Robot Intelligence Technology and Applications 6: Results from the 9th International Conference on Robot Intelligence Technology and Applications

✍ Scribed by Jinwhan Kim, Brendan Englot, Hae-Won Park, Han-Lim Choi, Hyun Myung, Junmo Kim, Jong-Hwan Kim

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 618
• Series: Lecture Notes in Networks and Systems, 429
• Category: Library

✦ Synopsis

This book aims at serving the researchers and practitioners in related fields with a timely dissemination of the recent progress on robotics and artificial intelligence. This book is based on a collection of papers presented at the 9th International Conference on Robot Intelligence Technology and Applications (RiTA), held at KAIST in Daejeon, Korea, in a hybrid format, on December 16–17, 2021. Humankind is getting through the third year of COVID-19 pandemic. While this pandemic has made everyone’s life so challenging, it has also expedited transition of our everyday lives into a new form, often called “the new normal.” Although many people often use the terminology, perhaps we still do not have a consensus about what it is and what is should be like. One thing that is clear is that robotics and artificial intelligence technologies are playing critical roles in this phase transition of our everyday lives. We see last-mile delivery robots on the street, AI-embedded service robots in the restaurants, uninhabited shops, non-face-to-face medical services, conferences and talks in metaverses and AI-based online education programs. For better readability, the total of 53 papers are grouped into four chapters: Chapter I: Motion Planning and Control; Chapter II: Design and Robot Application; Chapter III: Sensing, Perception and Recognition; and Chapter IV: Cognition, Autonomy and Intelligence. For those who have research on robot intelligence technology, we believe this book will help them understand the recent robot technologies and applications and enhance their study.

✦ Table of Contents

Preface
Contents
Motion Planning and Control
Design for a Dual-Redundant Flight Control Computer Using a Model Predictive Voting Algorithm
1 Introduction
2 Related Work
3 Methodology
3.1 Dual-Redundant Flight Control Computer System Design
3.2 Model Predictive Voting Algorithm (MPVA)
4 Simulation Results
5 Conclusions
References
TPL: Trajectory Planner for Target Tracking in Low-Light Environments
1 Introduction
2 Related Works
3 Methodology
3.1 System Overview
3.2 Waypoints Generation with Object Detection
3.3 Trajectory Construction
3.4 Tracking Controller
4 Experiment
4.1 Validation of Trajectory Planner and Controller
4.2 Real-World Experiment
5 Result and Discussion
6 Conclusion
References
Design of Foothold Decision Model in Convex Model Predictive Control for Legged Robots
1 Introduction
2 Dynamics and Control
2.1 3D Single Rigid Body Hybrid Model
2.2 Variation-Based Linearization
2.3 Vectorization
2.4 Model Predictive Control
3 Simulational Results
3.1 Comparative Simulation
3.2 Challenging Tasks
4 Conclusion
References
Learning Footstep Planning for the Quadrupedal Locomotion with Model Predictive Control
1 Introduction
1.1 Model Predictive Control
1.2 Reinforcement Learning
2 Methodology
2.1 Overview
2.2 Nonlinear Model Predictive Control
2.3 Learning Footstep Planning Policy
3 Result
3.1 Environment Setup
3.2 Comparison with Heuristic Planner
4 Conclusion
References
Study on Geometry Based Collision Avoidance for Formation Flight by Redefinition of Safety Radius
1 Introduction
2 Guidance Law of Formation Flight
