and trajectory estimation.

MMPI: a Flexible Radiance Field Representation by Multiple Multi-plane Images Blending

Yuze He, Wenping Wang

GenerationData SynthesisAutonomous DrivingNeural Radiance FieldImagePoint CloudBenchmark

🎯 What it does: Propose a flexible light field representation based on multi-directional multi-plane images (MPI) with adaptive mixing, enabling high-quality view synthesis for complex scenes.

MOAR Planner: Multi-Objective and Adaptive Risk-Aware Path Planning for Infrastructure Inspection with a UAV

Louis Petit, Alexis Lussier Desbiens

Optimization

🎯 What it does: Propose the MOAR path planner to achieve real-time adaptation to dynamic risks while optimizing safety, time, and energy consumption; search for optimal paths in graphs using a risk-aware cost function.

Mobile Bot Rotation Using Sound Source Localization And Distant Speech Recognition

Swapnil Sontakke, D. K T

RecognitionAutonomous DrivingRobotic IntelligenceAudio

🎯 What it does: Proposed a rotation module for mobile robots based on sound source localization and far-field speech recognition, enabling the robot to automatically turn according to distant voice commands.

Mobile Robot Oriented Large-Scale Indoor Dataset for Dynamic Scene Understanding

Yifan Tang, Long Zeng

Robotic IntelligenceImageVideoTabularBenchmark

🎯 What it does: Proposed and constructed a large-scale indoor dynamic scene dataset THUD for mobile robots, and evaluated tasks such as 3D object detection, semantic segmentation, and robot localization on it.

Model Design and Concept of Operations of Standard Interface for On-orbit Construction

Jingdong Zhao, Hong Liu

Robotic Intelligence

🎯 What it does: A complete on-orbit construction solution in space is proposed, including novel standard interfaces, compliant docking methods, and hierarchical path planning for multi-arm space robots.

Model Predictive Control for an Autonomous Underwater Robot with Fully Vectored Propulsion

Tianzhu Gao, Yantao Shen

OptimizationRobotic IntelligencePhysics RelatedOrdinary Differential Equation

🎯 What it does: Designed a fully omnidirectional vector propulsion underwater robot equipped with eight vector thrusters, built a software architecture based on ROS, established a 13-dimensional state-space marine dynamics model using Fossen's method, discretized it with the explicit fourth-order Runge-Kutta method, applied model predictive control (MPC) with physical constraints to achieve real-time prediction and optimization, and finally validated control performance through numerical simulations and point-to-point motion experiments.

Model Predictive Control with Graph Dynamics for Garment Opening Insertion during Robot-Assisted Dressing

Stelios Kotsovolis, Y. Demiris

Robotic IntelligenceGraph Neural NetworkGraph

🎯 What it does: Proposed a two-handed control method for the garment opening insertion task during robot-assisted dressing, trained in a simulation environment, and validated on a real medical training model.

Model-Based Approach for Lateral Maneuvers of Bird-Size Ornithopter

E. Sanchez-Laulhe, A. Ollero

Robotic IntelligencePhysics Related

🎯 What it does: Propose a model-based lateral maneuvering method for bird-sized winged aircraft, define a simplified steady-state turning equation, and achieve continuous circular trajectory tracking control.

Model-Based Runtime Monitoring with Interactive Imitation Learning

Huihan Liu, Yuke Zhu

ClassificationAnomaly DetectionWorld Model

🎯 What it does: Proposed a model-based runtime monitoring algorithm that can learn from deployment data to detect system anomalies and predict failures.

Model-Free Control of a Class of High-Precision Scanning Motion Systems with Piezoceramic Actuators

Y. Al-Rawashdeh, M. Janaideh

Physics Related

🎯 What it does: Proposed and experimentally verified a model-free control and synchronization technique for piezoelectric-driven micro-adjustment phase, aimed at achieving coordinated motion between coarse long-travel axes and fine short-travel axes;

Modeling and Analysis of Combined Rimless Wheel with Tensegrity Spine

Yuxuan Xiang, Fumihiko Asano

Robotic IntelligencePhysics Related

🎯 What it does: Proposed a quadruped model composed of a tension-integrated structural spine and edgeless wheels, and investigated its locomotion performance through passive dynamic gait simulation.

Modeling and Control of Intrinsically Elasticity Coupled Soft-Rigid Robots

Zachary Patterson, Daniela Rus

Robotic Intelligence

🎯 What it does: Studies the modeling and control of soft-rigid hybrid robots under intrinsic elastic coupling, proposing an elastic compensation controller and verifying its stability.

Modeling and Control of PADUAV: a Passively Articulated Dual UAVs Platform for Aerial Manipulation*

Jiali Sun, Yiqun Dong

OptimizationRobotic Intelligence

🎯 What it does: Proposed PADUAV—a 5-degree-of-freedom aerial platform based on two off-the-shelf quadrotors passively coupled on a rigid frame, and designed a 5-DOF geometric tracking control strategy that directly generates 2D force and 3D torque, followed by verification of the platform's feasibility through Gazebo+RotorS simulation.

