Artificial Intelligence for Digitising Industry: Applications

Description

This book provides in-depth insights into use cases implementing artificial intelligence (AI) applications at the edge. It covers new ideas, concepts, research, and innovation to enable the development and deployment of AI, the industrial internet of things (IIoT), edge computing, and digital twin technologies in industrial environments. The work is based on the research results and activities of the AI4DI project, including an overview of industrial use cases, research, technological innovation, validation, and deployment. This book’s sections build on the research, development, and innovative ideas elaborated for applications in five industries: automotive, semiconductor, industrial machinery, food and beverage, and transportation. The articles included under each of these five industrial sectors discuss AI-based methods, techniques, models, algorithms, and supporting technologies, such as IIoT, edge computing, digital twins, collaborative robots, silicon-born AI circuit concepts, neuromorphic architectures, and augmented intelligence, that are anticipating the development of Industry 5.0. Automotive applications cover use cases addressing AI-based solutions for inbound logistics and assembly process optimisation, autonomous reconfigurable battery systems, virtual AI training platforms for robot learning, autonomous mobile robotic agents, and predictive maintenance for machines on the level of a digital twin. AI-based technologies and applications in the semiconductor manufacturing industry address use cases related to AI-based failure modes and effects analysis assistants, neural networks for predicting critical 3D dimensions in MEMS inertial sensors, machine vision systems developed in the wafer inspection production line, semiconductor wafer fault classifications, automatic inspection of scanning electron microscope cross-section images for technology verification, anomaly detection on wire bond process trace data, and optical inspection. The use cases presented for machinery and industrial equipment industry applications cover topics related to wood machinery, with the perception of the surrounding environment and intelligent robot applications. AI, IIoT, and robotics solutions are highlighted for the food and beverage industry, presenting use cases addressing novel AI-based environmental monitoring; autonomous environment-aware, quality control systems for Champagne production; and production process optimisation and predictive maintenance for soybeans manufacturing. For the transportation sector, the use cases presented cover the mobility-as-a-service development of AI-based fleet management for supporting multimodal transport. This book highlights the significant technological challenges that AI application developments in industrial sectors are facing, presenting several research challenges and open issues that should guide future development for evolution towards an environment-friendly Industry 5.0. The challenges presented for AI-based applications in industrial environments include issues related to complexity, multidisciplinary and heterogeneity, convergence of AI with other technologies, energy consumption and efficiency, knowledge acquisition, reasoning with limited data, fusion of heterogeneous data, availability of reliable data sets, verification, validation, and testing for decision-making processes. Front Cover Artificial Intelligence for Digitising Industry Applications Dedication Acknowledgement Contents Preface Editors Biography List of Figures List of Tables List of Abbreviations Section 1 AI Automotive Abstract 1.0 AI Reshaping the Automotive Industry 1.0.1 Introduction and