arXivSub Start free trial

IROS 2023 Papers — Page 9

IEEE/RSJ International Conference on Intelligent Robots and Systems · 1195 papers

Open-Vocabulary Affordance Detection in 3D Point Clouds

Toan Ngyen, A. Nguyen

Object DetectionVision-Language-Action ModelPoint Cloud

🎯 What it does: Propose an Open-Vocabulary Affordance Detection (OpenAD) method that can detect an infinite number of affordances in 3D point clouds.

Optical Flow Boosts Unsupervised Localization and Segmentation

Xinyu Zhang, Abdeslam Boularias

Object DetectionSegmentationTransformerContrastive LearningOptical Flow

🎯 What it does: Fine-tune an unsupervised vision transformer (ViT) using optical flow information to enhance performance in unsupervised object localization and segmentation.

Optimal and Stable Multi-Layer Object Rearrangement on a Tabletop

Andy Xu, Jingjin Yu

OptimizationRobotic Intelligence

🎯 What it does: This paper investigates the multi-layer desktop object rearrangement (MORT) problem and proposes an optimal and stable rearrangement algorithm;

Optimal Cost-Preference Trade-Off Planning with Multiple Temporal Tasks

Peter Amorese, Morteza Lahijanian

Optimization

🎯 What it does: Proposed a new preference concept and constructed a Pareto compromise planning framework for multi-task scenarios with optimal behavior of preferences and resources, and implemented efficient Pareto-optimal plan generation based on A* search;

Optimal Decision Making in Robotic Assembly and Other Trial-and-Error Tasks

J. Watson, N. Correll

OptimizationRobotic IntelligenceConvolutional Neural Network

🎯 What it does: Studied robot tasks under uncertainty in perception, execution, and environment by constructing low-entropy success/failure indicators and high-entropy prediction data, achieving a closed-form solution that predicts failures in advance and restarts to reduce total completion time, validated in a robot slot assembly experiment.

Optimal Energy Tank Initialization for Minimum Sensitivity to Model Uncertainties

Andrea Pupa, Cristian Secchi

OptimizationRobotic Intelligence

🎯 What it does: Propose a new energy tank initialization strategy that utilizes closed-loop state sensitivity to derive the precise boundaries of energy behavior, and employs nonlinear optimization to find the optimal trajectory and minimum initial energy under robot models with parameter uncertainty, achieving localization tasks.

Optimization-Based VINS: Consistency, Marginalization, and FEJ

Chuchu Chen, Guoquan Huang

OptimizationSimultaneous Localization and Mapping

🎯 What it does: This paper provides a comprehensive analysis of the FEJ method's application in nonlinear optimization-based visual inertial navigation systems (VINS), and offers guidance on correctly implementing FEJ in four common marginalization architectures.

Optimizing Algorithms from Pairwise User Preferences

L. Keselman, Aaron Steinfeld

OptimizationRobotic Intelligence

🎯 What it does: Propose the SortCMA algorithm, which uses pairwise user preferences for high-dimensional algorithm parameter optimization, and applies it to tuning commercial depth sensors and robot social navigation without real annotations;

Optimizing the Extended Fourier Mellin Transformation Algorithm

W. Jiang, Sören Schwertfeger

Pose EstimationSimultaneous Localization and MappingImage

🎯 What it does: An optimized algorithm (o-eFMT) based on the extended Fourier-Mellin transform (eFMT) was developed for visual odometry, improving uncertainty extraction, the objective function, and graph optimization in the small loop (three consecutive frames).

Orbital Head-Mounted Display: A Novel Interface for Viewpoint Control during Robot Teleoperation in Cluttered Environments

Sjoerd Kuitert, Luka Peternel

Robotic Intelligence

🎯 What it does: Proposed an Orbital Head-Mounted Display (OHMD) interface, allowing operators to simultaneously control a 6-DoF camera platform and a robotic arm, while experiencing a remote environment through a VR headset.

Orientation Control with Variable Stiffness Dynamical Systems

Youssef Michel, Dongheui Lee

Robotic Intelligence

🎯 What it does: Propose a closed-loop control algorithm for robots in rotational motion and damping adaptation.

Output Feedback Formation Control of a School of Robotic Fish with Artificial Lateral Line Sensing

A. Wolek, D. Paley

Robotic Intelligence

🎯 What it does: Proposed an estimation and control framework based on artificial lateral line perception to achieve stable parallel motion of robotic fish schools;

Overtaking Moving Obstacles with Digit: Path Following for Bipedal Robots via Model Predictive Contouring Control

K. S. Narkhede, I. Poulakakis

OptimizationRobotic Intelligence

🎯 What it does: Proposes a Model Predictive Contour Control (MPCC) method for selecting gaits along a global path to maximize path traversal while balancing precise tracking and rapid passage.

