Automotive Safety-Assisted Driving Technology Based on Computer Artificial Intelligence Environment

Abstract

A reasonable driving behavior decision model can choose the appropriate driving behavior according to the actual situation, thus improving the safety and efficiency of driving. To achieve an intelligent and humanized driving experience, this study explores the decision-making process behind driving behaviors. We have established a decision-making model for driving behaviors rooted in the finite state machine (FSM) paradigm. This model selects the most suitable driving action based on the car's current state, the surrounding environment, and the driver's intention. Given the intricate and varied nature of driving behaviors, we have incorporated a deep reinforcement learning (DRL) algorithm. This enables the optimization of decision-making strategies through dynamic interactions between the driver and the environment. Our findings reveal that this model adeptly handles complexities in real-world driving scenarios, thereby enhancing driving safety. In automotive contexts, FSM ensures the selection of apt driving actions aligned with the vehicle's status, environmental cues, and the driver's intentions. This innovative model surpasses traditional decision-making frameworks, paving the way for advancements in intelligent driving technology, and demonstrating remarkable adaptability and potential for further optimization. © 2024 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.