Behavior-Aware Hierarchical Decision Strategy for High-Speed Autonomous Driving

This paper introduces a method for driving style modeling and classification using a GMM model to capture the relationships among various features derived from driving data collection. Subsequently, principal component analysis is applied to reduce the dimensionality of driving features, a Bayesian approach is taken to develop a statistical probability model for driver characteristics, and an analysis of the model’s statistical features is conducted. The Gibbs sampling algorithm is employed to estimate the parameters of the Gaussian model, uncovering similar characteristics within the data and categorizing driver intentions. Using the driving model, high-speed following is simulated. By optimizing the decision-making and control layers, the algorithm’s efficiency is enhanced and the dimensionality of its parameters is reduced. The feasibility of the intelligent driving decision strategy developed in this study was validated using the Highway-env simulation platform.