P2O-Calib: Camera-LiDAR Calibration Using Point-Pair Spatial Occlusion Relationship

Su Wang, Xuchong Qiu

Autonomous DrivingPoint Cloud

🎯 What it does: Proposes a novel un-calibrated target camera-LiDAR extrinsic calibration method based on 3D spatial occlusion relationships for 2D-3D edge point extraction and occlusion-guided point matching.

P4P: Conflict-Aware Motion Prediction for Planning in Autonomous Driving

Qiao Sun, Hang Zhao

Autonomous DrivingSafty and PrivacyPoint Cloud

🎯 What it does: Proposed a conflict-aware motion prediction method called P4P to enhance the safety of autonomous driving planning

PACT: Perception-Action Causal Transformer for Autoregressive Robotics Pre-Training

Rogerio Bonatti, Ashish Kapoor

Representation LearningRobotic IntelligenceTransformerSupervised Fine-TuningImagePoint Cloud

🎯 What it does: Proposes PACT (Perception-Action Causal Transformer), a self-supervised generative model based on Transformer architecture, for learning perception-action causal representations from robot data, and achieves performance improvements in multi-task scenarios (e.g., safe navigation, localization, and mapping) through fine-tuning of small task-specific networks.

PaintNet: Unstructured Multi-Path Learning from 3D Point Clouds for Robotic Spray Painting

Gabriele Tiboni, T. Tommasi

Robotic IntelligencePoint CloudBenchmark

🎯 What it does: Propose a framework capable of processing arbitrary 3D point clouds and predicting multiple unstructured trajectories for industrial painting tasks, and release the first public PaintNet dataset.

PanelPose: A 6D Pose Estimation of Highly-Variable Panel Object for Robotic Robust Cockpit Panel Inspection

Han Sun, Qixin Cao

Data SynthesisPose EstimationRobotic IntelligenceSimultaneous Localization and MappingImage

🎯 What it does: Proposed a 6D pose estimation method called PanelPose for handling highly variable panel objects in robot inspection scenarios within aviation cockpits.

PANet: LiDAR Panoptic Segmentation with Sparse Instance Proposal and Aggregation

Jianbiao Mei, Yong Liu

SegmentationAutonomous DrivingPoint Cloud

🎯 What it does: Propose the PANet framework, utilizing sparse instance proposal (SIP) and an instance aggregation module for LiDAR panoramic segmentation, removing the offset branch and enhancing performance on large objects.

Panoptic Mapping with Fruit Completion and Pose Estimation for Horticultural Robots

Yue Pan, C. Stachniss

Pose EstimationSimultaneous Localization and MappingAgriculture Related

🎯 What it does: Proposed an online multi-resolution panoramic mapping system for simultaneously performing fruit 3D shape reconstruction and pose estimation in multi-resolution maps.

PanopticNDT: Efficient and Robust Panoptic Mapping

Daniel Seichter, H. Groß

Computational EfficiencyRobotic IntelligenceSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposed an efficient and robust panoramic mapping method called PanopticNDT based on the Occupancy Normal Distribution Transform (NDT).

Parallel Cell Array Patterning and Target Cell Lysis on an Optoelectronic Micro-Well Device

Chunyuan Gan, Lin Feng

Biomedical Data

🎯 What it does: Developed an electrical method based on a photoelectrode microhole array, achieving cell arraying and electroporation lysis of target cells through light-induced depolarization forces.

Parallel-Jaw Gripper and Grasp Co-Optimization for Sets of Planar Objects

Rebecca H. Jiang, Alberto Rodriguez

OptimizationRobotic Intelligence

🎯 What it does: Proposed a joint optimization framework for planar parallel clamp-type grippers and grasping actions targeting multi-faceted objects

Part-level Scene Reconstruction Affords Robot Interaction

Zeyu Zhang, Hangxin Liu

OptimizationRobotic Intelligence

🎯 What it does: Proposes a component-level interactive scene reconstruction method, which reassembles objects using original shapes to accurately replicate observed physical scenes and support robot interaction with rigid and articulated objects.

Path-Following Control with Path and Orientation Snap-In

C. Hartl-Nesic, Andreas Kugi

Robotic Intelligence

🎯 What it does: Proposed an extension of multi-path path following control, and designed two physical human-robot interaction modes: path snap-in and orientation snap-in, using attractive forces to guide the robot's end-effector to the path or predefined posture.

Perceptive Hexapod Legged Locomotion for Climbing Joist Environments

Zixian Zang, A. Zakhor

Depth EstimationKnowledge DistillationRobotic IntelligenceReinforcement LearningImage

🎯 What it does: Developed a perceptual gait model utilizing self-perspective depth maps, enabling a six-legged robot to navigate narrow attic wooden truss structures and achieve zero-shot transfer from simulation to reality.

