Design and validation of interaction systems in autonomous vehicles

As autonornous vehicle trust through Hurnan-M portant area of researc how the vehicle perceiv driver's behavioral state control. To address th rnentary systerns that i visualization interface feature-based rnonitorinThe research follo lection, and experirnent e6 Neighborhood Elect sensors, GNSS localiza ating Systern (ROS) and alizing vehicle state, en The DMS project lever L5l 5 carnera cornbined such as Eye Aspect Rati tern integrates a calibra and Randorn Forest (RF The HMI systern scenarios, including autonornous path following, gear and speed rnonitoring, and obstacle de­ tection. The interface successfully cornmunicated critica} information such as driving mode, environrnental object detection, localization, and systern actuation feedback. For the bio­ rnetric systern, the classification rnodels achieved high accuracy using SVM when tested on unseen data. The validation process confirrned the ability to detect a range of driver states, including eye closure, head distraction, and yawning, by processing extracted features and derived behavioral rnetrics such as blink rate. The outcornes of this thesis highlight the irnportance of designing AV interfaces that cornmunicate its situational awareness, while being capable of acquiring user behavioral in­ forrnation. The findings support future integration of such systerns into shared rnobility con­ texts, with potential extensions including user feedback rnodalities, real-time driver alerting mechanisrns, and multi-user calibration capabilities for broader deployment.

https://orcid.org/0000-0001-5536-1426