3 Collision Avoidance for Formation Flight
3.1 Obstacle Detection and Estimation
3.2 Redefinition of Safe Radius
3.3 Collision Detection and Avoidance
4 Experiment
4.1 System Configuration
4.2 Flight Experiment
5 Conclusion
Reference
Socially Acceptable Route Planning and Trajectory Behavior Analysis of Personal Mobility Device for Mobility Management with Improved Sensing
1 Introduction
2 Socially Acceptable Global Path Analysis for PMD
2.1 Map Visualization of Shared Space in an Area
2.2 Traffic Flow and Betweenness Centrality (BC) for Finding Out Major Links
2.3 Design Criteria for Selecting Social Acceptable Route
2.4 Cost Function of Links for Global Path Search
2.5 Global Path Analysis
3 Socially Acceptable Local Path Management for PMD
3.1 Modelling of Socially Acceptable Local Navigation
3.2 Local Navigation Analysis
4 Conclusion
References
Automatic Parameterization of Motion and Force Controlled Robot Skills
1 Introduction
2 Related Work
2.1 Contact Skill Learning
2.2 Hybrid Position-Force Control
3 Skill Architecture
3.1 Skill Attributes
3.2 Skill Parameters
4 Skill Implementation
4.1 Slide
4.2 Contour
5 Skill Execution
6 Experiments
6.1 Experimental Setup
6.2 Results
7 Conclusions and Future Work
References
Hierarchical Fault Tolerant Control of a Hexacopter UAV Against Actuator Failure
1 Introduction
2 Problem Formulation
2.1 Hexacopter Dynamic Model
2.2 Problem Statement
3 Hierarchical Fault Tolerant Controller Design
3.1 Sliding Mode Controller
3.2 Adaptive Sliding Mode Controller
3.3 Proportional-Integral-Derivative Controller
3.4 Control Allocation
3.5 Hierarchical Structure of Controller
4 Numerical Simulation
5 Conclusion
References
Neural Network Self Tuning PI Control for Thin McKibben Muscles in an Antagonistic Pair Configuration
1 Introduction
2 Thin McKibben Muscle Biceps-Triceps Platform
3 Controller Design
3.1 Neural Network Self-tuning Algorithm
4 Results and Analysis
5 Conclusion
References
Improved Robustness Analysis of Reinforcement Learning Embedded Control Systems
1 Introduction
2 Robustness Analysis
2.1 Equilibrium Point Perturbation
2.2 Linearized System at xeq+
2.3 Robustness Analysis at xeq+
3 Example: Inverted-Pendulum Stabilisation
3.1 Dynamics and DDPG
3.2 Linearization
3.3 Worst Perturbation with Usual Linearization (6)
3.4 Robustness Analysis with Improved Linearization (17)
4 Conclusions
References
Optimal and Adaptive Control Design Using Recursive Least Square with a New Exponential Forgetting Factor
1 Introduction
2 Optimal Control Problem and Reinforcement Learning
2.1 Discrete Time Optimal Control Problem
2.2 Reinforcement Learning and Optimal Control Problem
2.3 Function Approximation
3 Recursive Least Square with New Exponential Forgetting Factor
3.1 New Exponential Forgetting Factor Algorithm
3.2 Temporal Difference Learning Based Optimal Control Using Recursive Least Square
4 Numerical Simulation
5 Conclusion
References
Mobility Hierarchy and Simulation of a Modular, Reconfigurable, Tetrahedral Robot System
1 Introduction
1.1 Assumptions
1.2 Simulated Behavior
2 Movement
2.1 Single Unit
2.2 Multiple Units
3 Environment with Obstacle
3.1 Timed Trials
4 Summary
5 Future Work
References
Modeling Vestibular Afferents for Neuromorphic Sensing and Eye Movement Control
1 Introduction
2 Methods
2.1 Human Afferent Model
2.2 Circuit Model Implementation
3 Results
3.1 Afferent Results
3.2 VOR
4 Conclusion
References
Design and Robot Applications
Development of a Robotic Cane for Mild Locomotion Assistance
1 Introduction
2 Implementation
2.1 Model of the Robotic Cane