Performance Comparison of Teleoperation Interfaces for Ultra-Lightweight Anthropomorphic Arms

Filip Zorić, A. Ollero

Pose EstimationRobotic Intelligence

🎯 What it does: This paper conducted a comparative evaluation of the performance of three teleoperation interfaces on an ultra-lightweight (<3 kg) dual-arm humanoid robot, aiming to achieve flexibility, accuracy, and agility in complex manipulation tasks;

Perirobot Space Representation for HRI: Measuring and Designing Collaborative Workspace Coverage by Diverse Sensors

Jakub Rozlivek, M. Hoffmann

Robotic IntelligenceSimultaneous Localization and MappingMultimodality

🎯 What it does: Defined the space around the robot that needs to be monitored, represented all sensor data as occupancy information, and used occupancy-based metrics to calculate the effectiveness of sensor coverage in the workspace.

Perseus AUV: Towards Linear Convoying of Agile A-Sized AUVs Through Acoustic Track-and-Trail

N. Rypkema, Michael Triantafyllou

Autonomous Driving

🎯 What it does: Designed and demonstrated a Perseus A-sized AUV equipped with a low-cost passive inverted ultra-short baseline (piUSBL) acoustic receiver system, capable of performing acoustic tracking and following a leading vehicle carrying an acoustic source, and achieving good directional stability and maneuverability through tuna-inspired variable fins; the system was validated through field experiments in Massachusetts' Charles River.

Persuasive Polite Robots in Free-Standing Conversational Groups

S. Zojaji, C. Peters

Robotic Intelligence

🎯 What it does: This paper experimentally investigates the use of six politeness behaviors based on Brown and Levinson's politeness theory by a humanoid Pepper robot in a free-standing conversation group, evaluating human participants' decisions to join positions (especially more distant ones) and their perception of the robot's politeness.

Physical Contact with Wall using a Multirotor UAV Equipped with Add-On Thruster for Inspection Work

Takamasa Kominami, K. Shimonomura

Robotic IntelligenceUltrasoundPhysics Related

🎯 What it does: Developed a multirotor drone equipped with a single horizontal thruster for physical contact non-destructive testing on vertical walls.

Physics-Informed Learning to Enable Robotic Screw-Driving Under Hole Pose Uncertainties

O. Manyar, Satyandra K. Gupta

Robotic IntelligencePhysics Related

🎯 What it does: Design a mobile manipulator system using active impedance control and passive elastic driving tools, combining physics-informed learning to automatically model screw tip motion, detect failures, and perform corrections;

PIMbot: Policy and Incentive Manipulation for Multi-Robot Reinforcement Learning in Social Dilemmas

Shahab Nikkhoo, Cong Liu

Reinforcement Learning

🎯 What it does: Propose PIMbot that adjusts the reward function in multi-robot collaboration through strategy and incentive manipulation methods to influence cooperative behavior in social dilemmas.

Placing by Touching: An Empirical Study on the Importance of Tactile Sensing for Precise Object Placing

Luca Lach, G. Chalvatzaki

Robotic IntelligenceMultimodality

🎯 What it does: Proposes an object placement method that combines tactile feedback with proprioception, using a neural network called PlaceNet to estimate rotation matrices and correct the gripper for stable placement.

Planning and Control for a Dynamic Morphing-Wing UAV Using a Vortex Particle Model

Gino Perrotta, Joseph L. Moore

OptimizationRobotic IntelligencePhysics Related

🎯 What it does: Proposed a vortex particle model and a model-based controller to account for unsteady aerodynamics during post-stall maneuvering in dynamic variable-wing UAVs.

PLPL-VIO: A Novel Probabilistic Line Measurement Model for Point-Line-Based Visual-Inertial Odometry

Zewen Xu, Xin Jin

Pose EstimationOptimizationSimultaneous Localization and Mapping

🎯 What it does: Proposes a visual-inertial odometry (VIO) system based on point-line features, and introduces a new line feature measurement model and uncertainty handling method.

Point2Point: A Framework for Efficient Deep Learning on Hilbert Sorted Point Clouds with Applications in Spatio-Temporal Occupancy Prediction

A. Pandhare

Autonomous DrivingComputational EfficiencyPoint Cloud

🎯 What it does: Proposes using Hilbert space-filling curves to construct a one-dimensional ordering that preserves locality for representing point clouds, and designs a neural network architecture called Point2Point that can effectively learn from Hilbert-ordered point clouds.

Poly-MOT: A Polyhedral Framework For 3D Multi-Object Tracking

Xiaoyu Li, K. Wang

Object TrackingAutonomous DrivingPoint Cloud

🎯 What it does: Proposes the Poly-MOT framework, which selects the most suitable tracking criteria for different object categories, employs multiple motion models and three custom similarity metrics, and adopts a two-stage data association strategy.

