基于多维高斯隐马尔科夫模型的驾驶员转向行为辨识方法

Driving Intention Recognition on Cornering Based on Multi-dimensional Gaussian Hidden Markov Model

在线控转向系统中,采用一种基于多维高斯隐马尔科夫模型的驾驶员转向行为辨识方法,可达到辅助驾驶员驾驶、屏蔽驾驶员错误操作和提高汽车主动安全性的目的。通过驾驶模拟器采集相应工况数据,经数据预处理后,应用Baum-Welch算法对多维高斯隐马尔科夫模型进行优化,且应用Labview进行在线辨识,准确率达到99.8%。

In steer-by-wire system,a driving turning intention recognition method based on multi-dimensional Gaussian Hidden Markov Model（HMM） can be used to assist driving or shield wrong driving operation and improve the active safety.By collecting corresponding operating data from driving simulator,which are preprocessed,and optimizing the multi-dimensional Gaussian HMM with Baum-Welch algorithm,and with the aid of on-line identification with LabVIEW,the accuracy can achieve 99.8%.