考虑驾驶习惯的驾驶员换道意图识别

Driver's lane change intention recognition considering driving habits

考虑了换道行驶过程中驾驶员的驾驶习惯因素,并基于多维高斯混合隐马尔可夫模型(MGHMM),构建了针对每名驾驶员的驾驶员换道意图识别模型,从而增强驾驶员换道意图的识别效果。结合Carmaker车辆动力学仿真软件搭建模拟驾驶平台,采集、分析和筛选模型观测数据。通过对比试验分析,验证了考虑驾驶习惯因素建模能够提高识别性能,同时分析出现误识别的因素,为后续研究工作提供参考。

Based on the Multi-dimension Gauss Hidden Markov Model(MGHMM),this paper establishes a lane change intention recognition model for each driver to consider driving habits during lane change,in order to identify the lane change intention earlier in the driver's lane change.Combined with Carmaker's vehicle dynamics simulation software a simulation driving platform is built up,and the model observation data are collected,analyzed and screened.Through comparative experimental analysis,it is verified that modeling with consideration of driving habits can improve recognition performance.The misidentification factors are analyzed to provide reference for subsequent research work.