Artificial intelligence based object detection and traffic prediction by autonomous vehicles – A review

Abstract

Autonomous vehicles (AV) are anticipated to have a significant positive impact on society by reducing accidents and optimizing traffic. However, they are primarily based on improvements in various artificial intelligence (AI) methods and processes. It’s critical to understand how AI functions in AV systems in order to attain the objective of full automation (self-driving). Several researchers have put a lot of work into examining various uses of AI in creating AV. However, only a limited number of articles provide a scientific examination of the methods currently used to integrate AI into AVs. This work offers a systematic review assessment to provide a clear picture of the current state of research in Explainable artificial intelligence (XAI) on AV safety and its theoretical background. Object detection by AVs is an important process that is part of several autonomous driving tasks, such as object tracking, trajectory prediction, and collision avoidance. This study investigates and uses object detection, semantic segmentation and traffic prediction with their layered architectures. Based on a review of current practices and technical innovations, this study also provides insights into future opportunities around using AI in conjunction with other emerging technologies. The applications are high definition (HD) maps related to big data and high-performance computing (HPC), test benches for augmented and virtual reality simulations, and AV with 5G communications. The results of the review are then summarized in the conclusion, together with recommendations for further research into the development of vehicles with a higher degree of automation.