基于专利引用和文本信息的自动驾驶技术演化

Evolution of Automatic Driving Technology Based on Patent Citation and Text Information

摘要:

通过分析自动驾驶技术领域专利的引用信息和文本信息，结合社区发现、LDA（latent Dirichlet allocation）主题建模、主路径分析方法，针对自动驾驶专利数据集的特点建立分析框架进行自动驾驶技术演化分析。应用上述方法可以有效识别自动驾驶技术领域的主题分布，弥补传统单一方法不能反映技术全貌的缺陷。我国需重点突破多源传感器融合感知、复杂环境智能决策控制、车路协同、人机交互等关键技术；在测试和自动泊车技术上处于技术领先；在定位、高精度地图、执行机构、运动控制、硬件计算平台技术上处于技术跟随。

Abstract:

By analyzing the citation information and text information of patents in the field of automatic driving technology， and in combination with community discovery， the LDA （latent Dirichlet allocation） topic modeling， and the main path analysis method， an analysis framework is established to analyze the evolution of automatic driving technology according to the characteristics of automatic driving patent data set. The application of the above method can effectively identify the theme distribution in the field of automatic driving technology， and compensate for the defect that the traditional single method cannot reflect the whole picture of the technology. China needs to focus on breaking through key technologies such as multi-source sensor fusion perception， intelligent decision-making control in complex environment， vehicle road coordination， and human-computer interaction， take the lead in testing and automatic parking technology， and follow in positioning， high-precision map， actuator， motion control， and hardware computing platform technologies.