Background 1.0.2 AI Developments and Future Trends for AI Technologies 1.0.3 AI-Based Applications Acknowledgements References 1.1 AI for Inbound Logistics Optimisation in Automotive Industry Abstract 1.1.1 Introduction and Background 1.1.2 Requirements – User Journeys 1.1.3 Data Flow Principles and Architecture of the MPDSS 1.1.4 Preliminary Analysis of Data and Dataset 1.1.4.1 Data Pre-processing and Visualisation 1.1.4.2 Classification Models 1.1.5 Conclusion Acknowledgements References 1.2 State of Health Estimation using a Temporal Convolutional Network for an Efficient Use of Retired Electric Vehicle Batteries within Second-Life Applications Abstract 1.2.1 Retired Electric Vehicle Batteries for Second-Life Applications 1.2.2 State of Health of Lithium-Ion Batteries 1.2.3 Data Measurement Using the Open-Source BMS foxBMS 1.2.4 Temporal Convolutional Neural Network for State of Health Prediction 1.2.4.1 Causal Convolutions and Receptive Field 1.2.4.2 Dilated Convolutions 1.2.4.3 Residual Block 1.2.5 Results 1.2.6 Conclusion Acknowledgements References 1.3 Optimising Trajectories in Simulations with Deep Reinforcement Learning for Industrial Robots in Automotive Manufacturing Abstract 1.3.1 Introduction 1.3.2 Background 1.3.3 Methodology 1.3.4 Conclusion and Outlook Acknowledgements References 1.4 Foundations of Real Time Predictive Maintenance with Root Cause Analysis Abstract 1.4.1 Introduction and Background 1.4.2 Foundations 1.4.2.1 Model-based Diagnosis 1.4.2.2 Machine Learning Based Diagnosis 1.4.2.3 Artificial Neural Networks for Diagnostics 1.4.3 Related Research 1.4.4 Conclusion Acknowledgements References 1.5 Real-Time Predictive Maintenance – Model-Based, Simulation-Based and Machine Learning Based Diagnosis Abstract 1.5.1 Introduction and Background 1.5.2 Application of Diagnosis Systems Based on Simplified DC e-Motor Model 1.5.2.1 Simplified DC e-Motor Model With Fault Injection Capabilities 1.5.2.2 Model-based Diagnosis for Simplified DC e-Motor 1.5.2.3 Simulation-Based Diagnosis for Simplified DC e-Motor 1.5.2.4 Machine Learning for Diagnosis of Simplified DC e-Motor 1.5.2.5 Comparisons and Limitations 1.5.3 Conclusion Acknowledgements References 1.6 Real-Time Predictive Maintenance – Artificial Neural Network Based Diagnosis Abstract 1.6.1 Introduction and Background 1.6.1.1 AI-based Diagnosis of E-motors 1.6.1.2 Artificial Intelligence in Vibration Diagnosis 1.6.2 Artificial Neural Network for e-Motor Diagnosis 1.6.2.1 Acausal e-Motor Model with Faults Injection Capability 1.6.2.2 Artificial Neural Network for Inter-turn Short Circuit Detection 1.6.2.2.1 Selection of suitable condition indicator for fault detection 1.6.2.2.2 Network structure selection 1.6.2.2.3 Data pre-processing 1.6.2.2.4 Preparation of datasets 1.6.2.2.5 MNN training 1.6.2.2.6 MNN validation 1.6.2.2.7 MNN deployment 1.6.3 Artificial Neural Network based Vibration Diagnosis 1.6.3.1 Vibration Diagnosis of Rotating Machines 1.6.3.2 AI Approaches in Vibration Diagnosis 1.6.3.3 MLP implementable in device at the edge 1.6.4 Conclusion Acknowledgements References Section 2 AI Semiconductor 2.0 AI in Semiconductor Industry Abstract 2.0.1 Introduction and Background 2.0.2 AI Developments in Semiconductor Industry 2.0.3 Future Trends for AI Technologies and Applications in Semiconductor Industry 2.0.4 AI-Based Applications Acknowledgements References 2.1 AI-Based Knowledge Management System for Risk Assessment and Root Cause Analysis in Semiconductor Industry Abstract 2.1.1 Introduction and Background 2.1.2 Research Areas 2.1.2.1 FMEA and FMEA Consistency Improvement 2.1.2.2 Causal Information Extracting from Free Text 2.1.2.3 Failure Analysis Process, Failure Analysis Reports, and Ontologies 