汽车前方静动目标状态转移机理与分类算法  

The State Transfer Mechanism and Classification Algorithm for Stationary and Moving Objects in Front of Vehicle

作　　者：高振海[1] 王竣[1] 王德平[2] 李红建[3] 

机构地区：[1]吉林大学,汽车仿真与控制国家重点实验室,长春130025 [2]中国第一汽车集团公司新能源汽车分公司,长春130122 [3]中国第一汽车集团公司技术中心,长春130011

出　　处：《汽车工程》2016年第5期638-645,共8页Automotive Engineering

基　　金：高等学校博士学科点专项科研基金(20120061110028);吉林省科技引导计划(20130413058GH);长江学者和创新团队发展计划(IRT1017)资助

摘　　要：为解决利用雷达回波实现静止目标和运动目标的准确识别这一驾驶辅助系统的关键技术问题,本文中基于地面目标运动状态转移机理提出了一种基于时间窗的汽车前方静动目标状态分类方法。在地面静动目标运动状态与转移机理分析的基础上,将目标分为静止目标、同向运动目标、反向运动目标、起停目标和未分类目标等5类,建立了在固定时间窗内的目标运动状态的转移状态机模型,并确定了目标状态转移的条件阈值和时间窗长度,最终在驾驶辅助试验车上进行了前方同向或反向行驶车辆、树木等静止物体和制动停车车辆等各种典型工况下的识别试验,为实现基于毫米波雷达的自适应巡航与自动紧急制动的驾驶辅助系统的工程化提供了技术支撑。To tackle the key technical issue of driving assistance system,i. e. the accurate detection of stationary and moving objects by radar echo,a time window-based vehicle frontal object state classification method is proposed in this paper based on the movement state transfer mechanism of ground objects. On the basis of analyses on ground object movement states and their transfer mechanism,Stationary and moving objects are classified into five categories： stationary object,moving object,oncoming object,stopped object and unclassified object. Then a transfer state machine model for the object movement states within a fixed time window is created,and the condition threshold and time window length for object state transfer are determined. Finally,the detection tests of moving,oncoming,braking and stopping vehicles and trees and other stationary objects under various working conditions are conducted on an experimental vehicle,providing technical supports for the engineering application of millimeterwave radar-based adaptive cruise and automatic emergent braking in driving assistance system.

关 键 词：驾驶辅助系统 状态机模型 车载毫米波雷达 静动目标分类 时间窗 

分 类 号：U463.6[机械工程—车辆工程]