AI-enabled Technologies for Autonomous and Connected Vehicles

Preface
Contents
Advances, Opportunities and Challenges in AI-enabled Technologies for Autonomous and Connected Vehicles
1 Introduction
2 Autonomous Vehicles: Current Technologies and Challenges
2.1 Levels of Autonomy
2.2 AI Technologies in Autonomous Vehicles
3 Connectivity and Mobility
3.1 Mobility Research
3.2 Prediction Model for V2X Communications
3.3 Big Data Research in Transportation
3.4 Automotive Cybersecurity
References
Sensors and Perception
Semi-autonomous Truck Platooning with a Lean Sensor Package
1 Overview
2 Literature Survey
3 Auburn Platooning System
3.1 Dynamic Base Real-Time Kinematic Positioning (DRTK)
3.2 Delphi Electronically Scanning Radar (ESR)
3.3 Dedicated Short Range Communications
3.4 Sensor Fusion
4 Testing Campaign
5 Results of Sensor Impairment
5.1 Effect of a Faulty Radar on Platooning
5.2 Effect of a Degraded GPS on Platooning
5.3 Effect of Radio Interference
5.4 Summary
6 Conditional Effects
6.1 Occlusions
6.2 Rain
6.3 Antenna Position
6.4 RF Interference
6.5 GPS Outage
6.6 Road Curvature
6.7 Grade
7 Conclusion
References
Environmental Perception for Intelligent Vehicles
1 Sensors
1.1 Development of Sensors in Intelligent Autonomous Vehicles
1.2 Camera
1.3 LiDAR
1.4 Radar
1.5 Future of Sensors in Intelligent Vehicles
2 Data Restoration
2.1 RGB Image Restoration
2.2 LiDAR Point Cloud Restoration
3 Semantic Segmentation
3.1 Semantic Segmentation for RGB Images
3.2 Semantic Segmentation for RGB-D Images
3.3 Semantic Segmentation for LiDAR Point Cloud
4 Object Detection
4.1 2D Object Detection
4.2 2D Object Detection of Fisheye Camera
4.3 3D Object Detection
5 Object Tracking
5.1 Object Tracking for RGB Images
5.2 Object Tracking for Point Cloud
6 Simultaneous Localization and Mapping
6.1 SLAM Overview
6.2 2D Visual Location and Mapping
6.3 3D Visual Location and Mapping
7 Multi-sensor Fusion
7.1 Multi-sensor Fusion Overview
7.2 LiDAR and Camera Fusion
7.3 LiDAR and Radar Fusion
7.4 Radar and Camera Fusion
References
3D Object Detection for Autonomous Driving
1 Introduction
2 Recent Advance in 3D Object Detection
2.1 Cloud Point-Based Method
2.2 Image-Based Method
3 Descriptor Enhanced Stereo R-CNN
3.1 Overview
3.2 Coarse 3D Box Estimation
3.3 Unsupervised Learning of the Local Descriptor
3.4 Descriptor-Enhanced 3D Box Alignment
4 Experiments
4.1 Implementation Details
4.2 Comparison with State-of-the-Art
4.3 Multi-class Comparison with 3DBBX
4.4 Ablation Study
4.5 Runtime
5 Conclusion and Future Work
References
Comparative Study on Transfer Learning for Object Classification and Detection
1 Introduction
2 State-of-the-Art Review in DNNs and Transfer Learning
3 Architecture and Characteristics of CNN Models
3.1 CNN Models for Object Classification
3.2 NN Models for Object Detection
4 Transfer Learning for Object Classification and Detection
4.1 Transfer Learning for Object Classification
4.2 Transfer Learning for Object Detection
5 Conclusion
References
Future Technology and Research Trends in Automotive Sensing
1 Introduction
2 Advancements in Radar and Lidar Sensing
3 Toward Energy Efficient Edge Computing via Optical Advances
4 Conclusion
References
Automated Driving Decisions and Control
Robust AI Driving Strategy for Autonomous Vehicles
1 Introduction
2 Decision Making: DRL Driving Strategy for Changing Lanes
2.1 Reinforcement Learning and Deep Reinforcement Learning an Introduction
