Mitigation Strategies for Driver Distraction and Drowsiness (DDD) at different levels of Automation
for a brighter
tomorrow
AI for Road User Assistance
Research at a glance
Driver distraction and drowsiness (DDD) are implicated in a large share of fatal and serious-injury crashes and remain a persistent risk even as Advanced Driver-Assistance Systems become widespread. My research develops AI-based methods to detect DDD and support its mitigation through driver behavioural precursors — the subtle changes that appear before a driver is visibly drowsy or distracted — across SAE automation levels L0 to L2. Using large naturalistic driving datasets, I combine computer-vision analysis of facial and ocular behaviour with vehicle kinematics and physiological signals. Current work links these indicators to measurable risk, such as reduced time-to-collision and harsh braking and examines how automation changes DDD patterns. The goal is to enhance DDD mitigation by improving early detection and strengthening the evidence base for driver monitoring systems (DMS) that intervene before risk materialises.
Research objectives
- Quantify how DDD indicators and their behavioural precursors evolve across different naturalistic driving situations and SAE automation levels (L0–L2) and link them to the risk of traffic conflicts and crashes.
- Develop vision-based DDD detection on large-scale naturalistic driving video databases, extracting behavioural and ocular indicators of driver state.
- Fuse behavioural, kinematic and physiological data streams to improve detection accuracy and generalisability across drivers and driving situations.
- Translate the findings into evidence-based recommendations for driver monitoring systems (DMS) and for the design of automation features that target DDD.
Publications
| Title | Authors | Venue | Year | Link |
|---|---|---|---|---|
| Characterizing driver drowsiness in automated driving: A machine learning analysis of naturalistic driving data. Accepted and presented; proceedings publication pending | Iyer, A., Afghari, A.P., Papadimitriou, E. | RSS | 2026 | – |
Conference contributions
| Title | Conference | Date | Link |
|---|---|---|---|
| Characterizing driver drowsiness in automated driving: A machine learning analysis of naturalistic driving data | RSS 2026 Naples, Italy | Jun 2026 | ↗ |
| Driver Distraction Analysis: A Big-Data Analytics Approach Using Naturalistic Driving Data | Transportation Research Symposium 2025 Rotterdam, Netherlands | May 2025 | – |