Polymer-Based Self-Calibrated Optical Fiber Tactile Sensor

Wen-Tzu Chen, Jia Pan

Robotic Intelligence

🎯 What it does: Designed and implemented a polymer-based self-calibrating optical fiber tactile sensor capable of separating normal force and shear force and calibrating according to the size of the contacted object.

Polynomial-Based Online Planning for Autonomous Drone Racing in Dynamic Environments

Qianhao Wang, Fei Gao

OptimizationRobotic Intelligence

🎯 What it does: Propose an online replanning framework for UAV racing in dynamic environments, using polynomial trajectory representation to optimize the balance between high speed and obstacle avoidance, and incorporating hard constraints to ensure passing through intermediate track points; for dynamic obstacles, parallel multi-topology planning is adopted to avoid local optima causing time loss, ultimately achieving first place in the DJI Robomaster UAV competition, completing the track in half the time of the second place.

POMDP-Guided Active Force-Based Search for Robotic Insertion

Chen Wang, Wei Zhang

Robotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a proactive dynamics search strategy based on POMDP, utilizing contact information for insertion tasks

PoseFusion: Robust Object-in-Hand Pose Estimation with SelectLSTM

Yuyang Tu, Jianwei Zhang

Pose EstimationRecurrent Neural NetworkImageMultimodalityBenchmark

🎯 What it does: Proposes PoseFusion, a multimodal fusion method combining vision and tactile sensing for robust hand-held object pose estimation, and creates a 6D hand-held object pose dataset containing RGB-D, tactile, and proprioceptive data.

Powered Knee and Ankle Prosthesis Control for Adaptive Ambulation at Variable Speeds, Inclines, and Uneven Terrains

Liam M. Sullivan, Tommaso Lenzi

Robotic Intelligence

🎯 What it does: Designed and implemented a knee and ankle electric prosthetic controller capable of continuously adapting to varying walking speeds, slopes, and uneven terrains.

Pre-and Post-Contact Policy Decomposition for Non-Prehensile Manipulation with Zero-Shot Sim-To-Real Transfer

Minchan Kim, Beomjoon Kim

Domain AdaptationRobotic IntelligenceReinforcement Learning

🎯 What it does: Proposed a non-grasping operating system that can move objects to target positions through multiple contact mode transitions and by leveraging environmental contact points, achieving zero-shot transfer from simulation to real-world environments

Pred-NBV: Prediction-Guided Next-Best-View Planning for 3D Object Reconstruction

Harnaik Dhami, Pratap Tokekar

RestorationGenerationTransformerPoint CloudMesh

🎯 What it does: Proposed the Pred-NBV method for 3D object reconstruction, using prediction to guide the planning of the next best view.

Predicting Center of Mass by Iterative Pushing for Object Transportation and Manipulation

Steven M. Hyland, C. Onal

Robotic Intelligence

🎯 What it does: Propose a method using small mobile robots to estimate the center of mass of approximately planar objects via iterative pushing

Predicting Energy Consumption and Traversal Time of Ground Robots for Outdoor Navigation on Multiple Types of Terrain

Matthias Eder, Gerald Steinbauer-Wagner

OptimizationRobotic IntelligenceConvolutional Neural Network

🎯 What it does: A data-driven supervised machine learning approach is used to construct a cost representation model for energy consumption and travel time across different outdoor terrains;

Primitive Skill-Based Robot Learning from Human Evaluative Feedback

Ayano Hiranaka, Ruohan Zhang

Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement Learning

🎯 What it does: Proposes a framework called SEED, which combines reinforcement learning with human evaluation feedback (RLHF) and skill-based reinforcement learning to address sparse rewards and safety issues in long-horizon robotic manipulation tasks.

Principled ICP Covariance Modelling in Perceptually Degraded Environments for the EELS Mission Concept

W. Talbot, V. Ila

Simultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposed a principled modeling approach for LiDAR point-to-plane ICP covariance, and conducted comparative evaluation of new and old models within the complete SLAM pipeline.

Printable Bistable Structures for Programmable Frictional Skins of Soft-Bodied Robots

Tung D. Ta, Yoshihiro Kawahara

Robotic Intelligence

🎯 What it does: A programmable friction skin based on bistable bellow structures is proposed, which can switch between two folded states, dynamically alter the contact points with the ground, and achieve tunable anisotropic friction properties.