2.1.2.4 Knowledge Representation 2.1.2.5 Refinement Algorithm 2.1.3 Reflections 2.1.4 Conclusion Acknowledgments References 2.2 Efficient Deep Learning Approach for Fault Detection in the Semiconductor Industry Abstract 2.2.1 Motivation: The Wafer Fault Classification Problem 2.2.2 Related Works 2.2.3 Target Platform Requirements 2.2.4 HW/SW System and Methodology 2.2.4.1 Industrial HW/SW System for On-Device Inference 2.2.4.2 Neural Network Building and Training Using N2D2 2.2.4.3 Neural Network Export and FPGA Implementation Used for Inference 2.2.5 Conclusion Acknowledgements References 2.3 Towards Fully Automated Verification of Semiconductor Technologies Abstract 2.3.1 Introduction 2.3.2 Background: Interpreting Semiconductor Technologies 2.3.2.1 Methodology: The Integrated Analysis Process 2.3.2.2 Example Analysis: From the Image to the Feature Extraction 2.3.3 Conclusion Acknowledgements References 2.4 Automated Anomaly Detection Through Assembly and Packaging Process Abstract 2.4.1 Introduction and Background 2.4.2 Dataset Description and Defect Types 2.4.3 Methodology 2.4.3.1 Anomaly Detection 2.4.3.2 Pseudo Anomaly Detection 2.4.3.3 Convolutional Neural Networks 2.4.4 Results and Discussion 2.4.5 Conclusion and Outlooks Acknowledgment References Section 3 AI Industrial Machinery 3.0 AI in Industrial Machinery Abstract 3.0.1 Introduction and Background 3.0.2 AI Developments and Future Trends in Industrial Machinery 3.0.3 AI-based Applications Acknowledgements References 3.1 AI-Powered Collision Avoidance Safety System for Industrial Woodworking Machinery Abstract 3.1.1 Introduction and Background 3.1.2 Review of Industrial-level Methods for Edge DNNs 3.1.2.1 Compression Techniques 3.1.2.2 Popular Frameworks and Tools 3.1.3 Materials and Methods 3.1.3.1 System Architecture 3.1.3.2 Dataset Collection 3.1.3.3 Detection Methods 3.1.3.4 Continual Learning Setup 3.1.4 Experimental Results 3.1.4.1 Evaluation Metrics 3.1.4.2 Continual Learning Scenario 3.1.4.3 Robustness Against Quantization 3.1.4.4 Latency, Energy and Memory Footprint on STM32H743ZI 3.1.5 Conclusion Acknowledgements References 3.2 Construction of a Smart Vision-Guided Robot System for Manipulation in a Dynamic Environment Abstract 3.2.1 Introduction and Background 3.2.2 Challenges of Enabling Robots to “See” 3.2.2.1 Modularity 3.2.2.2 Operability 3.2.2.3 Computer Vision Algorithms 3.2.2.4 Validation of Algorithms 3.2.3 Requirements 3.2.4 Proposed Solution 3.2.4.1 Hardware and Interface Components 3.2.4.1.1 Robot Interface 3.2.4.1.2 Industrial Robot 3.2.4.1.3 3D Cameras 3.2.4.1.4 Deep Edge Device 3.2.4.2 Software Components 3.2.4.2.1 Computer Vision Algorithms 3.2.4.2.2 Synthetic Data Generation 3.2.4.2.3 Object 3D Reconstruction 3.2.4.2.4 Validation Framework 3.2.4.2.5 Robot Control 3.2.4.3 Hardware/Software Partitioning 3.2.5 Demonstrator Setup and Initial Results 3.2.6 Conclusion and Future Work Acknowledgements References 3.3 Radar-Based Human-Robot Interfaces Abstract 3.3.1 Introduction and Background 3.3.2 Gesture Recognition Using a Machine Learning Approach 3.3.2.1 Concept and Experimental Setup 3.3.2.2 Inference Pipeline, Training Algorithm Transmit Part Receive Part 3.3.2.3 Data Recording and Results Machine Learning Detector Statistics Machine Learning Real Time Inference 3.3.3 Gesture Recognition Using a Spiking Neural Network 3.3.3.1 Concept and Experimental Setup 3.3.3.2 Inference Pipeline, Training Algorithm 3.3.3.3 Data Recording and Results 3.3.4 Proof of Concept Demonstration 3.3.5 Comparison and Conclusion Acknowledgements References 3.4 Touch Identification on Sensitive Robot Skin Using Time Domain Reflectometry and Machine Learning Methods Abstract 3.4.1 Introduction and Background 3.4.2 