Research on Personalized Predictive Cruise Control Based on Driving Style and Driving Intention Identification

With the advancement of intelligent driving, research on predictive cruise control (PCC) has been increasing. However, existing studies predominantly focus on predictive cruise control based on the information of the vehicle itself. This paper studied the PCC system in relation to driving style and driving intention. First, the driving style was recognized through the implementation of Inverse Reinforcement Learning (IRL), followed by the prediction of the driving intention using a bi-directional long short-term memory-CNN (BILSTM-CNN) model. Subsequently, the predictive cruise control function was implemented using MPC, the safe following distance was determined based on the driving style, and the acceleration range was constrained in consideration of the driving intention. The safety and cruising performance of the system were validated through changes in relative speed and relative distance. Fuel consumption was reduced by 12.1% compared to the adaptive cruise control system, demonstrating the excellent fuel economy performance of our developed PCC system. Our PCC control strategy not only maintains good followability under various driving conditions, including cruising, acceleration, and deceleration, but also caters to different drivers' personalized needs by considering their driving style and intention.