Preface

Nowadays, artificial intelligence is playing an essential role in electronics, which demands potential innovations to enhance the performance and quality of digital applications. This book focuses on sensor technology and computer vision, where machine learning (ML) and deep learning (DL) are able to utilize input data and images for prediction, classification, and data visualization.

The initial chapters discuss the indepth research on data utilization in AI from various sensors, especially IMU (Inertial Measurement Unit), light detection and ranging (lidar), and radio detection and ranging (radar). IMU sensor is a common and powerful sensor providing motion data from accelerometers, gyroscopes, and magnetometers. With MEMS (Micro-electromechanical Systems) technology, the IMU sensors are compacted in a small size, with lower power consumption and high-quality factors. ML models handle these IMU data for process optimization, risk prevention, product improvement, fault diagnosis, human activity recognition, and automation. Furthermore, IMU data can be combined with Lidar and radar sensors to detect objects and navigate their surroundings in self-driving cars and AI robotic systems to avoid obstacles or pick up the demanded items. In addition, reinforcement learning algorithms play an important role in self-driving robots, together with simultaneous localization and mapping (SLAM) technology for high-resolution 3D maps of the environment.

On the other hand, computer vision has been developed for image recognition, motion tracking, and object classification. Many electronic devices can implement robust AI algorithms, such as convolutional neural networks (CNN), you only look once (YOLO), etc., to support healthcare and automated vehicle control. Moreover, deep learning also provides solutions for human pose estimation (HPE), which evaluates human posture to support people in rehabilitation.

After deep analysis and research on classification and computer vision, ML regression can be taken into account in terms of prediction uncertainty. The aim is to examine the uncertainty of deep neural network (DNN) prediction, specifically in MEMS IMU data in this case. From this study, we are able to have a profound view of ML applications for high-technology sensors.

The last chapter discusses the incorporation of ML into augmented reality (AR) in the automotive industry. AR adopts the existing real-world environment and transfers virtual information to the top, practically enhancing the car industry in terms of the design process, manufacturing, and customer experience. The techniques discussed in previous chapters will be linked to this part via AI applications in AR, such as object recognition, SLAM, HPE, gesture recognition, and DL models.

Based on the above contents, this book includes the following chapters:

1. Current State, Challenges, and Data Processing of AI in Sensors and Computer Vision.
2. Human Activity Recognition and Health Monitoring by Machine Learning Based on IMU Sensors
3. Reinforcement Learning in Robot Automation by Q-learning.
4. Deep Learning Techniques for Visual Simultaneous Localization and Mapping Optimization in Autonomous Robot
5. Deep Learning in Object Detection for the Autonomous Car
6. Human Pose Estimation for Rehabilitation by Computer Vision
7. Prediction Uncertainty of Deep Neural Network in Orientation Angles from IMU Sensors
8. Machine Learning in Augmentation Reality for Automotive Industry.

This book depicts the input data processing, AI model structure, training process, model test/validation, and final performance of the whole system in use. After reading this book, readers will comprehend the working principles, pros, and cons of AI technology in the highly trending topics of the scientific field.