ACC系统用目标换道预测方法

A target lane changing prediction method for ACC system

为提高车辆自适应巡航控制(ACC)系统的有效性,利用雷达传感器、车道线识别传感器、车载陀螺仪、车辆总线设备等搭建车载试验平台,获取真实交通流环境中自车与前方车辆运动状态时的表征数据。基于自车与前方车辆的距离、前方车辆的横向速度与纵向速度参数,采用隐马尔科夫理论,建立前方车辆换道意图预测模型。用实测数据检验该模型。结果表明:用该模型能够准确快速预测前方车辆的车道变换与车道保持行为。在4.5 s的时间窗口宽度下,直线路段的预测准确率达到97%;在3.5 s的时间窗口宽度下,曲线路段的预测准确率达到96%。

In order to improve the effectiveness of vehicle ACC system,a millimeter-wave radar sensor,lane line sensor,car gyroscope,and controller area network instrument were used to build a test platform.Experiments were carried out in a real traffic environment. Moving state characterization data on vehicle in guestion and vehicle in front of it were obtained. On the basis of parameters of distance between two vehicles,lateral speed and longitudinal velocity of vehicle in front,a lane change intent prediction model was built for the vehicle in front by using of hidden Markov theory. Measured data were used to test the model.The result shows that predictive model can be used to accurately predict the lane change and maintain behavior of vehicle in front vehicle,that with time window opening width of 4. 5 s,the maximum prediction accuracy for linear section road is 97%,and that with width of 3. 5 s,the maximum prediction for accuracy for curve section road is 96%.