Enhancing Safety in Autonomous Vehicles through Advanced AI-Driven Perception and Decision-Making Systems

Abstract

Accidents, congested roads, consumption of energy, and emissions may all decrease significantly with the rise of self-driving cars, providing a promising response to social and environmental issues. In this study, researchers investigate the way the integration of innovative AI-driven perception and decision-making systems into AVs impacts safety. AVs have the potential to change transportation by reducing accidents caused mainly by human error. They may operate on their own or in conjunction with human drivers. The primary goal is to investigate and enhance AV safety by creating highly sophisticated perception and decision-making technologies driven by machine learning. Pragmatism research philosophy, experimental research design, and inductive approach serve as the selected methods of the study. In addition, secondary qualitative data analysis methods help evaluate the entire study. The outcomes of the study demonstrated that the advancement of autonomous vehicles depends significantly on the creation of AI-driven sensors. “Vehicle-to-vehicle (V2X) networks” for communication and safety features increase security, while integrated camera systems with acoustic and thermal sensors improve sensing capacities. Deep learning techniques, especially “convolutional neural networks (CNN)” and “fully convolutional networks (FCN)”, facilitate accurate object recognition and segmentation. Vehicles’ ability to handle difficult road conditions is further improved by a real-time risk evaluation and trajectory planning based on human-like behavioral modeling.