基于集成学习的换道决策与状态估计方法研究

Situation Assessment and Decision Making for Lane Change Assistance Using Ensemble Learning Methods

为提高换道辅助系统阈值的精确度,提出了两种可应用于换道辅助系统的集成学习方法:随机森林和Ada Boost算法。采用实车试验数据建立和验证算法,结果表明:相比于以往研究中使用的贝叶斯或者决策树分类方法,这两种集成学习方法具备较高的分类精度和较低的假阳性率。采用Ada Boost算法的车道保持精度可达99.1%,随机森林精度为98.7%,相应的真阳性率分别为96.3%和94.6%。

In order to increase the accuracy of safety critical lane change events, two ensemble learning methods for lane change assistance system are put forward. Detailed vehicle trajectory data from the real vehicle test. The results showed that both ensemble learning methods produced higher classification accuracy and lower false positive rates than the Bayes/Decisiontree classifier used in the literature.