MoDem-V2: Visuo-Motor World Models for Real-World Robot Manipulation

Patrick E. Lancaster, Vikash Kumar

Robotic IntelligenceReinforcement LearningWorld Model

🎯 What it does: Developed the MoDem-V2 system, which can directly learn contact-rich robotic manipulation in real-world environments without installed sensors.

Modular Growing Mechanism with Multi-axis Deformation

Dongdong Du, Barbara Mazzolai

Robotic Intelligence

🎯 What it does: Proposed and implemented a modular robot concept that can grow by adding transformable modules at the tip, and validated its feasibility through a two-module system;

Monocular Localization with Semantics Map for Autonomous Vehicles

Jixiang Wan, Ming Yang

Autonomous DrivingSimultaneous Localization and MappingImagePoint Cloud

🎯 What it does: A lightweight visual semantic localization algorithm is proposed, which uses stable semantic features instead of low-level texture features. In the offline phase, a semantic map is constructed by detecting semantic objects such as ground markers, lane lines, and poles using a camera or LiDAR. In the online phase, localization is achieved through data association between semantic features and map objects.

MonoOcc: Digging into Monocular Semantic Occupancy Prediction

Yupeng Zheng, Qichao Zhang

Autonomous DrivingKnowledge DistillationConvolutional Neural NetworkTransformerImageBenchmark

🎯 What it does: We propose MonoOcc, an improved monocular semantic occupancy prediction framework that enhances performance using auxiliary semantic loss, image-conditioned cross-attention modules, and knowledge distillation.

MoPA: Multi-Modal Prior Aided Domain Adaptation for 3D Semantic Segmentation

Haozhi Cao, Lihua Xie

SegmentationDomain AdaptationVision Language ModelPoint Cloud

🎯 What it does: Proposes a multi-modal prior-assisted domain adaptation method called MoPA to improve the recognition of rare objects in 3D semantic segmentation

MORALS: Analysis of High-Dimensional Robot Controllers via Topological Tools in a Latent Space

Ewerton R. Vieira, Kostas E. Bekris

Robotic IntelligenceAuto Encoder

🎯 What it does: Propose the MORALS method, combining autoencoder neural networks and Morse Graphs, to estimate the region of attraction (RoA) of high-dimensional robot controllers in a learned latent space.

MoRC—A Modular Robot Controller

Carsten Oldemeyer, Tobias Bellmann

Robotic Intelligence

🎯 What it does: Proposed and implemented MoRC—a modular robot controller based on the FMI standard, capable of controlling any industrial robot with electrical drives through customizable vendor-independent control cabinets and multi-rate real-time control components.

MORPH: Design Co-optimization with Reinforcement Learning via a Differentiable Hardware Model Proxy

Zhanpeng He, M. Ciocarlie

OptimizationRobotic IntelligenceReinforcement LearningWorld Model

🎯 What it does: Propose the MORPH method, which co-optimizes hardware design parameters and control strategies in a simulated environment through reinforcement learning;

Morphable-SfS: Enhancing Shape-from-Silhouette Via Morphable Modeling

Guoyu Lu

GenerationMesh

🎯 What it does: Proposed a Shape-from-Silhouette method combining Morphable Model with multi-view synthesis to reconstruct complete 3D object models from a single image.

MORPHeus: a Multimodal One-armed Robot-assisted Peeling System with Human Users In-the-loop

Ruolin Ye (Cornell University), T. Bhattacharjee (Cornell University)

Robotic IntelligenceReinforcement Learning from Human FeedbackMultimodality

🎯 What it does: Proposed and demonstrated a single-arm robot-assisted peeling system, incorporating a multi-modal active perception module, a human-in-the-loop long-term planner, and a compliant controller.

MOSAIC: Learning Unified Multi-Sensory Object Property Representations for Robot Learning via Interactive Perception

Gyan Tatiya, Jivko Sinapov

Knowledge DistillationRepresentation LearningRobotic IntelligenceTransformerMultimodality

🎯 What it does: A unified object attribute representation learning framework called MOSAIC based on multi-modal perception is studied, which aligns the knowledge of foundation models from visual, tactile, and auditory modalities, and experiments are conducted on robot interaction data to evaluate performance in object classification and grasping tasks.

Motion Memory: Leveraging Past Experiences to Accelerate Future Motion Planning

Dibyendu Das, Xuesu Xiao

Autonomous DrivingOptimizationComputational Efficiency

🎯 What it does: Propose the Motion Memory method, which utilizes past planning experiences to accelerate new planning.

