Clustering and Analysis of the Driving Style in the Cut-in Process

For a long period, autonomous vehicles (AVs) and human-driven vehicles (HDVs) need to share roads in mixed traffic flow, where the cut-ins of the HDVs towards the AVs can frequently occur. To better understand and address the cut-in behavior, it is crucial to comprehend the driving style of this behavior. Thus, this paper investigates how to classify and analyze the driving style of the cut-in process. The features of the driver behavior and driving context are selected from the speed-change and lane-change phases of the cut-in process. The principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE) methods are employed to reduce the dimensionality of the features. The k-means++ algorithm is applied to cluster the driving style of the cut-ins. To acquire the cut-in data with different driving styles, driver-in-the-loop experiments were conducted with eight subjects in two classes of cut-in scenarios. The clustering results show that the t-SNE method outperforms the PCA method and the best clustering performance is achieved when the number of clusters is set to three. Based on the clustering results, a statistical analysis is conducted to illustrate the characteristics of three different cut-in driving styles, i.e., aggressive, normal, and conservative.