The paper introduces DME-Driver, an autonomous driving system that integrates human decision logic and 3D scene perception to enhance performance and reliability. DME-Driver features a Decision-Maker Executor (DME) structure, combining a Large Vision Language Model (LVLM) for logical decision-making and a planning-oriented perception model for accurate control signal generation. The LVLM is trained using extensive real-world driving data and human driver behavior logic, enabling it to simulate human-like logical assessments and provide reliable prior information to the perception model. The Executor model translates these logical decisions into precise vehicle control signals, ensuring effective and context-aware responses in various driving situations. The system's effectiveness is evaluated through a comprehensive dataset, HBD, which integrates human driver behavior logic and detailed environmental perception. Empirical results demonstrate that DME-Driver achieves state-of-the-art accuracy in autonomous driving planning while significantly enhancing interpretability and robustness.The paper introduces DME-Driver, an autonomous driving system that integrates human decision logic and 3D scene perception to enhance performance and reliability. DME-Driver features a Decision-Maker Executor (DME) structure, combining a Large Vision Language Model (LVLM) for logical decision-making and a planning-oriented perception model for accurate control signal generation. The LVLM is trained using extensive real-world driving data and human driver behavior logic, enabling it to simulate human-like logical assessments and provide reliable prior information to the perception model. The Executor model translates these logical decisions into precise vehicle control signals, ensuring effective and context-aware responses in various driving situations. The system's effectiveness is evaluated through a comprehensive dataset, HBD, which integrates human driver behavior logic and detailed environmental perception. Empirical results demonstrate that DME-Driver achieves state-of-the-art accuracy in autonomous driving planning while significantly enhancing interpretability and robustness.