Multi-Anticipative Piecewise-Linear Car-Following Model

TL;DR

This paper extends the piecewise linear car-following model to include multi-anticipative driving, demonstrating improved transient behavior and trajectory smoothing, with parameter identification based on real traffic data.

Contribution

It introduces a multi-anticipative extension to the existing model, analyzing stability, stationary regimes, and transient effects, validated with real traffic data.

Findings

Multi-anticipative model matches single-anticipative stationary behavior.

Transient traffic shows reduced velocity and acceleration variance.

Model parameters are identified using NGSIM data.

Abstract

We propose in this article an extension of the piecewise linear car-following model to multi-anticipative driving. As in the one-car-anticipative model, the stability and the stationary regimes are characterized thanks to a variational formulation of the car-dynamics. We study the homogeneous driving case. We show that in term of the stationary regime, the multi-anticipative model guarantees the same macroscopic behavior as for the one-car-anticipative one. Nevertheless, in the transient traffic, the variance in car-velocities and accelerations is mitigated by the multi-anticipative driving, and the car-trajectories are smoothed. A parameter identification of the model is made basing on NGSIM data and using a piecewise linear regression approach.