Prioritized Planning for Target-Oriented Manipulation via Hierarchical Stacking Relationship Prediction

Zewen Wu, Nanning Zheng

Robotic IntelligenceImage

🎯 What it does: Propose a goal-oriented priority planning method based on hierarchical stacking relationship prediction

Privacy-Preserving and Uncertainty-Aware Federated Trajectory Prediction for Connected Autonomous Vehicles

Muzi Peng, Lili Su

Autonomous DrivingFederated LearningSafty and PrivacySequential

🎯 What it does: Proposed FLTP (Federated Learning Trajectory Prediction) and ALFLTP (FLTP with Active Learning) algorithms, achieving privacy-preserving and uncertainty-aware trajectory prediction on connected autonomous vehicles.

Proactive Model Predictive Control with Multi-Modal Human Motion Prediction in Cluttered Dynamic Environments

Lukas Heuer, K. Arras

Autonomous DrivingOptimizationRobotic IntelligenceMultimodality

🎯 What it does: Proposes a new framework that integrates multimodal human motion prediction into nonlinear model predictive control for robot navigation in crowded dynamic environments.

Proactive Opinion-Driven Robot Navigation Around Human Movers

Charlotte Cathcart, N. Leonard

Robotic Intelligence

🎯 What it does: Proposed, analyzed, and experimentally verified an active robot social navigation method driven by the robot's 'views' formed from the direction and degree of movement of human pedestrians, which evolve over time according to nonlinear dynamics and utilize attention to social cues to determine passage strategies.

Probabilistic Guarantees for Nonlinear Safety-Critical Optimal Control

Prithvi Akella, A. Ames

OptimizationSafty and Privacy

🎯 What it does: Provides three algorithms to generate three categories of probabilistic guarantees for general nonlinear safety-critical finite-time optimal controllers: optimality of the solution, recursive feasibility, and maximum controller runtime.

Probabilistic Slide-support Manipulation Planning in Clutter

Shusei Nagato, K. Harada

Robotic Intelligence

🎯 What it does: Proposed a dual-arm robot sliding-support strategy, where one arm slides the target object out of a cluttered environment while the other arm supports surrounding objects to prevent clutter collapse, achieving safe and efficient extraction of the target object.

Probabilistic Traversability Model for Risk-Aware Motion Planning in Off-Road Environments

Xiaoyi Cai, J. How

Autonomous Driving

🎯 What it does: Proposed a probabilistic off-road traversability model based on the empirical distribution of trajectory parameters, performing risk-aware path planning under the no-slip assumption.

Programable On-Chip Fabrication of Magnetic Soft Micro-Robot

Yuke Li, T. Arai

Robotic Intelligence

🎯 What it does: Proposed a programmable preparation method for magnetic soft microrobots through an on-chip photopolymerization system, assembling superparamagnetic nanoparticles via their magnetic anisotropy and fixing them using hydrogel photopolymerization, successfully fabricating joint rotation mechanisms and snake-like microrobots that achieve the intended motion.

Projecting Robot Intentions Through Visual Cues: Static vs. Dynamic Signaling

Shubham D. Sonawani, H. B. Amor

Robotic IntelligenceImageVideo

🎯 What it does: Compare the effectiveness of static and dynamic visual signals in collaborative object sorting tasks, assess their impact on human behavior, and quantify the information transfer between visual signals and human behavior using information theory methods.

Proprioception and Reaction for Walking Among Entanglements

Justin K. Yim, Aaron M. Johnson

Robotic Intelligence

🎯 What it does: Proprioceptive method and reactive strategy for legged robots to perceive entanglement in entangled environments and perform untangling during the swing phase

Proprioception and Tail Control Enable Extreme Terrain Traversal by Quadruped Robots

Yanhao Yang, Aaron M. Johnson

Robotic Intelligence

🎯 What it does: Developed proprioception-based gait planning and two-degree-of-freedom tail control, enabling quadruped robots to dynamically traverse extreme terrains with sudden height variations.

Proprioceptive External Torque Learning for Floating Base Robot and its Applications to Humanoid Locomotion

Daegyu Lim, Jaeheung Park

Robotic IntelligenceRecurrent Neural NetworkTime SeriesSequential

🎯 What it does: Learn and estimate external joint torques and contact torques of a floating base robot using only proprioceptive sensors such as encoders and IMUs, and verify their application in gait control.

Protective Skin Mechanism with an Exhaustive Arrangement of Tiny Rigid Bodies for Soft Robots: Evaluation of Puncture Resistance, Elasticity, and Descaling Resistance of the Scale Mechanism

K. Tadakuma, S. Tadokoro

Robotic Intelligence

🎯 What it does: Proposed a skin protection mechanism for soft robots using densely packed small rigid bodies, and tested by sewing small pieces with Kevlar threads on elastic sheets.

