Artificial intelligence-driven embedded sensing, digital twin, and modular electrified vehicle systems for intelligent sustainable mobility

Engineering notes

Key topics: autonomous driving, route planning, lidar, sensor fusion, reinforcement learning, large language model, deployment, radar, perception, planning, control. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Abstract The rapid transition toward sustainable and electrified transportation has accelerated the integration of artificial intelligence (AI) and embedded intelligence across modern mobility platforms. Despite significant advancements in electric and hybrid transportation systems, existing review studies predominantly focus on isolated aspects such as battery management, autonomous perception, or energy optimization, lacking a unified perspective on how AI-driven embedded systems collectively shape next-generation intelligent mobility ecosystems. This paper presents a comprehensive and forward-looking review of AI-enabled embedded architectures for electric and hybrid transport systems, while also incorporating original analytical insights on deep reinforcement learning (DRL)-based energy management and range optimization frameworks. The study critically examines AI-assisted battery state-of-charge estimation, predictive battery management systems (BMS), real-time sensor fusion, embedded vehicle health monitoring, digital twins (DTs), IoT-enabled diagnostics, and vehicle-to-everything (V2X) communication technologies. Furthermore, the review investigates AI-optimized propulsion architectures, modular power electronic interfaces, and intelligent perception systems integrating LiDAR, radar, and vision sensors for autonomous driving applications. Unlike conventional surveys, this work bridges the gap between embedded AI hardware, intelligent sensing, energy-aware control, and scalable mobility infrastructure within a unified framework. In addition, DRL-based dispatch strategies for battery energy storage systems (BESS) are comparatively analyzed against traditional rule-based and model predictive approaches, demonstrating improved adaptability and operational efficiency under dynamic and high-temperature operating conditions. The paper also highlights emerging opportunities associated with generative AI and large language models (LLMs) for vehicle health diagnostics, adaptive route planning, and intelligent energy management. Finally, critical research challenges related to explainability, computational complexity, battery aging, sensor reliability, and large-scale deployment are discussed to outline future directions toward practical, sustainable, and autonomous AI-driven transportation ecosystems.

Links and sources

Comments