Evaluation of Driver Distraction with Changes in Gaze Direction Based on a Vestibulo-Ocular Reflex Model

With the aim of improving parameter identification and, eventually, evaluating driver distraction with changes in gaze direction, we applied a genetic algorithm (GA) method to identify parameters for an existing vestibulo-ocular reflex (VOR) model. By changing the initial inputs to the GA and fixing two parameters pertaining to the horizontal direction, we achieved improved parameter identification with a lower mean-square error. The influence of driver distraction on eye movement with changes in gaze direction was evaluated from the difference between the predicted and observed VOR in the vertical axis. When a driver was given an additional mental workload, the mean-square error between the measured and simulated values was bigger than that in the absence of the mental workload. This confirmed the relationship between driver distraction and eye movement in the vertical direction. We hope that this method can be applied in evaluating driver distraction.

基于仿生机理的头眼协调系统

针对多数头眼系统易受环境因素或复杂坐标变换影响的问题,基于人类眼球生理学和解剖学原理,根据眼球运动的神经回路,通过对重要神经核功能以及视觉信息传递处理机制的研究,建立一个具有眼球自适应前庭动眼反射与平滑追踪运动控制模型。结合人类行为特征,提出一种头眼协调控制方法,并运用Matlab进行控制系统仿真模拟。仿真结果表明,该头眼协调运动控制方法不仅可以应用于机器人仿生视觉系统中,而且具有良好的鲁棒性。

头眼协调运动控制仿生模型

深入分析灵长类动物在注视点转移过程中眼球运动、头部运动及头眼协调运动的关系,将头眼协调运动的神经生理机制引入工程领域,改善机器人视觉系统的头眼协调问题。提出一种仿生型机器人头眼协调运动控制策略,将注视转移过程分为初始的快相和随后的慢相两个阶段。快相组合了高速眼球扫视运动(saccade)眼球运动和缓慢的头部运动,两者协调配合迅速切换注视点至新的目标;慢相通过前庭动眼反射(VOR),依靠头和眼的等量反向旋转运动维持目标稳定的同时,调整头部位置,使其朝向目标。建立了注视转移的仿生控制模型。仿真结果与生理学实验数据进行了对比分析,证明了所提控制策略与控制模型的可行性。