Evaluation and optimization of adaptive cruise control in autonomous vehicles using the car learning to act simulator: A performance evaluation under various weather conditions

This is the authors' abstract. We don't add key points for this paper.

Adaptive Cruise Control (ACC) can automatically change the speed of the ego vehicle to maintain a safe distance from the following vehicle. The primary purpose of this study is to use cutting-edge computing approaches to locate and track vehicles in real-time under various conditions to achieve an efficient ACC. This research examines the extension of ACC by employing depth cameras, Radio Detection and Ranging (radar) sensors, Global Positioning System (GPS), and collision sensors within Autonomous Vehicles (AV) to respond in real-time to the changing weather conditions using the Car Learning to Act (CARLA) simulation platform. The ego vehicle uses a proportional-integration-derivative (PID) controller to decide whether to accelerate or decelerate depending on the speed of the leading (ahead) vehicle and its safe distance from the vehicle. The simulation shows that controlling autonomous vehicles reduces the speed of the leading vehicle and ego vehicle when it rains, especially at night. In addition, longer travel times were observed for both vehicles under rainy conditions than under dry conditions at night in towns 4 and 10. In addition, PID control prevents the leading vehicles from rear collisions. However, the spacing between the ego vehicle and the leading vehicle was not maintained in the safe zone in either town. Therefore, this study used the Hill Climb (HC) and Tabu Machine Learning (ML) algorithms to optimize the speed of the ego vehicle to maintain a safe spacing distance between it and the leading vehicle by training and developing CARLA using a Bayesian Network (BN).