2.2 Robot Components
2.3 Robot Prototype
3 Tests and Results
4 Conclusions
References
2-DOF Rolling Joint with the Novel Interior Reinforcement Structures
1 Introduction
2 Simplified 2-DOF Rolling Joint Model
3 Modified Internal Structures
3.1 Link Chain Configuration
3.2 The Novel Interior Reinforcement Structure
4 Joint Design
5 Experiments
6 Conclusion
References
A Novel Approach for Reducing Actuators in Soft Continuum Robots and Manipulators
1 Introduction
1.1 SCR&M Body and Cross-Section
1.2 SCR&M Actuators
2 Current SCR&M Configurations
3 Actuator Reduction Technique
3.1 Tip Path Model
3.2 Planar Workspace
3.3 Volumetric Workspace
4 Conclusion
References
Designing a Contactless Office-Assistant Robot Using User Differentiation
1 Introduction
2 System Design Methodology
2.1 Design Requirements
2.2 Overall System
2.3 Face Recognition Implementation
2.4 Tracking Implementation
2.5 Use of Secondary Display
2.6 Speech Recognition (STT)/Voice Synthesis (TTS) Implementation
2.7 Task Manager Implementation
3 Evaluation and Results
3.1 Demonstration Video
3.2 Verification of Robot System Design
4 Conclusion and Future Work
References
Designing an Interactive Indoor Delivery Robot and Its Implications
1 Introduction
2 Overall System
3 Design of Navigation System
3.1 Map Building in a Hallway
3.2 Using Lidar Sensor in a Window-Wall Space
4 Design of HRI System
5 Evaluation and Result
5.1 Verification of Navigation
5.2 Verification of HRI
6 Conclusion
References
Digital Twins Development of Automatic Storage and Retrieval Station in a Production Line and an Integrated Robotic Manipulator
1 Introduction
2 Methodology
2.1 Research Methodology
2.2 Validation Approach
2.3 System Architecture
2.4 Design and Implementation
3 Results and Discussion
4 Conclusion and Future Work
References
Error Models and Position Estimations of PRPaR Mechanisms
1 Introduction
2 Inverse Kinematic
3 Error Model
3.1 Kinematic Error Model
3.2 Interpolation Model
4 Experiment Process
4.1 Error Measure and Line Segment Distinction
4.2 Kinematic Error Model
4.3 Interpolation Model
4.4 Combine Two Models
5 Experimental Results and Discussion
5.1 Kinematic Error Model and Interpolation Model
5.2 Combine Two Models
6 Conclusion
References
Calibration Algorithm of Cam Flat-Faced Follower Mechanism
1 Introduction
2 Cam Mechanism Working Principle
3 Manufacturing Tolerance and Calibration Algorithm
4 Results and Discussion
5 Conclusion
References
Mobile-Based Motorized Cutter Mechanism for Palm Oil Fresh Fruit Bunch Harvesting
1 Introduction
2 System Design and Fabrication
3 Fabrication of Motorized Cutter Mechanism
4 Experimental Results
5 Conclusion
References
Sensing, Perception and Recognition
Towards Autonomous Pipeline Inspection with Hierarchical Reinforcement Learning
1 Introduction
2 Background
2.1 Reinforcement Learning
3 Related Work
4 Methodology
4.1 Reinforcement Learning for the PIRATE Robot
4.2 Hierarchical RL and Policy Decomposition
5 Experimental Design
5.1 Autonomous Navigation in Pipeline Networks
6 Results and Discussions
6.1 Different Observation Groups
6.2 Comparison of the RL and HRL Performances
6.3 Comparison with Human Control
6.4 Transferring the Policies to the Real Robot
7 Conclusions
References
U-VIO: Tightly Coupled UWB Visual Inertial Odometry for Robust Localization
1 Introduction and Related Works
1.1 Multi-sensor Fusion-based Localization
1.2 UWB Based Localization
1.3 Graph Structure's Loop Closing
1.4 Contributions
2 UWB Visual-Inertial Odometry
2.1 Definition of State Variables
2.2 UWB Visual-Inertial Odometry Front-end