Prototypical Contrastive Transfer Learning for Multimodal Language Understanding

Seitaro Otsuki, Komei Sugiura

Domain AdaptationRepresentation LearningData-Centric LearningContrastive LearningMultimodality

🎯 What it does: Propose a novel multimodal language understanding transfer learning method called Prototypical Contrastive Transfer Learning (PCTL), and apply it to locate target objects in a home environment based on free-form natural language instructions.

Provably Correct Sensor-Driven Path-Following for Unicycles Using Monotonic Score Functions

Be Clark, Hasan A. Poonawala

Robotic IntelligenceScore-based Model

🎯 What it does: Designed a stable controller based on sensor measurements for path tracking in robots with single-wheel kinematics (e.g., wheeled mobile robots);

ProxMaP: Proximal Occupancy Map Prediction for Efficient Indoor Robot Navigation

V. Sharma, Pratap Tokekar

Computational EfficiencyRobotic IntelligenceSimultaneous Localization and Mapping

🎯 What it does: Designed a self-supervised occupancy prediction technique ProxMaP to predict the occupancy status of areas near the robot to accelerate navigation

Pseudo Inputs Optimisation for Efficient Gaussian Process Distance Fields

Lan Wu, Teresa Vidal-Calleja

OptimizationComputational EfficiencySimultaneous Localization and Mapping

🎯 What it does: Proposes an efficient distance field representation based on Gaussian processes, utilizing pseudo-input optimization to generate accurate distance fields and iteratively building maps in dynamic environments.

Pseudo-Stereo++: Cycled Generative Pseudo-Stereo for Monocular 3D Object Detection in Autonomous Driving

Ahmed Elhagry, Abdulmotaleb El-Saddik

Object DetectionAutonomous DrivingConvolutional Neural NetworkContrastive LearningImage

🎯 What it does: Proposed a Cyclic Generative Pseudo Stereo (CGPS) architecture that generates right views from monocular left views, creating pseudo stereo pairs for stereo 3D detectors to use.

PTDRL: Parameter Tuning Using Deep Reinforcement Learning

Elias Goldsztejn, R. Brafman

Hyperparameter SearchReinforcement Learning

🎯 What it does: Propose the PTDRL parameter tuning strategy, which automatically selects parameters from a fixed parameter set to maximize expected rewards, thereby improving indoor social navigation performance.

Push to Know! - Visuo-Tactile Based Active Object Parameter Inference with Dual Differentiable Filtering

Anirvan Dutta, Mohsen Kaboli

Robotic IntelligenceMultimodality

🎯 What it does: Estimate key physical parameters of objects (shape, friction coefficient, mass, center of gravity, inertia) through non-grasping pushing operations combined with visual and tactile perception.

PuSHR: A Multirobot System for Nonprehensile Rearrangement

Sidharth Talia, S. Srinivasa

OptimizationRobotic Intelligence

🎯 What it does: This study addresses the problem of non-grasping object rearrangement using vehicle-type pushing robots, constructing the PuSHR multi-robot system. It optimizes task allocation and trajectory planning in the offline phase, and achieves decentralized trajectory tracking in the online phase.

Pyramid Semantic Graph-Based Global Point Cloud Registration with Low Overlap

Zhijian Qiao, S. Shen

Pose EstimationAutonomous DrivingOptimizationGraph Neural NetworkPoint Cloud

🎯 What it does: Propose a global point cloud registration framework based on semantic graphs to address registration problems under low overlap conditions.

QDP: Learning to Sequentially Optimise Quasi-Static and Dynamic Manipulation Primitives for Robotic Cloth Manipulation

David Blanco Mulero, Ville Kyrki

OptimizationRobotic IntelligenceReinforcement Learning

🎯 What it does: Propose the QDP method to optimize the speed, trajectory, and pickup/drop-off positions of fabric manipulation primitives, achieving adaptability to different fabric properties.

Quadratic Dynamic Matrix Control for Fast Cloth Manipulation

Edoardo Caldarelli, Carme Torras

OptimizationRobotic Intelligence

🎯 What it does: Integrate Quadratic Dynamic Matrix Control (QDMC) with chance constraints for robotic fabric manipulation.

Quadrupedal Footstep Planning Using Learned Motion Models of a Black-Box Controller

I. Taouil, Sven Behnke

OptimizationRobotic IntelligenceImage

🎯 What it does: This paper learns a motion model that maps high-level velocity commands from a black-box locomotion controller to the center of mass (CoM) and toe movements. These models are combined with a variant of the A* algorithm to plan CoM trajectories, footstep sequences, and corresponding high-level velocity commands based on visual information, enabling safe quadruped robot locomotion over irregular terrain.

Quantized Distillation: Optimizing Driver Activity Recognition Models for Resource-Constrained Environments

Calvin Tanama, Alina Roitberg

RecognitionAutonomous DrivingOptimizationComputational EfficiencyKnowledge DistillationConvolutional Neural Network

🎯 What it does: Propose a lightweight framework that improves 3D MobileNet by combining knowledge distillation and model quantization to achieve driver behavior recognition in resource-constrained environments.