Motion planning for 4WS vehicle with autonomous selection of steering modes via an MIQP-MPC controller

Ngoc Thinh Nguyen, Floris Ernst

Autonomous DrivingOptimizationAgriculture Related

🎯 What it does: Propose a model predictive controller (MPC) based on mixed-integer quadratic programming (MIQP) that can autonomously select and switch between parallel positive steering and symmetric negative steering modes for four-wheel steering vehicles to meet trajectory tracking and heading requirements in agricultural applications.

Motions in Microseconds via Vectorized Sampling-Based Planning

Wil B. Thomason, L. Kavraki

Computational EfficiencyRobotic Intelligence

🎯 What it does: Accelerate sampling-based motion planning through fine-grained parallelism and vectorized subroutines, reducing planning time from milliseconds to microseconds and increasing the solving rate to kilohertz.

MOTPose: Multi-object 6D Pose Estimation for Dynamic Video Sequences using Attention-based Temporal Fusion

Arul Selvam Periyasamy, Sven Behnke

Object DetectionPose EstimationTransformerVideo

🎯 What it does: Proposes the MOTPose method, which utilizes attention mechanisms for temporal fusion to achieve joint inference for multi-object 6D pose estimation and object detection;

MPCGPU: Real-Time Nonlinear Model Predictive Control through Preconditioned Conjugate Gradient on the GPU

Emre Adabag, B. Plancher

OptimizationRobotic Intelligence

🎯 What it does: Proposed a real-time nonlinear model predictive control (NMPC) solver called MPCGPU based on GPU acceleration, with the core utilizing a preconditioned conjugate gradient (PCG) solver;

MPS: A New Method for Selecting the Stable Closed-Loop Equilibrium Attitude-Error Quaternion of a UAV During Flight

Francisco M. F. R. Gonçalves, N. O. P'erez-Arancibia

OptimizationPhysics Related

🎯 What it does: Proposed and verified a model predictive selection (MPS) method for real-time selection of stable closed-loop balance attitude error quaternions during high-speed yaw flight.

MTG: Mapless Trajectory Generator with Traversability Coverage for Outdoor Navigation

Jing Liang, Dinesh Manocha

Autonomous DrivingOptimizationRobotic IntelligenceAuto Encoder

🎯 What it does: Propose a learning-based map-agnostic trajectory generation algorithm for outdoor robot navigation

MTRadSSD: A Multi-Task Single-Stage Detector for Object Detection and Free Space Analysis in Radar Point Clouds*

Yinbao Li, J. Jiao

Object DetectionAutonomous DrivingPoint Cloud

🎯 What it does: Proposes MTRadSSD, a multi-task single-stage detector based on radar point clouds, which utilizes an instance-aware sampling strategy for multi-class vehicle, pedestrian, and cyclist detection, and generates polygon-represented free space through a free space analysis in bird's-eye view (BEV) using a kernel density estimation (KDE)-based occupancy map tool.

Multi-agent Path Finding for Cooperative Autonomous Driving

Zhongxia Yan, Cathy Wu

Autonomous DrivingOptimization

🎯 What it does: Developed a multi-agent path planning algorithm applicable to signal-free intersections for cooperative autonomous driving.

Multi-Agent Strategy Explanations for Human-Robot Collaboration

Ravi Pandya, H. Admoni

Explainability and InterpretabilityRobotic IntelligenceLarge Language Model

🎯 What it does: Studying how to generate multi-agent strategy explanations for human-agent collaboration

Multi-Camera Asynchronous Ball Localization and Trajectory Prediction with Factor Graphs and Human Poses

Qingyu Xiao, Matthew C. Gombolay

Object TrackingPose EstimationOptimizationConvolutional Neural NetworkVideo

🎯 What it does: Achieve real-time asynchronous 3D tennis ball localization using multi-camera systems and factor graphs, and predict trajectories by estimating hidden states such as velocity and rotation; during the early flight phase, compute rotational priors by combining human pose information with temporal convolutional networks (TCN) to improve prediction accuracy.

Multi-class Road Defect Detection and Segmentation using Spatial and Channel-wise Attention for Autonomous Road Repairing

Jongmin Yu, Shan Luo

Object DetectionSegmentationConvolutional Neural NetworkImage

🎯 What it does: Proposed an end-to-end multi-class pavement defect detection and segmentation method.

Multi-Confidence Guided Source-Free Domain Adaption Method for Point Cloud Primitive Segmentation

Shaohu Wang, Zhengtao Zhang

SegmentationDomain AdaptationPoint Cloud

🎯 What it does: Proposed a source-free domain adaptation method based on multi-confidence for point cloud semantic segmentation using a pseudo-label self-training framework.

Multi-Granular Transformer for Motion Prediction with LiDAR

Yi Gan, Lingting Ge

Autonomous DrivingTransformerPoint CloudBenchmark

🎯 What it does: Proposed a Multi-Granular Transformer (MGTR) framework for motion prediction by encoding and decoding multi-granularity context features of different traffic participants, combined with LiDAR point cloud semantic features.

