摘要
驾驶决策过程中,驾驶行为常受到人、车、路、环境等多源信息的刺激和影响。由于信息处理能力有限,驾驶员对多源信息无法同时实现知识获取与表示,以致有时不能准确、快速地进行驾驶决策,易引发交通事故。文章利用决策树能融知识表示与获取于一身的优点,将决策树用于不同驾驶行为决策机制的研究,以实现对驾驶员行为的模拟再现。仿真结果表明,用决策树构建的驾驶决策识别模型有较高的推理速度,能实时、准确地识别当前的驾驶行为和预测下一时刻的驾驶决策,为智能车辆中自动驾驶系统的仿真和实现提供了理论指导和可行性依据。
In driving decision-making process,driving behavior is usually affected by many elements,such as human,vehicle, road and environment.Because of the limitation of information processing capabilities,knowledge representation and acquisition are not be able to realized simultaneously,and driving decision can not be made so quickly and correctly that traffic incident usually happens.Decision tree was used to study the decision mechanism of driving behavior in order to simulate driving behavior.Simulation results show that the recognition model of driving decision based on decision tree possesses a high reasoning speed.Current driving behavior and next driving decision can be recognized and forecasted exactly.The theory base and feasibility can be provided for automatic driving system in intelligent vehicles.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2008年第2期415-419,448,共6页
Journal of System Simulation
基金
山东省自然科学基金项目(Y2006G32)
山东理工大学科研基金重点资助项目(2004KJZ02)
关键词
驾驶员行为
决策机制
决策树
信息熵
分类规则
交通流
智能运输系统(ITS)
driving behavior
decision mechanism
decision tree
information entropy
classification rule
traffic flow
intelligent transportation systems（ITS）