2.2 DRL for Autonomous Driving
2.3 Vehicle Dynamics
2.4 Simulation Results
2.5 Summary
3 Executing DRL Decision with Motion Control Algorithm
3.1 Longitudinal Motion Control
3.2 Lateral Motion Control
3.3 Summary
4 Generic Safety Filter Design with Control Barrier Functions
4.1 Control Barrier Functions
4.2 Calculation of Barrier Constraints
4.3 Contextual Selection of Decoupled CBF
4.4 Examples
4.5 Summary
5 Integrated Driving Policy with DRL, Motion Control, and CBF Safety Filter
5.1 Training Architecture with CBF Safety Filter
5.2 Summary
6 Summary and Conclusion
References
Artificially Intelligent Active Safety Systems
1 Introduction
1.1 SAE Definitions
2 Active Safety Technology
2.1 Collision Warning
2.2 Collision Intervention
2.3 Driving Control Assistance
2.4 Parking Assistance
2.5 Other Driving Assistance
2.6 Beyond Assistance
3 Active Safety Potential
4 Systems on the Road Today
4.1 Methods
4.2 Results
4.3 Discussion
4.4 Conclusion
5 Promising AI Applications
5.1 Deep Learning
5.2 Reinforcement Learning
6 Final Thoughts
References
Model Predictive Control for Safe Autonomous Driving Applications
1 Introduction
1.1 Notation
2 Problem Description
3 Model Predictive Flexible Trajectory Tracking Control
4 Tracking an Infeasible Reference
5 Safety-Enforcing MPC
6 Simulations
6.1 ISS: Stability with Infeasible Reference
6.2 MPFTC: Ensuring Safety of the Controller
7 Conclusions
References
Energy-Efficient Autonomous Driving Using Cognitive Driver Behavioral Models and Reinforcement Learning
1 Introduction
2 Lane Changes for Energy-Efficient AV Driving
3 Powertrain Modeling for Battery Electric Vehicles
3.1 Model Description
3.2 Model Calibration and Validation
4 Controller Design
4.1 Game-Theoretic Traffic Environment
4.2 Observation and Action Spaces
4.3 Reward Function
4.4 Training Algorithm
4.5 Training Process
4.6 Autonomous Vehicle Control Policy for Benchmarking
5 Results
5.1 Training for RL-Based Policies
5.2 Control Performance
6 Conclusions
References
Self-learning Decision and Control for Highly Automated Vehicles
1 Introduction
2 Scalability
3 Performance
4 Interpretability
5 Mixed Model
6 Emergency Handling
7 Conclusion
References
Advanced Driver Assistant Systems
MAGMA: Mobility Analytics Generated from Metrics on ADAS
1 Introduction
2 Constraints
2.1 Data Collection Costs
2.2 Data Generation and Collection Issues
3 Determination of Actual Feature Experience
4 Determination of Expected Feature Experience
5 ADAS Feature Customer Experience Metrics
5.1 Binary Feature Availability
5.2 Nuanced Feature Availability
5.3 Clustering for Outlier Discovery
6 Conclusion
References
Driver Assistance Systems and Safety—Assessment and Challenges
1 Introduction
2 The Scenario Approach
2.1 The Idea
2.2 Elements of a Scenario Based Evaluation
3 Scenario Generation and Selection
3.1 Scenario Generation
3.2 Scenarios from Crash Databases
3.3 Automated Scenario Catalogue Learning
4 Scenario Parametrization
4.1 Numerical Assessment Methods
5 Representation of the Surrounding Traffic
5.1 Trajectory Prediction
5.2 Threat Assessment
6 Metrics of Risk
7 Outlook
7.1 Induced Effects on Other Vehicles
7.2 Shields and Emergency Systems
References
Factors Influencing Driver Behavior and Advances in Monitoring Methods
1 Introduction
2 Factors Influencing Driver Behavior
2.1 Types of Human Factors
2.2 Types of Environmental Factors
2.3 Driver Profile
