A Novel Approach to Detect Drowsy Driving Resulting From Sleep Loss and Sleep Pressure

睡眠不足已经成为一种现代流行病，对白天的功能有重大影响，在美国每年约有6000起致命车祸是由于昏昏欲睡驾驶，在加拿大每年约有100起。一个可能危及生命的结果是驾驶员疲劳加剧，经常导致严重的机动车事故。这项研究的主要目标是开发一种非侵入性、实时和非侵入性的车载系统，以便在机动车事故发生之前检测驾驶员的疲劳。我们提出的系统使用红外线(IR)眼睛跟踪相机技术来测量嗜睡前的眼睛运动。我们的新型人工智能算法在线对这些眼球运动进行分类，以检测嗜睡的前兆，用于在驾驶时提醒司机。在这里，我们利用驾驶模拟技术，研究眼动特征与驾驶员昏昏欲睡的行为和电生理预警信号之间的关系。这将使我们能够开发一种方法，在发生损伤和判断失误之前检测嗜睡情况并提醒司机。我们计划开发一种新的方法，仅使用眼睛跟踪数据就可以有效地检测嗜睡：1)变化的睡眠压力，2)白天或晚上不同时间昏昏欲睡的自然变化。最终，这将导致“预警系统”的商业化，作为对运输卡车运输业、公共交通和个人乘用车昏昏欲睡损害的一种对策。我们预计，这项工作将为研究和应用由于各种其他原因，如酒精、吸毒和分心驾驶而导致的驾驶适合性开辟进一步的途径。

Sleep loss has become a modern-day epidemic, with a significant impact on daytime functioning, resulting in about 6000 fatal car crashes/year due to drowsy driving in the US and about 100/year in Canada. One potentially life-threatening outcome is increased driver fatigue, often resulting in serious motor vehicle accidents. The main objective of the proposed research is the development of a non-invasive, real-time and unobtrusive in-vehicle system to detect driver fatigue before a motor vehicle accident occurs. Our proposed system measures eye movements that precede drowsiness using infra-red (IR) eye tracking camera technology. These eye movements are classified online by our novel artificial intelligence algorithms to detect the antecedents of somnolence, to be used to alert the driver while driving. Here, using driving simulation technology, we aim to study the relationship between eye movement characteristics and the behavioural and electrophysiological warning signs of impending driver sleepiness. This will allow us to develop a method to detect sleepiness and alert drivers before impairment and lapses in judgement occur. We plan to develop a novel method to effectively detect drowsiness using only eye tracking data in: 1) varying sleep-pressure, and, 2) natural variations in drowsiness at different times of day or night. Ultimately, this will result in the commercialization of an "early warning system", as a countermeasure to drowsiness impairment in the transport-trucking industry, public transit and in individual passenger vehicles. We expect that this work will open up further avenues for research and applications about the fitness to drive due to various other causes, such as alcohol, drug use and distracted driving.