Multi-Level Action Tree Rollout (MLAT-R): Efficient and Accurate Online Multiagent Policy Improvement

Andrea Henshall, S. Karaman

Reinforcement Learning

🎯 What it does: Proposes the multi-layer action tree rollout (MLAT-R) algorithm for online policy improvement in multi-agent environments, evaluated on a challenging problem where a baseline policy fails to reach the goal.

Multi-Level Progressive Reinforcement Learning for Control Policy in Physical Simulations

Kefei Wu, Xiaopei Liu

Robotic IntelligenceReinforcement LearningPhysics Related

🎯 What it does: Propose a multi-level progressive reinforcement learning framework that utilizes coarse-to-fine multi-resolution physical simulation to accelerate control policy learning, with experimental validation on 2D aerial and 3D underwater robots.

Multi-level Reasoning for Robotic Assembly: From Sequence Inference to Contact Selection

Xinghao Zhu, Anoop Cherian

Robotic IntelligenceTransformerSequential

🎯 What it does: Proposed a multi-level robotic assembly planning framework, including assembly sequence reasoning, motion planning, and contact optimization, and used the Part Assembly Sequence Transformer (PAST) model to recursively infer assembly sequences from target blueprints.

Multi-LIO: A Lightweight Multiple LiDAR-Inertial Odometry System

Qi Chen, Jian Pu

Autonomous DrivingComputational EfficiencySimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposed a real-time, computationally efficient multi-LiDAR-inertial odometry system called Multi-LIO, and validated its performance through experiments.

Multi-modal 3D Human Tracking for Robots in Complex Environment with Siamese Point-Video Transformer

Shuo Xin, Yong Liu

Object TrackingTransformerVideoPoint Cloud

🎯 What it does: Proposed PVTrack, a multi-modal 3D human tracking model for robotic scenarios, combining point clouds and RGB videos to achieve information complementarity.

Multi-modal jumping and crawling in an autonomous, springtail-inspired microrobot

Shashwat Singh, Ryan St. Pierre

Robotic Intelligence

🎯 What it does: Designed and manufactured an autonomous, bio-inspired micro-robot inspired by spring-tailed arthropods, capable of both crawling and jumping.

Multi-Model 3D Registration: Finding Multiple Moving Objects in Cluttered Point Clouds

David Jin, Luca Carlone

Pose EstimationPoint Cloud

🎯 What it does: This paper proposes a multi-model 3D registration method based on EM that can simultaneously recover the motion of all objects between two frames in point clouds containing multiple objects and cluttered backgrounds.

Multi-Object RANSAC: Efficient Plane Clustering Method in a Clutter

Seunghyeon Lim, Byoung-Tak Zhang

Robotic IntelligenceImagePoint Cloud

🎯 What it does: Using an RGB-D camera for plane clustering in cluttered scenes, and validating the method's effectiveness through robot suction gripper experiments.

Multi-Object Tracking by Hierarchical Visual Representations

Jinkun Cao, Kris Kitani

Object Tracking

🎯 What it does: Proposes a new visual hierarchy representation paradigm and designs an attention mechanism to fuse these hierarchical visual representations for multi-object tracking.

Multi-objective Cross-task Learning via Goal-conditioned GPT-based Decision Transformers for Surgical Robot Task Automation

Jiawei Fu, Qi Dou

Robotic IntelligenceTransformerLarge Language ModelSequentialBiomedical Data

🎯 What it does: Proposes a GPT-based learning framework that utilizes a goal-conditioned decision transformer and cross-task pre-training to achieve task automation in surgical robots.

Multi-Profile Quadratic Programming (MPQP) for Optimal Gap Selection and Speed Planning of Autonomous Driving

Alexandre Miranda Añon, David Isele

Autonomous DrivingOptimization

🎯 What it does: A multi-configuration quadratic programming-based speed planning algorithm is proposed, which explores interaction gaps in the time-space domain through breadth-first search, and generates optimal speed trajectories for each gap using quadratic programming to achieve smooth and safe speed planning for autonomous driving.

Multi-query TDSP for Path Planning in Time-varying Flow Fields

James Ju Heon Lee, Robert Fitch

OptimizationTime Series

🎯 What it does: Propose a multi-query time-dependent shortest path (TDSP) framework that constructs a shortest path tree analogy using a piecewise linear cost function to achieve optimal splicing of subproblem solutions;

Multi-Radar Inertial Odometry for 3D State Estimation using mmWave Imaging Radar

Jui-Te Huang, Michael Kaess

Autonomous DrivingOptimizationSimultaneous Localization and MappingPoint CloudTime Series

🎯 What it does: Studied multi-radar inertial odometry based on dual millimeter-wave imaging radar and consumer-grade inertial measurement units (IMU), proposing a fixed delay smoothing optimization method to achieve high-precision 3D state estimation.