2.3 UWB Visual-Inertial Odometry Back-end
3 Experimental Results
3.1 Experimental Setup
4 Conclusion
References
Point and Line Feature-Based VIO for Mobile Devices
1 Introduction
2 Factor Graph Formulation
2.1 Cost Function
2.2 Prior Factor
2.3 IMU Factor
2.4 Point Factor
2.5 Line Factor
3 Implementation
4 Experiments
4.1 Indoor Experiment
4.2 Large Dataset Experiment
5 Conclusion
References
Deep Learning Based Parking Slot Detection and Tracking: PSDT-Net
1 Introduction
2 Proposed Algorithm
2.1 Parking Slot Detector
2.2 Parking Slot Tracker
3 Experiment
3.1 Experiment Framework
3.2 Quantitative Performance
3.3 Effect of Auxiliary Marking Points
3.4 Overall Performance
4 Conclusion
References
Robust Lane Detection by Cascaded Deep Learning Network
1 Introduction
2 Related work
2.1 Lane Detection
2.2 Spatio-Temporal Deep Learning Network
3 Proposed Approach
3.1 Cascaded Lane Detection Framework
3.2 ENet-SAD
3.3 Supervised CRRN
4 Experiments
4.1 Datasets
4.2 Evaluation Metrics
4.3 Implementation Details
4.4 Results
5 Conclusion
References
Elimination of Race Condition During GPU Acceleration of Probabilistic Height Map
1 Introduction
1.1 Probabilistic Height Map
1.2 Race Condition by GPU Acceleration
2 Algorithm for Eliminating Race Condition
2.1 The Number of Calculation Cycle
2.2 Problem Statement
2.3 Detailed Description of the Algorithm
2.4 Proof of Guarantee Condition
3 Result
4 Conclusion
References
Various-Level Spatio-Temporal Alignment for Cross-Domain Action Recognition
1 Introduction
2 Related Work
3 Various-Level Spatio-Temporal Alignment(VSTA)
4 Experiments
5 Conclusion
References
Revisiting Self-supervised Monocular Depth Estimation
1 Introduction
2 Related Works
2.1 Self-supervised Depth Estimation
2.2 Visual Representation Learning
3 Methodology
3.1 Learning Approaches
3.2 Pretrained Features (CNN Weights)
4 Experiments
4.1 Settings
4.2 Implementation Details
4.3 Learning Approaches
4.4 Pretrained Features
4.5 Scale Consistency Analysis
5 Conclusion
References
Feature-Aided SMC-PHD Filter for Nonlinear Multi-target Tracking in Cluttered Environments
1 Introduction
2 Related Work
2.1 Feature-Aided Data Association
2.2 Target Loss in Particle PHD Filters
3 Problem Formulation
3.1 Multi-target Representation and Simulation Setup
3.2 Target Features
3.3 Background: SMC-PHD Filter and Assumptions
4 Feature-Aided SMC-PHD for Increased Nonlinearity and Clutter Resilience
4.1 Augmenting the PHD of the Posterior with Feature Likelihood
4.2 Feature-Aided Likelihood
5 Adaptive FA-SMC-PHD: An Adaptive Sampling Scheme to Mitigate Degeneracy
5.1 Adaptive Importance Sampling Mechanism
5.2 Introducing Artificial Feature Measurements
6 Numerical Simulations
6.1 Simulation Setup
6.2 Intermediate Performance of the FA-SMC-PHD
6.3 Alleviating Degeneracy Using the Adaptive Sampling Mechanism
6.4 Adaptive FA-SMC-PHD Performance Assessment
7 Conclusions
References
Hand Gesture and Arm Movement Recognition for Multimodal Control of a 3-DOF Helicopter
1 Introduction
2 System Architecture
2.1 Hand Gesture Recognition System
2.2 3-DOF Helicopter Testbench
3 Motion Control System
3.1 Model of the Helicopter
3.2 Position Controller
3.3 HGR and Position Controller Integration
4 Experimental Results
5 Conclusions and Future Work
References
Data-Driven Preflight Diagnosis of Hexacopter Actuator Fault Based on Principal Component Analysis of Accelerometer Signals
1 Introduction
2 Fault Scenario and System Configuration
2.1 Actuator Fault Injection
2.2 System Configuration
3 Feature Extraction
3.1 Feature Extraction from Accelerometer Signals
