注视熵和马尔科夫链的弯道诱导设施信息量研究

Research on information volume of guidance facilities on bends based on gaze entropy and Markov chain

为探究驾驶人在不同诱导设施信息量的草原公路弯道行车时的注意力分配情况,建立5种不同层级诱导设施信息量(Q0、Q1、Q2、Q3、Q4)的草原公路弯道虚拟场景,并对其进行模拟驾驶试验;对驾驶人视野分区,分析驾驶人在不同信息量条件下注视熵率值变化特征;应用马尔科夫链计算驾驶人在不同信息量条件下注视点在各个区域间的一步转移概率和平稳分布概率。结果表明:草原公路弯道诱导设施信息量的变化对驾驶人注视熵率值影响显著,驾驶人在Q2(10~20 bits/km)信息量条件下注视熵率值最大;驾驶人需要对信息目标多次重复注视才能获取足够有效行车信息,90%以上的概率集中于道路左侧和远方区域,其中道路远方区域是驾驶人保证正常驾驶的重要观察区域。

In order to study drivers’attention distribution while driving on bends of grassland highways with different guidance facilities information volume,five virtual scenes of bends with diverse levels of information volume(Q0,Q1,Q2,Q3,Q4)were established,and driving simulation experiments were carried out.Then,drivers’visual fields were divided,and their gaze entropy changes under different volume conditions were analyzed.Finally,Markov chain was used to calculate one-step transfer probability and stable distribution probability of gaze in different regions.The results show that information volume change of guidance facilities on grassland highway bends has a significant impact on drivers’gaze entropy which would be the largest under Q2 condition(10-20 bits/km).They need to repeatedly gaze at target to obtain sufficient and effective driving information,and more than 90%of probability is concentrated in left and far areas of roads,of which the latter one is an significant zone for drivers to ensure normal driving.