Enhancing Road Safety Through Real-Time AI-Powered Drowsiness Detection and Alert system Using EEG Eye-Blink Artifacts
PI: Daniel Otchere (ICDS)
Tuition support for the graduate student(s) working on this project will be covered by the ICDS funding requested in this proposal, while I will like to collaborate with an ICDS co-hire to support this proposal in terms of additional funding.
Project Overview
This proposal seeks to develop an innovative AI-powered system for detecting driver drowsiness through real-time analysis of EEG eye-blink artifacts. 17.6% of all fatal crashes in the United States between 2017 and 2021 involved a drowsy driver, our solution addresses critical gaps in current camera-based systems by leveraging the higher accuracy of EEG signals while overcoming computational challenges for real-world deployment. The project aligns with ICDS’s AI and Data Science hubs while fostering interdisciplinary collaboration across engineering and transportation safety domains.
Team and Collaboration
As the sole PI from ICDS, I am actively seeking collaborators from Electrical Engineering and Computer Science (for EEG hardware integration and AI), Biomedical Engineering (for signal processing expertise), and Penn State’s Larson Transportation Institute (for real-world validation). This cross-disciplinary approach will strengthen both the technical robustness and practical applicability of our solution while expanding ICDS’s research network.
Technical Approach
The project will develop an AI model optimized for real-time EEG blink detection on edge devices. We will focus on three key innovations: (1) novel feature extraction from EEG time-series data, (2) efficient model compression for deployment on low-power hardware, and (3) integration with vehicle alert systems. Our methodology includes validation using Penn State’s driving simulator and a 20-participant study.
Junior Researcher Role
We seek a graduate student (50% RA for 2 semesters) or postdoc (25% effort) with skills in Python/PyTorch, time-series analysis, EEG signal analysis, and interest in edge AI deployment. The researcher will divide effort between core algorithm development (10 hrs/week), interdisciplinary collaboration (8 hrs/week), and ICDS community engagement (2 hrs/week).
Funding and Resources
Tuition support will be provided through ICDS seed funds and departmental resources, with the PI and Co-PI actively pursuing supplemental NSF/DoT funding. The project will leverage Penn State’s existing driving simulator and collaborate with the RISE team on open-source tool development.
Timeline and Deliverables
Short-term goals (6-12 months) include analysis of EEG signals, feature extraction and model prototype and submission to IEEE Transactions on Intelligent Vehicles. Long-term objectives focus on securing external funding (NSF Smart and Connected Communities) and industry partnerships. The project will generate publishable algorithms, an open-source dataset, and patentable edge-computing solutions.
ICDS Alignment
This work embodies ICDS’s mission by applying cutting-edge AI to a pressing societal challenge. It creates natural bridges between ICDS and other colleges while offering tangible student training opportunities in translational data science. Our planned ICDS seminar on “AI for Safer Transportation” will further share insights with the broader community.
Conclusion
By combining ICDS’s AI expertise with Penn State’s engineering and transportation resources, this project offers a unique opportunity to advance both fundamental knowledge and practical applications in driver safety. We welcome junior researchers interested in working at the intersection of AI, healthcare technology, and public safety innovation.