Multi-Resolution Planar Region Extraction for Uneven Terrains

Yinghan Sun, Wei Zhang

SegmentationComputational EfficiencyPoint Cloud

🎯 What it does: Propose a multi-resolution planar region extraction strategy for extracting planar regions from unordered point cloud measurements in irregular terrain.

Multi-Robot Autonomous Exploration and Mapping Under Localization Uncertainty with Expectation-Maximization

Yewei Huang, Brendan Englot

OptimizationRobotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Propose an autonomous exploration algorithm for decentralized multi-robot teams, considering the uncertainty in mapping and localization for range sensor robots.

Multi-Robot Cooperative Navigation in Crowds: A Game-Theoretic Learning-Based Model Predictive Control Approach

Viet-Anh Le, Andreas A. Malikopoulos

OptimizationRobotic IntelligenceRecurrent Neural Network

🎯 What it does: This paper proposes a multi-robot collaborative navigation control framework that combines local model predictive control (MPC) and social long short-term memory (Social LSTM) models to enable coordinated movement of robots in crowded environments.

Multi-Robot Cooperative Socially-Aware Navigation Using Multi-Agent Reinforcement Learning

Weizheng Wang, Byung-Cheol Min

Robotic IntelligenceTransformerReinforcement LearningBenchmark

🎯 What it does: Built a multi-robot social navigation environment based on Dec-POSMDP and proposed the SAMARL benchmark to enhance the safety and collaborative navigation of multi-robot systems in public spaces.

Multi-robot Human-in-the-loop Control under Spatiotemporal Specifications

Yixiao Zhang, Dimos V. Dimarogonas

Robotic Intelligence

🎯 What it does: Proposes a coordination strategy for multi-robot systems in human-robot interaction scenarios, defining tasks using Signal Temporal Logic (STL) and ensuring task completion through Control Barrier Function (CBF) constraints; generating collision-avoidance trajectories via Nonlinear Model Predictive Control (NMPC), while integrating human-in-the-loop (HIL) models and a novel task allocation protocol; experimental results validate the feasibility and applicability of the method.

Multi-Robot Informative Path Planning from Regression with Sparse Gaussian Processes

Kalvik Jakkala, Srinivas Akella

OptimizationRobotic Intelligence

🎯 What it does: Proposed a multi-robot information path planning method based on sparse Gaussian processes and utilizing gradient descent optimization for path

Multi-robot Search in a 3D Environment with Intersection System Constraints

Yan-Shuo Li, Kuo-Shih Tseng

OptimizationRobotic Intelligence

🎯 What it does: Propose a task allocation method that reformulates the multi-robot search problem as submodular function maximization with intersection system constraints, combining coverage functions and load balance functions to achieve more efficient robot scheduling through routing and clustering constraints.

Multi-Robot Task Allocation Under Uncertainty Via Hindsight Optimization

N. Dhanaraj, Satyandra K. Gupta

OptimizationRobotic Intelligence

🎯 What it does: Propose a hierarchical method to address uncertainty in multi-robot task allocation, where the low-level employs deterministic task allocation optimization, and the high-level generates potential failure combinations through hindsight optimization and invokes the low-level to find the optimal task sequence.

Multi-Sample Long Range Path Planning under Sensing Uncertainty for Off-Road Autonomous Driving

Matt Schmittle, Byron Boots

Autonomous Driving

🎯 What it does: Proposed the DREAMS algorithm for long-distance dynamic replanning of off-road vehicles under perceptual uncertainty.

Multi-task Learning for Real-time Autonomous Driving Leveraging Task-adaptive Attention Generator

Wonhyeok Choi, Sunghoon Im

Object DetectionSegmentationDepth EstimationAutonomous DrivingPoint Cloud

🎯 What it does: Proposed a real-time multi-task network that can simultaneously perform monocular 3D object detection, semantic segmentation, and dense depth estimation.

Multi-Task Learning of Active Fault-Tolerant Controller for Leg Failures in Quadruped robots

Tai-Wei Hou, Lihua Zhang

Robotic Intelligence

🎯 What it does: Designed a hierarchical fault-tolerant control scheme and employed a multi-task learning architecture to actively detect and overcome two types of leg joint failures, training three joint task strategies for healthy, power loss, and locked conditions.

