CAJ | 학술논문

针对目前智能汽车仿真中应用最广的环境感知方法(建立数据库，动态更新提取信息)的不足，提出了一种分类、分层的数据结构定义与存储方法，使数据库不但能描述复杂的道路情况，而且修改容易；在此基础上，将碰撞检测算法引入信息动态提取过程中，实现分层的感知信息提取。在驾驶模拟器上进行的车道标线、交通标志感知试验的结果表明，本文中提出的方法能实时对环境信息进行全面可靠的检测，并给出真实传感器的检测误差。
침대목전지능기차방진중응용최엄적배경감지방법(건립수거고，동태경신제취신식)적불족，제출료일충분류、분층적수거결구정의여존저방법，사수거고불단능묘술복잡적도로정황，이차수개용역；재차기출상，장팽당검측산법인입신식동태제취과정중，실현분층적감지신식제취。재가사모의기상진행적차도표선、교통표지감지시험적결과표명，본문중제출적방법능실시대배경신식진행전면가고적검측，병급출진실전감기적검측오차。
[ Abstract] In view of the inadequacy of the most commonly used way of environmental perception in current intelligent vehicle simulation, i. e. database setup and dynamic information update and extraction, a definition and storage method of classified and hierarchical data structure is proposed, which can not only describe complex road condition, but is also easy to modify. On this basis, a collision detection algorithm is introduced into the extraction process of dynamic information to realize the extraction of layered perception information. A perception test of lane markings and traffic signs is conducted on driving simulator with a result showing that the method proposed can fulfill thorough and reliable detection on environment information in real time with the detection error of real sensor given.