基于驾驶行为基元的典型驾驶场景识别与提取方法

Identification and Extraction of Typical Driving Scenarios Based on Driving Behavior Primitives

为了从海量自然驾驶数据中有效提取典型驾驶场景,提出一种基于驾驶行为基元的典型驾驶场景识别与提取方法,搭建了具有多类场景覆盖及通用程度高的典型驾驶场景自动化识别与提取框架,解决了当前方法仅适用特定单一场景提取,无法从大规模数据中高效提取场景的问题。

In order to effectively extract typical traffic scenes from massive natural driving data,a method for recognizing and extracting driving scenes is proposed based on driving behavior primitives.To address the challenges of being limited to extracting only a specific single scene and the inefficiency of scene retrieval from large-scale data acquisition,the paper developed an automatic recognition and extraction framework for typical driving scenes.This framework ensures multi-class scene coverage and features high degree of universality.