Multimodal Indoor Localization Using Crowdsourced Radio Maps

Zhaoguang Yi, C. X. Lu

Simultaneous Localization and MappingMultimodality

🎯 What it does: The study utilizes crowdsourced radio maps as an alternative to floor plans, constructing a new framework to address issues of map inaccuracy and sparse coverage, while combining uncertainty-aware WiFi localization neural networks and custom Bayesian fusion techniques;

Multimodal Object Query Initialization for 3D Object Detection

Mathijs R. van Geerenstein, D. Gavrila

Object DetectionTransformerImageMultimodalityPoint Cloud

🎯 What it does: Propose EfficientQ3M, an efficient, modular, and multimodal object query initialization scheme for Transformer-based 3D object detection models, combined with a 'modal balance' Transformer decoder, enabling queries to access all sensor modalities throughout the entire decoding process;

Multimodal Transformers for Real-Time Surgical Activity Prediction

Keshara Weerasinghe, H. Alemzadeh

RecognitionComputational EfficiencyTransformerMultimodalityBiomedical Data

🎯 What it does: Proposed a multi-modal Transformer architecture based on short-term kinematics and video clips for real-time identification and prediction of surgical actions and trajectories, along with ablation studies and end-to-end evaluation.

Multimodal Visual-Tactile Representation Learning through Self-Supervised Contrastive Pre-Training

Vedant Dave, Elmar Rueckert

ClassificationRepresentation LearningContrastive LearningMultimodality

🎯 What it does: Utilizing self-supervised contrastive learning to fuse visual and tactile information for material attribute classification and predicting successful grasping

Multiple Update Particle Filter: Position Estimation by Combining GNSS Pseudorange and Carrier Phase Observations

Taro Suzuki

Autonomous DrivingOptimizationSimultaneous Localization and Mapping

🎯 What it does: Propose a multiple update particle filter method targeting sharp peak likelihood functions, used for position estimation by combining GNSS pseudorange and carrier phase observations.

MuRoSim – A Fast and Efficient Multi-Robot Simulation for Learning-based Navigation

Christian Jestel, Oliver Urbann

Robotic IntelligenceReinforcement LearningPoint Cloud

🎯 What it does: Developed MuRoSim, a multi-robot simulation environment specialized for LiDAR navigation and deep reinforcement learning (DRL) training, validated through extensive real-world experiments to demonstrate its realism and transferability.

Mutual Information-calibrated Conformal Feature Fusion for Uncertainty-Aware Multimodal 3D Object Detection at the Edge

A. Stutts, A. Trivedi

Object DetectionAutonomous DrivingComputational EfficiencyAuto EncoderMultimodalityPoint Cloud

🎯 What it does: Developed a lightweight, multimodal 3D object detection framework that combines conformal inference with information theory to achieve uncertainty estimation without Monte Carlo methods, and fuses RGB and LiDAR features through the multivariate Gaussian product of VAE.

N-MPC for Deep Neural Network-Based Collision Avoidance exploiting Depth Images

M. Jacquet, Kostas Alexis

Autonomous DrivingOptimizationSafty and PrivacyImage

🎯 What it does: Proposes a framework combining deep neural networks based on depth images with nonlinear model predictive control (N-MPC) for collision avoidance in drone trajectory tracking.

N-QR: Natural Quick Response Codes for Multi-Robot Instance Correspondence

Nathan Glaser, Z. Kira

Robotic IntelligenceImageAgriculture Related

🎯 What it does: Propose a N-QR (Natural Quick Response codes) for multi-robot instance correspondence, enabling fast and reliable image matching in large-scale heterogeneous robot teams.

Nature-Inspired Bubble Magnetic Microrobots for Multimode Locomotion, Cargo delivery, Imaging, and Biosensing

Zichen Xu, H. Yu

Robotic IntelligenceBiomedical DataUltrasound

🎯 What it does: Proposed a micro-robot based on magnetic bubbles capable of multifunctional applications such as cargo transportation, multimodal movement, micro-manipulation, medical imaging, and biosensing.

NaviFormer: A Data-Driven Robot Navigation Approach via Sequence Modeling and Path Planning with Safety Verification

Xuyang Zhang, Jianmin Ji

Robotic IntelligenceTransformerReinforcement Learning

🎯 What it does: Proposed a sequence modeling-based robot navigation method called NaviFormer, integrated with regularized safety verification to achieve data-driven continuous learning and safe navigation.

NeRF-Enhanced Outpainting for Faithful Field-of-View Extrapolation

Rui Yu, Sharon X. Huang

GenerationNeural Radiance FieldImage

🎯 What it does: Propose a NeRF-enhanced image extrapolation technique (NEO), which utilizes NeRF-generated extended view images to train scene-specific image extrapolation models for achieving faithful FOV extrapolation of real scenes.

NeRF-VINS: A Real-time Neural Radiance Field Map-based Visual-Inertial Navigation System

Saimouli Katragadda, Guoquan Huang

Pose EstimationAutonomous DrivingNeural Radiance FieldSimultaneous Localization and MappingImageMultimodality

🎯 What it does: Designed and implemented a real-time tightly coupled NeRF-based visual-inertial navigation system (NeRF-VINS), which synthesizes new perspectives using NeRF and fuses them with IMU, monocular images, and synthesized images to achieve efficient 3D motion tracking.

