Safety Evaluation and Improvement of Autonomous Vehicles

点击翻译按钮获取中文摘要

Advanced driver assistance systems (ADAS) promise accident-free roads, however, until level-5 autonomy (the fully automated vehicle) is reached, ultimate driving responsibility relies on the human driver. Past research shows that drivers easily become complacent about the semi-automated system. Hence, the safety implications of the automated driving features need to be fully understood to enhance road safety. The proposed research will develop tools to measure the cognitive load of drivers for safety evaluation and will assess the safety benefits and risks associated with a vast array of ADAS in order to help the Ontario Ministry of Transportation develop policies and regulations on ADAS use on Ontario roads. Further, the proposed research will benchmark a commercial driver monitoring system being developed by Dreyev, one of the partner organizations, to assess the potential of cognitive load measures to enhance the performance of state-of-the-art driver monitoring systems. In addition, the proposed research will develop signal processing and machine learning techniques to enhance the quality of cognitive load information obtained from physiological measures, such as pupil size, eye-gaze, and heart rate. Then, these measures will be analyzed in conjunction with the existing signals used by Dreyev for effective driver monitoring. Finally, the proposed research will demonstrate the benefits of cognitive load measures, collected from non-invasive and inexpensive physiological sensors to improve the performance and accuracy of driver monitoring systems.