3.2 Principal Component Analysis for Accelerometer (PCA)
3.3 Supervised Machine Learning
4 Fault Classification Results
5 Conclusion and Future Work
References
The Classification of Oral Squamous Cell Carcinoma (OSCC) by Means of Transfer Learning
1 Introduction
2 Methodology
3 Result and Discussion
4 Conclusion
References
The Diagnosis of Diabetic Retinopathy: An Evaluation of Different Classifiers with the Inception V3 Model as a Feature Extractor
1 Introduction
2 Methodology
3 Results and Discussion
4 Conclusion
References
Cognition, Autonomy and Intelligence
Deep Learning Based Real-Time Biodiversity Analysis Using Aerial Vehicles
1 Introduction
2 Related Research
3 Proposed Methodology
3.1 Hardware Implementation
3.2 Computational Pipeline Framework of YOLOv3 Model
3.3 Evaluation Parameters for the Framework
3.4 Lighter Versions of YOLOv3 Model
4 Experimental Results
4.1 Dataset Training
4.2 Model Training
4.3 Comparison of Algorithms
4.4 Test Results
5 Conclusion and Future Work
References
Field Friction Recognition and State Inference in AI Soccer
1 Introduction
2 Methods
2.1 Friction-Changing AI Soccer Environment and Data Collection
2.2 Friction State Classification
2.3 Autonomous State Inference
3 Experiments
3.1 Friction State Classification
3.2 Autonomous State Inference
4 Conclusion
References
MMH-GGCNN: Multi-Modal Hierarchical Generative Grasping Convolutional Neural Network
1 Introduction
2 Background
2.1 Grasp Representation
2.2 GG-CNN
3 MMH-GGCNN
3.1 Multi-modal Feature Encoding
3.2 Feature Decoding
3.3 Hierarchical Grasp Generation
4 Experiments
4.1 Experimental Setting
4.2 Quantitative Results
4.3 Qualitative Results
5 Conclusion
References
s-DRN: Stabilized Developmental Resonance Network
1 Introduction
2 Stablized Developmental Resonance Network
2.1 Global Weight Update
2.2 Node Activation
2.3 Template Matching
2.4 Template Learning
2.5 Node Grouping
2.6 Computational Efficiency Analysis
3 Experiments
3.1 Experiment Setting
3.2 Results and Analysis
4 Conclusion
References
Comparison of Deep Q-Learning, Q-Learning and SARSA Reinforced Learning for Robot Local Navigation
1 Introduction
2 Q-Learning
3 Sarsa
4 Deep Q-Learning (DQN)
4.1 Separate Target Network
4.2 Experience Replay
4.3 DQN Process Workflow
4.4 Network Architecture
5 Experimental Setup
5.1 Q-Learning and Sarsa Parameters
5.2 DQN Parameters
5.3 Training Setup
6 Results and Discussion
7 Conclusion
References
The Diagnostics of Osteoarthritis: A Fine-Tuned Transfer Learning Approach
1 Introduction
2 Methodology
2.1 Dataset
3 Results and Discussion
4 Conclusion
References
A Novel Reinforcement Learning Sampling Method Without Additional Environment Feedback in Hindsight Experience Replay
1 Introduction
2 Related Work
3 Background
3.1 Deep Deterministic Policy Gradient
3.2 Hindsight Experience Replay
3.3 Energy-Based Prioritization
4 Method
4.1 Trajectory Behaviour Principles
4.2 Trajectory Behaviour Sampling in Pushing Robotic Task
4.3 Trajectory Behaviour Sampling in PickAndPlace Robotic Task
5 Results
6 Discussion and Conclusion
References
Trajectory-Based Anomaly Classification of 6-DOF Guided Missile Using Neural Networks
1 Introduction
2 Problem Formulation
2.1 6-DOF Missile Simulator
2.2 Anomaly Definition
3 Methodology
3.1 Data Pre-processing
3.2 Neural Network Structures
4 Results
4.1 Simulation Environment
4.2 Training Model Setting
4.3 Simulation Results
5 Conclusion
References
Solving Delivery Assignment in Hybrid-Transit Network Using Multi-agent Reinforcement Learning
1 Introduction
2 Preliminaries
