Abstract: Compared to high-speed and elevated road sections, driver's decision systems should face more complex and multi-change traffic road environments in urban roads. This paper proposes a method of combining the ontology knowledge and Bayesian Network (BN). By logic reasoning to the road environment body model, migrates to the structural parameters of BN, and constructs a complete construction Bayesian network according to historical driving cases and expert experience. Finally the best driving mode of the current scene is judged by BN. The present method can make the vehicle better judge the current optimal driving behavior.

Key words: Unmanned driving; Ontology reasoning; Bayesian network; Behavioral decision

摘要： 相较于高速与高架路段，无人驾驶行为决策系统在城区路段要面对更为复杂和多变的 交通道路环境。文章提出一种将本体知识推理和贝叶斯网络（BN）相结合的方法，通过对道 路环境本体模型加以逻辑推理，迁移为 BN 的结构参数，并根据历史驾驶案例和专家经验构 建完整的贝叶斯网络。最后通过 BN 判断当前场景的最佳驾驶模式。本方法能使车辆较好地 判断当前最优驾驶行为。

关键词: 无人驾驶；本体推理；贝叶斯网络；行为决策