A comprehensive review of data, features, and modeling approaches for estimating remaining driving range in electric vehicle

Assessing a vehicle’s remaining driving range (RDR) is essential for alleviating range anxiety, improving energy management, and supporting sustainable mobility in public spaces for electric vehicles (EVs). Recent advancements in data-driven modeling have transformed approaches to RDR estimation, moving from conventional rule-based methods for intelligent systems capable of learning complex relationships among driving styles, battery dynamics, and interactions with external environments. Several studies have demonstrated that using real-time sensor data in conjunction with machine learning (ML) and Deep learning (DL) models enhances the accuracy of RDR estimation and captures the multifaceted relationships among driving speed, acceleration, temperature, road topology, and state-of-charge (SoC) variables. In addition to accuracy, interpretability and trustworthiness remain significant challenges, highlighting the need for causal inference and uncertainty quantification as an essential component in the development of future predictive systems.