2.1 Task-Assignment Problem
2.2 Multi-agent Reinforcement Learning
3 Problem Formulation
4 Multi-agent Reinforcement Learning Model
4.1 Environment
4.2 State
4.3 Observation
4.4 Action
4.5 Reward
5 Evaluation
6 Discussion
7 Conclusion
References
A Novel Human-Robot Interaction Framework Based on Telegram and Programming by Demonstration
1 Introduction
2 Robot Programming by Demonstration
2.1 Task-Parameterized Gaussian Mixture Model (TP-GMM)
3 Telegram Bot-Based Interaction Framework
4 Experimental Results
5 Conclusion
References
A Review of Nonsmooth Systems for Social Robotics
1 Introduction
2 Literature Review
2.1 Social Robotics
2.2 A Summary of Nonsmooth Systems
3 Example: Nonsmooth Approach to Path Planning
4 Example: Social Norms
5 Conclusions
References
Towards Linguistic and Cognitive Competence for Socially Interactive Robots
1 Introduction
2 Literature Survey
3 Implementation
3.1 Speech Recognition and Speech Generation
3.2 Text Sentiment and Contexts
3.3 Linguistic Competences
3.4 Cognitive Competences
3.5 Workflow of the System Proposed
3.6 Emotion-Specific vs. Event-Specific Robot Behaviour
4 Experimentation and Evaluation
4.1 Experimental Setup
4.2 Evaluation
5 Conclusion
References
Understanding Professional Care Providers Readiness Towards the Adoption of Care Robots for Elderly Care
1 Introduction
2 Literature Review
2.1 Technology Acceptance Models in the Context of Care Robots for Elderly Care
3 Methodology
3.1 Study Design and Data Collection
3.2 Fuzzy Cognitive Maps
4 Results
5 Discussion
6 Conclusion
Appendix
References
VR Karaoke Using Expressive 3D Avatars
1 Introduction
2 Methodology
2.1 Implementation of Avatar and Natural Facial Expression Changes
2.2 Scale Analysis
2.3 Scale and Beat Scoring for Monitoring the User’s Singing Performance
2.4 Lyrics Scoring Algorithm
3 Evaluation and Result
4 Discussion
5 Conclusion
References
Real-Time Model Predictive Safety Assessment for Level 3 Autonomous Driving
1 Introduction
2 Model Predictive Safety Evaluation
2.1 Overview
2.2 Risks, Constraints, Feasibility and Safety
2.3 Model Predictive Safety Assessment
3 Dynamic Programming for Real-Time Processing
3.1 Real-Time Longitudinal Safety Assessment
3.2 Real-Time Lateral Safety Assessment
3.3 Discussion on Employed Control Policies and Scalability
4 Experimental Results
5 Conclusions
References
A Cooperative Protocol for Vehicle Merging Using Bi-dimensional Artificial Potential Fields
1 Introduction
2 Vehicle Modeling
2.1 Rear Axle Bicycle Model
2.2 Vehicle Merging Model
3 Controller Design
3.1 Potential Function Design
3.2 Lateral and Longitudinal Control
4 Simulation Results
5 Conclusions
References
Optimizing the Product Locations in Automated Cube Storage Warehouse Using an Enhanced Genetic Algorithm
1 Introduction
2 Problem Statement
3 Methodology
3.1 Representation
3.2 Operators
3.3 Fitness Function
4 Simulation and Result
4.1 Parameter Setting
4.2 Results
5 Conclusion
References
An Evasive Strategy for Safe Autonomous Navigation Using Bayesian Networks and CMA-ES
1 Introduction
2 Probabilistic Multi-Controller Architecture (P-MCA)
3 Evasive Strategy
3.1 Problem Statement
3.2 Single Hypothesis Evasive Prediction Profile
3.3 Multi-hypothesis Evasive Prediction Profile
3.4 Multi-risk Management
3.5 Multi-objective Function
3.6 Constraints Definition
3.7 Solving the Optimization Problem Based on CMA-ES
4 Simulation Results
4.1 Reacting to Multiple Anomaly Happening Simultaneously
4.2 Reacting to Multiple Anomaly Happening at Different Moments
5 Conclusion
References
Author Index