A Generative Car-following Model Conditioned On Driving Styles

TL;DR

This work tackles the challenge of realistic car-following by capturing both inter-driver heterogeneity (different driving styles) and intra-driver dynamics (style changes over time). It introduces a time-varying IDM parameter framework to calibrate each driver’s evolving behavior and pairs it with a neural-process-based CF model that generates acceleration distributions conditioned on a learned driving-style latent. The aggressiveness index $H$, derived from time-varying IDM parameters, is mapped to the NP latent $r$ to provide an interpretable link between physical parameters and the generative style. Empirical results show improved calibration accuracy, stable CF predictions, and realistic trajectory generation for both observed and unseen driving styles, enabling more faithful traffic simulations and safer autonomous-vehicle planning. The approach supports generating traffic with mixed human and automated agents and can inform design of human-like controllers for AVs.

Abstract

Car-following (CF) modeling, an essential component in simulating human CF behaviors, has attracted increasing research interest in the past decades. This paper pushes the state of the art by proposing a novel generative hybrid CF model, which achieves high accuracy in characterizing dynamic human CF behaviors and is able to generate realistic human CF behaviors for any given observed or even unobserved driving style. Specifically, the ability of accurately capturing human CF behaviors is ensured by designing and calibrating an Intelligent Driver Model (IDM) with time-varying parameters. The reason behind is that such time-varying parameters can express both the inter-driver heterogeneity, i.e., diverse driving styles of different drivers, and the intra-driver heterogeneity, i.e., changing driving styles of the same driver. The ability of generating realistic human CF behaviors of any given observed driving style is achieved by applying a neural process (NP) based model. The ability of inferring CF behaviors of unobserved driving styles is supported by exploring the relationship between the calibrated time-varying IDM parameters and an intermediate variable of NP. To demonstrate the effectiveness of our proposed models, we conduct extensive experiments and comparisons, including CF model parameter calibration, CF behavior prediction, and trajectory simulation for different driving styles.

Figures (7)

• Figure 1: The training process of the NP-based CF model.
• Figure 2: NP-based CF model.
• Figure 3: (a) Histogram of $\tilde{r}$; (b) histogram of $H$; and (c) relationship between $\tilde{r}$ and aggressiveness index $H$.
• Figure 4: Comparison of the calibration and prediction RMSE of spacing using different methods.
• Figure 5: An example of calibrated time-varying IDM parameters and their Pearson correlation coefficients.