Neural Implicit Swept Volume Models for Fast Collision Detection

Dominik Joho, Kirill Safronov

Computational EfficiencyRobotic Intelligence

🎯 What it does: Proposed a new neural implicit sweeping body model for continuous representation of arbitrary robot motion, enabling fast computation of signed distances at any point in the task space to accelerate collision detection; simultaneously combining deep learning with geometric collision checking algorithms to achieve both speed and accuracy.

Neural Informed RRT*: Learning-based Path Planning with Point Cloud State Representations under Admissible Ellipsoidal Constraints

Zhe Huang, K. Driggs-Campbell

Robotic IntelligencePoint Cloud

🎯 What it does: Propose Neural Informed RRT*, combining the asymptotic optimality of RRT* with the acceleration advantages of rule-based information sampling. Utilize point cloud representation for free states, infer guided states near the optimal path through Neural Focus and PointNet++, and introduce Neural Connect to construct connectivity among guided states to improve planning efficiency.

Neural Potential Field for Obstacle-Aware Local Motion Planning

M. Alhaddad, Aleksandr Panov

Autonomous DrivingOptimizationRobotic IntelligenceConvolutional Neural Network

🎯 What it does: Propose a Neural Potential Field model that returns differentiable collision costs under arbitrary obstacle maps and robot footprints, integrating it into MPC for local motion planning.

Neural Radiance Fields for Unbounded Lunar Surface Scene

Xu Zhang, Jihao Yin

Data SynthesisNeural Radiance FieldPoint Cloud

🎯 What it does: Synthesize new views of the lunar surface using NeRF, integrating 3D hash grid and 2D plane grid representations

Neural Rearrangement Planning for Object Retrieval from Confined Spaces Perceivable by Robot’s In-hand RGB-D Sensor

Hanwen Ren, A. H. Qureshi

Robotic IntelligenceImage

🎯 What it does: This paper proposes a neural network-based object retrieval framework capable of re-planning within confined spaces to manipulate unknown, arbitrarily shaped objects to achieve target object acquisition, using a robot-mounted handheld RGB-D camera for active environmental perception.

NGEL-SLAM: Neural Implicit Representation-based Global Consistent Low-Latency SLAM System

Yunxuan Mao, Yiyi Liao

Neural Radiance FieldSimultaneous Localization and MappingImage

🎯 What it does: Proposed NGEL-SLAM, combining traditional modules of feature tracking and loop closure with multi-neural implicit fields to achieve a globally consistent and low-latency SLAM system.

Night-Rider: Nocturnal Vision-aided Localization in Streetlight Maps Using Invariant Extended Kalman Filtering

Tianxiao Gao, Hui Kong

Object DetectionAutonomous DrivingSimultaneous Localization and MappingImageMultimodality

🎯 What it does: Propose a nighttime visual aided localization system based on streetlight maps, adopting a novel data association and matching scheme based on target detection; fusing IMU, odometry, and camera measurements via Invariant Extended Kalman Filter (InEKF) to achieve consistent state estimation at night, and designing a tracking recovery module; experiments show that the relative error is less than 0.2% in four nighttime environments.

NNgTL: Neural Network Guided Optimal Temporal Logic Task Planning for Mobile Robots

Ruijia Liu, Xiang Yin

OptimizationRobotic Intelligence

🎯 What it does: Studied the task planning problem for mobile robots under linear temporal logic (LTL) specifications and proposed a neural network-based sampling strategy;

Noisy Few-shot 3D Point Cloud Scene Segmentation

Hao Huang, Yi Fang

SegmentationMeta LearningGraph Neural NetworkContrastive LearningPoint Cloud

🎯 What it does: A novel few-shot 3D point cloud scene semantic segmentation method is proposed through a meta-learning framework, utilizing multi-prototype graph construction, graph structure-based denoising, subgraph bagging schemes for semi-supervised transfer learning, and triplet contrastive loss to enhance prototype feature discriminability.

NoMaD: Goal Masked Diffusion Policies for Navigation and Exploration

A. Sridhar, Sergey Levine

Robotic IntelligenceTransformerDiffusion model

🎯 What it does: Train a unified diffusion strategy that can perform both goal-oriented navigation and target-free exploration, addressing the challenge of traditional methods requiring separate models.

Non-Axiomatic Reasoning for an Autonomous Mobile Robot

Patrick Hammer, Jana Tumova

Robotic Intelligence

🎯 What it does: Integrate the non-axiomatic reasoning system NARS into a mobile robot to achieve planning and decision-making, completed within ROS nodes

Non-Intrusive LiDAR Protection Module Emulating Bio-Inspired Wiping Motion for Outdoor Unmanned Vehicles

Youngrae Kim, Dong-Woo Yun

Autonomous DrivingPoint Cloud

🎯 What it does: Developed a non-intrusive LiDAR protection module that employs bio-inspired wiping motion for outdoor unmanned vehicles.

