Safety enhancement by detecting driver impairment through analysis of real-time volatilities [R44]

The overall goal of the project was to focus on understanding early detection of driver impairment using streaming biometric information coupled with data on vehicle performance and surrounding contexts. During Phase 1 of the project, the team focused on developing a framework for driver impairment detection through analysis of driver biometric information along with vehicle and road infrastructure factors, with streaming data. This project contributed by implementing the framework and the model developed in Phase 1 to detect impaired driving and any abnormality in the driver, vehicle, and roadway/environment system performance. The model can be used by transportation stakeholders to reduce the probability of crashes. The motivation behind our research is to enhance safety by monitoring driver actions and detecting impairment. To accomplish this task, our team conducted experiments in a simulated environment where we requested the participants to emulate specific distracted driving behaviors, e.g., texting, reading, looking at scenery, drowsiness, and drinking. We took a multimodal approach to data collection, monitoring, and analysis. Specifically, the data included driver biometric signals, vehicle dynamics and telemetry, and external environmental conditions, e.g., traffic flow, simulated weather, day/night conditions. The outcomes of this project include embedding leading indicators of impairment in Advanced Driver Assistance System (ADAS) that can greatly enhance safety, given the substantial interest from major automotive and information technology companies, especially for applications in fleet vehicles.