Personalized Adaptive Cruise Control and Impacts on Mixed Traffic

TL;DR

This paper introduces a personalized adaptive cruise control system that learns individual driver preferences using inverse reinforcement learning and studies its effects on mixed traffic scenarios, including safety and fuel efficiency.

Contribution

It proposes a novel PACC design that adapts to driver behavior through IRL and evaluates its impact on mixed traffic dynamics and safety.

Findings

The learned driver model accurately replicates individual driving styles.

Impacts on traffic vary significantly among drivers due to personal preferences.

PACC can improve safety and fuel economy depending on driver tuning.

Abstract

This paper presents a personalized adaptive cruise control (PACC) design that can learn driver behavior and adaptively control the semi-autonomous vehicle (SAV) in the car-following scenario, and investigates its impacts on mixed traffic. In mixed traffic where the SAV and human-driven vehicles share the road, the SAV's driver can choose a PACC tuning that better fits the driver's preferred driving behaviors. The individual driver's preferences are learned through the inverse reinforcement learning (IRL) approach by recovering a unique cost function from the driver's demonstrated driving data that best explains the observed driving style. The proposed PACC design plans the motion of the SAV by minimizing the learned unique cost function considering the short preview information of the preceding human-driven vehicle. The results reveal that the learned driver model can identify and replicate the personalized driving behaviors accurately and consistently when following the preceding vehicle in a variety of traffic conditions. Furthermore, we investigated the impacts of the PACC with different drivers on mixed traffic by considering time headway, gap distance, and fuel economy assessments. A statistical investigation shows that the impacts of the PACC on mixed traffic vary among tested drivers due to their intrinsic driving preferences.