【作者】 管欣； 李思深； 贾鑫； 陈永尚； 詹军；

【Author】 Guan Xin;Li Sishen;Jia Xin;Chen Yongshang;Zhan Jun;Jilin University,State Key Laboratory of Automotive Simulation and Control;

【机构】 吉林大学汽车仿真与控制国家重点实验室；

【摘要】 智能汽车对目标车道进行选择时,为了保证驾驶安全性,需要对候选车道风险进行评估。本文提出一种考虑他车横向驾驶习惯的风险评估方法,采用在线最大期望算法对前方车辆横向驾驶习惯进行统计,采用高斯混合—隐式马尔科夫算法对前方车辆驾驶意图进行识别;根据识别结果,利用驾驶习惯统计数据及高斯过程回归算法对车道占有特性进行计算,根据车道占有特性对车道风险进行评估。最后,利用智能汽车仿真平台,对所采用的算法进行验证。更多还原

【Abstract】 When the intelligent vehicle makes target lane decision,in order to ensure driving safety,it is necessary to assess the risk of the candidate lanes.This paper proposes a lane risk assessment approach which consider other vehicles’ lateral driving style.The online EM algorithm is used to calculate the lateral driving styles of the preceding vehicles.The GMM-HMM algorithm is used to identify the intentions of the preceding vehicles.According to the recognition results,the driving style statistics and the Gaussian process regression algorithm are used.to calculate the lane occupancy characteristics.The risk assessment based on the lane occupancy characteristics are conducted.In the end,the Autonomous Vehicle Simulator Platfonn is used to verify the algorithm.更多还原