State of the Art 3.4.3 Problem Definition 3.4.4 Concepts and Methods 3.4.5 Proof of Concept of the Novel Sensor System 3.4.5.1 Experimental Acquisition of Training Data 3.4.5.2 Training Procedure 3.4.6 Results 3.4.7 Conclusions Acknowledgements References Section 4 AI Food and Beverage 4.0 AI in Food and Beverage Industry Abstract 4.0.1 Introduction and Background 4.0.2 AI Developments in Food and Beverage Industry 4.0.3 Future Trends for AI Technologies and Applications 4.0.4 AI-Based Applications Acknowledgements References 4.1 Innovative Vineyards Environmental Monitoring System Using Deep Edge AI Abstract 4.1.1 Introduction 4.1.2 Related Work 4.1.3 Edge Intelligence 4.1.4 Communication Technology – LoRaWAN 4.1.5 Environmental Monitoring System 4.1.6 Conclusion Acknowledgements References 4.2 AI-Driven Yield Estimation Using an Autonomous Robot for Data Acquisition Abstract 4.2.1 Introduction 4.2.2 Artificial Intelligence for Grape Detection 4.2.3 Towards an Automated Protocol for Yield Estimation 4.2.4 Assessing the Vine Vitality Using an Embarked LiDAR 4.2.5 Conclusions Acknowledgements References 4.3 AI-Based Quality Control System at the Pressing Stages of the Champagne Production Abstract 4.3.1 Introduction and Background 4.3.2 Methodology 4.3.3 Results and Discussion 4.3.4 Conclusion Acknowledgements References 4.4 Optimisation of Soybean Manufacturing Process Using Real-time Artificial Intelligence of Things Technology Abstract 4.4.1 Introduction 4.4.2 Soybean Production Process Description 4.4.3 Overall Manufacturing System Architecture and Platform 4.4.4 Process Parameters Monitoring 4.4.5 Edge Processing and AI-based Framework for Real-Time Monitoring 4.4.6 Embedded Intelligent Vision and Multi-sensors Fusion Approach 4.4.6.1 Embedded Vision IIoT Systems Evaluation 4.4.7 Experimental Setup 4.4.7.1 Experimental Evaluation and Results 4.4.8 Summary and Future Work Acknowledgements References 4.5 AI and IIoT-based Predictive Maintenance System for Soybean Processing Abstract 4.5.1 Introduction 4.5.2 Maintenance Foundations 4.5.3 Principles of Predictive Maintenance 4.5.4 Soybean Production Process and Maintenance Policies 4.5.4.1 Vibration Analysis 4.5.5 AI-based Predictive Maintenance Framework Methodology 4.5.6 Industrial Integrated System for Soybean Production Equipment 4.5.7 Experimental Set-up and Implementation 4.5.8 Summary and Future Work Acknowledgements References Section 5 AI Transportation 5.0 Applications of AI in Transportation Industry Abstract 5.0.1 Introduction and Background 5.0.2 AI Developments in Transportation Industry 5.0.3 Future Trends for Applications in Transportation Industry 5.0.4 AI-Based Applications Acknowledgements References 5.1 AI-Based Vehicle Systems for Mobility-as-a-Service Application Abstract 5.1.1 Introduction and Background 5.1.2 AI-Based 3D Object Detection and Tracking for Automated Driving 5.1.2.1 Camera and LiDAR Sensor Data Fusion 5.1.2.2 Experiments and Results 5.1.2.3 Evaluation of the Algorithm and Vehicle Integration 5.1.3 Autonomous Control Prototyping in Simulated Environments 5.1.3.1 Reinforcement Learning Control for Mobile Vehicles 5.1.3.2 The Architecture – Immediate Actions Time-Horizon 5.1.4 Conclusion Acknowledgements References 5.2 Open Traffic Data for Mobility-as-a-Service Applications – Architecture and Challenges Abstract 5.2.1 Introduction and Background 5.2.2 Data Acquisition 5.2.2.1 Bus Traces 5.2.2.2 Traffic Cameras 5.2.2.3 Loop Detectors 5.2.3 Data Processing at the Edge 5.2.3.1 Object Detection 5.2.3.2 Bus GPS Trace 5.2.4 Data Processing in the Cloud 5.2.4.1 Data Quality Monitoring 5.2.4.2 Data Quality Observations 5.2.5 Conclusion Acknowledgements References List of Contributors Index Back Cover