学术讲座

报告题目： Predicting Unintentional Vehicle Lane Departure Using the Support Vector Machine and Neural Network

报 告 人：应浩，美国密歇根州底特律市韦恩州立大学电气与计算机工程系教授

报告时间： 2019 年 5 月 29 日   下午 3 点

报告地点：信息楼 417 学术会议室

 

报告人介绍：应浩，美国密歇根州底特律市韦恩州立大学电气与计算机工程系教授， IEEE Fellow 。出版了两本模糊控制书籍、 110 多篇期刊论文和 160 多篇会议论文。他的论文被广泛引用（谷歌学者 H 指数 47 ）。担任《 IEEE Transactions on Fuzzy Systems 》 和《 IEEE Transactions on Systems, Man, and Cybernetics: Systems 》等 9 家国际期刊的副主编或编委会成员，并在过去 12 年的 11 年中一直在 IEEE 计算智能学会的模糊系统技术委员会任职。他曾在 2016 年和 2017 年担任 IEEE 系统、人和控制论协会的成员，三次国际会议的项目主席，以及项目 / 技术委员会成员。

 

报告摘要： Abstract — Advanced driver assistance systems, such as unintentional lane departure warning systems, have recently drawn much attention and efforts. We explored utilizing the nonlinear binary support vector machine (SVM) technique as well as the three-layer neural network with the back-propagation learning scheme to predict unintentional lane departure. We developed a two-stage training scheme to improve SVM’s prediction performance in terms of minimization of the number of false positive prediction errors. Experiment data generated by a Virtual Test Track Experiment (VIRTTEX) simulator, which is a hydraulically powered 6-degrees-of-freedom moving base driving simulator at Ford Motor Company, were used. All the 100+ vehicle variables were sampled at 50 Hz and there were 16 drowsy drivers (about three-hour driving per driver) and six control drivers (approximately 20 minutes driving each). A total of 3,508 unintentional lane departures occurred for the drowsy drivers and 23 for the control drivers.  Our study involving these 22 drivers with a total of over 7.5 million prediction decisions demonstrates that: (1) excellent SVM prediction performance, measured by numbers of false positives (i.e., falsely predicted lane departures) and false negatives (i.e., lane departures  failed  to  be predicted), were achieved when the prediction horizon was 0.6 s or less, (2) lateral position and lateral velocity worked the best as SVM input variables among the nine variable sets that we explored, (3) the radial basis function performed the best as the SVM kernel function, and (4) the SVM produced more accurate lane departure prediction than the neural network did.