A phase-aware AI car-following model for electric vehicles with adaptive cruise control: Development and validation using real-world data

TL;DR

The paper addresses the challenge of accurately modeling EV car-following dynamics within mixed traffic by introducing a Phase-Aware AI (PAAI) framework that augments traditional physics-based models with AI corrections and explicit phase recognition. It develops two AI-based CF models—the Baseline AI and the PAAI model—trained on high-fidelity ACC trajectory data from ICE and EV vehicles, and demonstrates that phase-aware residual corrections substantially improve acceleration, speed, and spacing predictions compared with conventional models. Key findings show EV-specific dynamics, such as rapid torque response and regenerative braking, lead to asymmetric acceleration/deceleration patterns and higher jerk, which the PAAI framework captures more accurately, yielding significant RMSE reductions, especially in speed and spacing. The work advances EV traffic modeling by enabling more accurate, data-driven mixed-traffic simulations, with implications for traffic management, safety analysis, and infrastructure planning as EV penetration grows.

Abstract

Internal combustion engine (ICE) vehicles and electric vehicles (EVs) exhibit distinct vehicle dynamics. EVs provide rapid acceleration, with electric motors producing peak power across a wider speed range, and achieve swift deceleration through regenerative braking. While existing microscopic models effectively capture the driving behavior of ICE vehicles, a modeling framework that accurately describes the unique car-following dynamics of EVs is lacking. Developing such a model is essential given the increasing presence of EVs in traffic, yet creating an easy-to-use and accurate analytical model remains challenging. To address these gaps, this study develops and validates a Phase-Aware AI (PAAI) car-following model specifically for EVs. The proposed model enhances traditional physics-based frameworks with an AI component that recognizes and adapts to different driving phases, such as rapid acceleration and regenerative braking. Using real-world trajectory data from vehicles equipped with adaptive cruise control (ACC), we conduct comprehensive simulations to validate the model's performance. The numerical results demonstrate that the PAAI model significantly improves prediction accuracy over traditional car-following models, providing an effective tool for accurately representing EV behavior in traffic simulations.

Figures (10)

• Figure 1: Speed and spacing profiles for ICE vehicles A and B under the 'speed dips' condition with minimum and maximum ACC settings. The first row shows speed profiles for Vehicle A (min/max) and Vehicle B (min/max) from left to right; the second row represents spacing profiles for Vehicle A (min/max) and Vehicle B (min/max) from left to right.
• Figure 2: Speed and spacing profiles from CF experiments for short (approximately 25 m), medium (approximately 35 m), long (approximately 45 m), and extra-long (approximately 55 m) space-gap settings at 55 mph, representing typical highway free-flow conditions, as defined by initial headway distances.
• Figure 3: Acceleration comparison between ICE vehicles and EVs.
• Figure 4: Relative speed comparison for ICE vehicles and EVs: (a) box plot revealing gap-dependent variability, (b) probability density distributions showing asymmetric EV patterns at larger gaps, and (c) KS-CDF plot confirming systematic behavioral differences between vehicle types.
• Figure 5: Speed comparison for ICE vehicles and EVs: (a) box plot showing conservative ICE behavior versus precise EV speed maintenance, (b) probability density distributions revealing concentrated EV peaks at target speed, and (c) KS-CDF plot quantifying the distinct speed-maintenance strategies.