RACECAR - The Dataset for High-Speed Autonomous Racing

A. Kulkarni, Madhur Behl

Autonomous DrivingMultimodalityPoint Cloud

🎯 What it does: Created and released the RACECAR high-speed full-scale autonomous racing car dataset, collected multi-modal sensor data, covering 11 track scenarios, 27 events, and 6.5 hours of racing;

RADA: Robust Adversarial Data Augmentation for Camera Localization in Challenging Conditions

Jialu Wang, Andrew Markham

Autonomous DrivingAdversarial AttackImage

🎯 What it does: Propose a robust adversarial data augmentation method called RADA to enhance the robustness of camera localization under complex conditions.

RAIST: Learning Risk Aware Traffic Interactions via Spatio-Temporal Graph Convolutional Networks

Videsh Suman, Aniket Bera

Autonomous DrivingGraph Neural Network

🎯 What it does: Propose a self-perspective driving framework based on spatiotemporal traffic graphs, and train graph edges using spatiotemporal graph convolutional networks (ST-GCN) to learn risk-aware traffic interaction representations

RAMP: Hierarchical Reactive Motion Planning for Manipulation Tasks Using Implicit Signed Distance Functions

V. Vasilopoulos, Volkan Isler

OptimizationRobotic Intelligence

🎯 What it does: Proposes RAMP—a hierarchical motion planning framework that combines sampling-based and reactive methods for manipulation tasks; generates trajectories using a novel MPPI controller, which are then asynchronously followed by a local vector field controller; achieves rapid path planning, task constraint satisfaction, and obstacle-safe responses in desktop cleaning scenarios.

Range-based GP Maps: Local Surface Mapping for Mobile Robots using Gaussian Process Regression in Range Space

Margaret Hansen, David Wettergreen

Robotic IntelligenceSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Proposed and implemented a Gaussian Process (GP) map based on LiDAR range, representing terrain by directly modeling the sensor range in spherical space.

Rapid Grasping of Fabric Using Bionic Soft Grippers with Elastic Instability

Zechen Xiong, Hod Lipson

Robotic Intelligence

🎯 What it does: Designed and demonstrated an elastic unstable soft gripper inspired by the pinching action of the human thumb and index finger, capable of achieving wide-range rapid closure;

RaPlace: Place Recognition for Imaging Radar using Radon Transform and Mutable Threshold

Hyesu Jang, Ayoung Kim

RecognitionRetrievalImage

🎯 What it does: Propose a radar-based place recognition method that calculates similarity using sinogram images from the Radon transform and frequency domain cross-correlation.

RaSpectLoc: RAman SPECTroscopy-dependent robot LOCalisation

C. Thirgood, Simon J. Hadfield

Robotic IntelligenceSimultaneous Localization and MappingBenchmark

🎯 What it does: Proposed and implemented the first robot localization algorithm based on material composition, which utilizes a Raman spectrometer to acquire material information and integrates it with visual, structural, and semantic features to improve localization accuracy.

Re-Evaluating Parallel Finger-Tip Tactile Sensing for Inferring Object Adjectives: An Empirical Study

Fangyi Zhang, Peter Corke

ClassificationRobotic IntelligenceTabular

🎯 What it does: Investigate the contribution of individual taxels in fingertip tactile sensors when inferring object adjectives.

Re-Thinking Classification Confidence with Model Quality Quantification

Yancheng Pan, Huijing Zhao

ClassificationExplainability and Interpretability

🎯 What it does: Proposed a confidence estimation method based on model quality quantization, and developed two evaluation metrics, MQ-Repres and MQ-Discri, as well as the MQ-Conf confidence estimation;

Reachability-Aware Collision Avoidance for Tractor-Trailer System with Non-Linear MPC and Control Barrier Function

Yucheng Tang, Björn Hein

Autonomous DrivingOptimization

🎯 What it does: Proposes a method combining reachability-aware model predictive control (MPC) with discrete control barrier functions (CBF) for backward obstacle avoidance in tractor-trailer systems.

Reactive and Safe Co-Navigation with Haptic Guidance

Mela C. Coffey, Alyssa Pierson

Autonomous DrivingSafty and Privacy

🎯 What it does: Propose a human-robot collaborative navigation algorithm where the robot uses tactile feedback to achieve collision avoidance and path suggestions, supporting humans in high-level path decision-making.

Read the Room: Adapting a Robot's Voice to Ambient and Social Contexts

Emma Hughson, Angelica Lim

Robotic IntelligenceAudio

🎯 What it does: Researched and implemented the process of robots adjusting their voice style according to social and environmental contexts in different environments (such as quiet, dim, bright, or noisy), primarily validated in dining scenarios.

