Honors thesis · University of Toledo · 2023
Driver Attention Autonomous Transition System
DAATS connects camera-based driver drowsiness detection with a transition from manual to autonomous vehicle control.
Read the project report ↗KEEN Design Award nominationDepartmental senior design nomination
Featured inUToledo News ↗The Blade ↗
Project overview
We co-developed a prototype using OpenCV, dlib, and Eye Aspect Ratio to detect prolonged eye closure. Driver-specific recognition and temporal filtering helped distinguish fatigue indicators from ordinary blinking.
The vision processor communicated over TCP/IP with a ROS-based controller on a Jetson Xavier-powered vehicle. A multiplexer coordinated manual and autonomous control, supported by steering-wheel inputs and LiDAR obstacle detection.
My contribution
Computer-vision development, communication between the vision processor and robotic controller, performance benchmarking, and technical report co-authorship.
Evaluation
The prototype recorded an average 29.7 ms internal control-switching latency across 15 trials. Hardware comparisons on Raspberry Pi and MacBook Pro M1 examined processing delays and real-time detection bottlenecks.
Presented at the University of Toledo College of Engineering Senior Design Expo.
Report & team
Dharsan Guruparan, Jay Alikkal, Sanskar Lamsal, and Unmesh Khanolkar
Faculty supervisor: Dr. Gerald Heuring
University of Toledo · December 2023