3 Driver Behavior Monitoring Methods
3.1 Intrusive Measuring Methods
3.2 Camera Based Measuring Methods
3.3 Ergonomic and Body Posture Based Measuring Methods
3.4 Vehicle Dynamics Based Measuring Methods
3.5 Hybrid Measuring Methods
4 Smart Detection Algorithms
4.1 Supervised Classification
4.2 Unsupervised Classification
5 Deep Learning Neural Network Classifiers
5.1 Supervised LSTM Classifier
5.2 Unsupervised LSTM Classifier
6 Summary
References
Connected Autonomous Vehicles, Mobility, and Security
Towards Learning-Based Control of Connected and Automated Vehicles: Challenges and Perspectives
1 Introduction
2 Multi-Agent Coordination Problem
3 Modeling Agent Dynamics
3.1 Double Integrator Model
3.2 Bicycle Model
3.3 Further Extensions
3.4 Challenges and Perspectives
4 Optimal Agent Coordination
4.1 Local Objectives and Constraints
4.2 Collision Avoidance
4.3 Centralized Optimal Control Problem
4.4 Distributed Solution
5 Towards Learning-Based Control
5.1 Model Uncertainties
5.2 Mixed-Traffic Scenarios
6 Conclusion
References
Virtual Rings on Highways: Traffic Control by Connected Automated Vehicles
1 Introduction
2 Traffic Control by Connected Automated Vehicles
2.1 Simplified Models for Longitudinal Vehicle and Traffic Dynamics
2.2 Vehicle Control Influencing Traffic
2.3 Traffic Control
3 Benefits of Connectivity
3.1 Simulation Results
3.2 Energy Efficiency
4 Dynamics of Traffic Flow
4.1 Linearized Dynamics
4.2 Transfer Functions
5 Stability of Traffic Flow
5.1 Stability Conditions
5.2 Relationship of the Stability Conditions
6 Conclusions
References
Socioeconomic Impact of Emerging Mobility Markets and Implementation Strategies
1 Introduction
2 Theoretical Preliminaries
2.1 An Introduction to Mechanism Design
2.2 The Vickrey-Clarke-Groves Mechanism
3 The Emerging Mobility Market
3.1 Mathematical Formulation of the Emerging Mobility Market
3.2 The Optimization Problem Statement of the Emerging Mobility Market
4 Methodology for the Design of Mobility Incentives
5 Properties of the Mobility Market
6 Conclusion
References
A Real-Time Seq2Seq Beamforming Prediction Model for C-V2X Links
1 Backgrounds
2 System Model
3 Implementation and Field Test
4 Real-Time Seq2Seq Beamforming Prediction Model
5 Conclusion
References
Big Data in Road Transport and Mobility Research
1 Introduction
2 Big Data and Methods in Transportation Research
2.1 Big Data Sources
2.2 Methods
3 Transportation Research Using Big Data
3.1 Use of AI/ML to Add to Information in Large Transportation Datasets
3.2 Use of Sensors and Algorithms to Obtain Safety-Relevant Data at Large Scales
3.3 Machine Learning and Artificial Intelligence in Safety Research
3.4 Travel Demand Estimation
3.5 Route Choice Models
4 Pros and Cons of Big Data and ML/AI
4.1 Big Data Challenges
4.2 Going Forward
References
Machine Learning for Automotive Cybersecurity: Challenges, Opportunities and Future Directions
1 Introduction
2 Security Analysis of CAVs
3 In Vehicular Networks (IVN)
3.1 LIN Protocol
3.2 MOST Protocol
3.3 FlexRay Protocol
3.4 CAN Bus Protocol
3.5 Ethernet Protocol
4 Security Analysis of IVN Architecture
4.1 LIN
4.2 FlexRay
4.3 CAN (Controller Area Network)
4.4 Ethernet
5 Machine Learning as Defence Mechanism
5.1 CAN Data Acquisition
5.2 Data Processing
5.3 Algorithms to Detect a CAN Bus Attack
5.4 Challenges of ML in IVN Research
5.5 Future Opportunities in This Domain
6 Conclusion
References