Real is Better than Perfect: Sim-to-Real Robotic System in Secondary School Education

Jiasi Gao, Guyue Zhou

Robotic Intelligence

🎯 What it does: Built a simulation-to-real robot system where students can optimize algorithms in a simulation environment and verify them in a remote physical laboratory, equipped with an automatic submit-test-reset subsystem to enable 24/7 testing support.

Real-time Dynamic Bipedal Avoidance

Tianze Wang, Christian Hubicki

OptimizationRobotic Intelligence

🎯 What it does: Proposes a real-time motion planning and multi-body control framework to enable dynamic biped robots to avoid collisions with multiple moving obstacles and automatically switch between stance and gait collision avoidance modes.

Real-Time Elevation Mapping with Bayesian Ground Filling and Traversability Analysis for UGV Navigation

Han Xie, Qiang Liu

Autonomous DrivingSimultaneous Localization and MappingPoint Cloud

🎯 What it does: Using sparse point cloud data from a single LiDAR, real-time generation of dense terrain height maps and traversability analysis is achieved, capable of distinguishing vertical obstacles, hanging objects, etc.

Real-Time Failure-Adaptive Control for Dynamic Robots

Jacob Hackett, Christian Hubicki

OptimizationRobotic Intelligence

🎯 What it does: Propose a real-time failure-adaptive control framework that can instantly learn failure probabilities and replan behaviors when dynamic robots (e.g., drones) encounter failures.

Real-Time Model-Free Deep Reinforcement Learning for Force Control of a Series Elastic Actuator

Ruturaj S. Sambhus, A. Leonessa

Robotic IntelligenceReinforcement Learning

🎯 What it does: On the Hardware Series Elastic Actuator (SEA) swing system, a Proximal Policy Optimization (PPO) algorithm is used to train a deep reinforcement learning (DRL) policy to achieve force tracking control with a frequency range of 0.05 Hz to 0.35 Hz and an amplitude of 50 N. The system is trained continuously for over 21 hours without human intervention, with safety mechanisms implemented during training to ensure hardware safety.

Real-Time Motion Planning Framework for Autonomous Vehicles with Learned Committed Trajectory Distribution

Minsoo Kim, Jaeheung Park

Autonomous Driving

🎯 What it does: Proposed a real-time motion planning framework that uses a neural network to predict a committed trajectory distribution as a sampling bias to achieve near-optimal continuous path planning;

Real-Time NMPC for an Automated Valet Parking with Load-Based Safety Constraints and a Path-Parametric Model

B. B. Carlos, Benoît Pelourdeau

Autonomous DrivingOptimization

🎯 What it does: Propose a real-time nonlinear model predictive control (NMPC) framework for an automatic valet parking system, aiming to reduce retrieval time and ensure vehicle and robot safety through acceleration constraints.

Real-Time Pose Estimation of Rats Based on Stereo Vision Embedded in a Robotic Rat

Xiaowen Guo, Qing Shi

Pose EstimationRobotic IntelligenceImage

🎯 What it does: A real-time rat pose estimation system based on stereo vision, designed with a lightweight high-resolution network RRKDNet for keypoint detection, and achieved pose reconstruction through stereo vision models and robot coordinate transformation, completing real-time simulation experiments.

Real-Time RRT* with Signal Temporal Logic Preferences

Alexis Linard, Jana Tumova

OptimizationRobotic Intelligence

🎯 What it does: In dynamic obstacle environments, Signal Temporal Logic (STL) constraints and preferences are integrated into the real-time rapidly exploring random tree (RT-RRT*) motion planning algorithm, and a cost function is proposed to guide the algorithm toward the most robust (i.e., STL-compliant) solution.

Real-Time Simultaneous Multi-Object 3D Shape Reconstruction, 6DoF Pose Estimation and Dense Grasp Prediction

Shubh Agrawal, Volkan Isler

Pose EstimationComputational EfficiencyRobotic IntelligenceImagePoint Cloud

🎯 What it does: Propose a real-time method to achieve multi-object 3D shape reconstruction, 6DoF pose estimation, and dense grasp point prediction from a single RGB-D image.

Real-Time Trajectory-Based Social Group Detection

Simindokht Jahangard, Hamid Rezatofighi

Object TrackingComputational EfficiencyRecurrent Neural NetworkGraph Neural NetworkGraphTime SeriesSequential

🎯 What it does: A real-time social group detection framework based on motion trajectories is proposed, treating individuals as nodes in a graph, using LSTM to encode trajectories, constructing edges based on inter-trajectory distances, and achieving group identification through an improved graph transformer and graph clustering loss.