A Full-Chain Dataset for Intelligent New Energy Vehicles based on High-fidelity Digital Twin Platform
摘要
With the rapid development of intelligent new energy vehicles, there is an increasing demand for datasets that can jointly characterize traffic environments, vehicle control behaviors, and powertrain energy responses. Existing public datasets often focus on perception tasks or isolated research directions, making them insufficient for system-level closed-loop studies. To address this gap, we present 3ITVP (Intelligent Transportation, Intelligent Vehicle, Intelligent Power), a full-chain dataset built on a high-fidelity digital twin platform. The dataset approaches from the perspectives of five types of vehicles: ordinary cars, including both electric and fuel vehicles; light trucks; semi-trailers; logistics vehicles and buses, exploring their driving conditions and behaviors. Rigorously time-synchronized multimodal data is provided, and systematically covers key aspects of intelligent transportation, autonomous driving and new energy technologies. In addition to long-horizon continuous driving behaviors, 3ITVP includes a wide range of special event cases, supporting research on perception, decision-making, planning and control, energy management, and Sim-to-Real transfer.