基于瞳孔变动视觉特征的驾驶人换道意图识别

Driver Lane Change Intention Recognition Based on Pupil Characteristics

针对驾驶人换道意图未能准确识别,导致车辆因换道事故频出的问题,采用瞳孔变动视觉特征的驾驶人换道意图识别方法。根据交感神经和副交感神经共同控制的瞳孔状态特点,构建能够检测到瞳孔及其大小变动的鲁棒基线策略,利用n阶多项式逼近的极小平方法,获取瞳孔变动与刺激强度的关系式,将换道过程分割为换道意图和换道执行,同时提取换道意图中的参数指标,以初始时窗宽度为基础,获取其中的待识别样本实际值,采用赋值方法得到相应的基本概率赋值,基于多证据合成规则公式,融合各形式的信任分配概率数据,从而取得基本信任概率分配,通过可信度赋值决策准则验证融合结果,获得意图识别结果,经过不等式对其的判定,得到最终的换道意图。以草原道路上车辆驾驶人为实验对象进行仿真,所提方法能够有效地表明驾驶人的换道意图,识别精准度高。

If the driver's intention of changing lanes is not accurately recognized,it will lead to frequent accidents.In this article,a method to recognize driver's intention of changing lanes based on visual characteristic of pupil change was put forward.According to the state characteristics of pupil controlled by the sympathetic nerve and the parasympathetic nerve,a robust baseline strategy of detecting changes of pupil was established.Moreover,the least square method of n-order polynomial approximation was used to obtain the relationship between pupil change and stimulus intensity.The process of lane change was divided into intention and execution.At the same time,the parameter indexes in intention were extracted.Based on the initial width of time window,the actual value of the sample to be identified was obtained,and the corresponding basic probability values were obtained by the assignment method.Based on the formula of multi-evidence combination rule,various forms of trust distribution probability data were fused to obtain the basic trust probability distribution.In addition,the fusion result was proved through the credibility assignment decision,so that the result of intention recognition was obtained.Finally,intention of changing lanes was obtained through inequalities.The Simulation was based on the driver on the grassland road.Simulation results prove that the proposed method can effectively indicate the driver's intention of changing lanes,so it has high recognition accuracy.