Non-singular Fast Terminal Adaptive Visual Tracking Control with Reduced Tuning Parameters for an Aerial Vehicle Under Perturbations

Gustavo Olivas-Martínez, Herman Castañeda

Robotic IntelligenceOptical FlowImage

🎯 What it does: A robust image-based visual servo design is proposed for visual target tracking of quadrotor drones in turbulent wind environments.

Non-Verbal Cues on Robot-Group Persuasion

Alexandra Gonçalves, Alexandre Bernardino

Robotic Intelligence

🎯 What it does: Designed and experimentally tested the nonverbal cues (gaze and gestures) of the robot Vizzy in group persuasion.

NPC: Neural Predictive Control for Fuel-Efficient Autonomous Trucks

Jiaping Ren, Wei Li

Data SynthesisAutonomous DrivingOptimizationTransformer

🎯 What it does: Propose a pure data-driven neural predictive control method (NPC) that achieves fuel consumption optimization without requiring a vehicle physical model.

Nullspace Adaptive Model-Based Trajectory-Tracking Control for a 6-DOF Underwater Vehicle with Unknown Plant and Actuator Parameters: Theory and Preliminary Simulation Evaluation

Annie M. Mao, L. Whitcomb

Autonomous DrivingOptimizationRobotic Intelligence

🎯 What it does: Proposed a model-based null space adaptive trajectory tracking control algorithm that can simultaneously estimate unknown plant and actuator parameters;

nvblox: GPU-Accelerated Incremental Signed Distance Field Mapping

A. Millane, R. Siegwart

Robotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Developed the nvblox library, utilizing GPU to achieve incremental construction of robot voxel maps and computation of Euclidean Signed Distance Fields (ESDF)

NYC-Indoor-VPR: A Long-Term Indoor Visual Place Recognition Dataset with Semi-Automatic Annotation

Diwei Sheng, Chen Feng

RecognitionRetrievalImageVideoBenchmark

🎯 What it does: Introduces the NYC-Indoor-VPR dataset and proposes a semi-automated annotation method to establish geographic location labels for indoor visual place recognition (VPR), followed by a benchmark evaluation of multiple state-of-the-art VPR algorithms on this dataset.

OASIS: Optimal Arrangements for Sensing in SLAM

Pushyami Kaveti, David M. Rosen

OptimizationSimultaneous Localization and MappingBenchmark

🎯 What it does: Conduct an information-theoretic study on sensor placement for mobile robots, proposing the OASIS method and formalizing it as an E-optimal subset selection problem.

Object Permanence Filter for Robust Tracking with Interactive Robots

Shaoting Peng, Nadia Figueroa

Object TrackingRobotic Intelligence

🎯 What it does: Proposed the Object Permanence Filter (OPF), which integrates object permanence assumptions and rules into a particle filter to achieve robust tracking in multi-object, multi-agent interaction scenarios.

Object-centric Cross-modal Feature Distillation for Event-based Object Detection

Lei Li, Dengxin Dai

Object DetectionKnowledge DistillationMultimodality

🎯 What it does: Propose a cross-modal feature distillation method that employs an object-based slot attention mechanism to enhance the real-time object detection performance of event cameras.

Object-Centric Instruction Augmentation for Robotic Manipulation

Junjie Wen, Jian Tang

Robotic IntelligenceTransformerLarge Language ModelVision Language ModelMultimodality

🎯 What it does: Propose the Object-Centric Instruction Augmentation (OCI) framework, which embeds object location information into language instructions using multimodal large language models (MLLM), and integrates pre-trained MLLM's visual-language features into robot control policy networks through a feature reuse mechanism, enhancing the robot's performance in tasks such as 'grasping-placing'.

Observation Time Difference: an Online Dynamic Objects Removal Method for Ground Vehicles

Rongguang Wu, Zheng Fang

Autonomous DrivingComputational EfficiencyPoint Cloud

🎯 What it does: A ground vehicle online dynamic object removal method based on observation time difference is proposed. Dynamic objects are divided into two categories: suddenly appearing and suddenly disappearing, which are removed using downward retrieval and upward retrieval techniques, respectively.

Observer-based Controller Design for Oscillation Damping of a Novel Suspended Underactuated Aerial Platform

Hemjyoti Das, Christian Ott

OptimizationTime SeriesPhysics Related

🎯 What it does: A controller based on an observer was designed for oscillation damping of a novel suspended, undriven aerial platform. The platform employs a spherical double pendulum model, using only available onboard IMU data to construct an extended Kalman filter (EKF) for state estimation. Optimal state feedback controllers and PD+ controllers were separately designed in joint space and task space to suppress oscillations.